Search

This thesis proposes an optimisation strategy for scalable Big Data processing in a heterogeneous Cloud. The resource needs of A Large Ion Collider Experiment (ALICE) at the European Organization for Nuclear Research (CERN) are reviewed as a motivating example. The thesis examines how to efficiently process and optimise the processing of resourceintensive tasks on a heterogeneous Cloud infrastructure distributed in five data centres to meet the needs of the ALICE experiment at the Tier 2 level. The objective was to perform research on a much larger number of tasks and resources of a significantly larger capacity than prior studies, which focused on a smaller number of tasks and resources with a lower capacity. The proposed and developed processing model for ALICE Monte Carlo production is based on a centralised software-defined management approach for the use of heterogeneous resources. Algorithms for assigning tasks to heterogeneous virtual resources have been analysed and proposed. The proposed algorithms are based on the selected Evolution Strategies metaheuristic that has not yet been used in this domain, namely Evolution Strategies algorithm, Evolution Strategies algorithm with Longest Job First broker policy, and Evolution Strategies algorithm with Shortest Job First broker policy. The Cloud system model is implemented using the open-source CloudSim simulator. ALICE Monte Carlo production job requirements are imported into the simulation model as a workload created in Standard Workload Format (SWF) adapted for the Cloud simulator. The results of the simulation performance of the reference implementation under different loads were analysed and compared with the Genetic Algorithm from the same group of algorithms. The obtained results show multiple improvements. The proposed data processing model enables centralised software management of heterogeneous Cloud infrastructure, optimises measured metrics, improves resource usage, and achieves the system's scalability. Ova disertacija predlaže strategiju optimizacije za skalabilnu obradu velikih podataka u heterogenom oblaku (engl. Cloud). U radu se kao motivirajući primjer razmatraju potrebe za resursima Eksperimenta na velikom ionskom sudaraču (engl. A Large Ion Collider Experiment, ALICE) na Europskoj organizaciji za nuklearna istraživanja (engl. European Organization for Nuclear Research, CERN). Rad istražuje kako učinkovito obraditi i optimizirati obradu resursno-intenzivnih zadataka na heterogenoj infrastrukturi u oblaku raspoređenoj u pet podatkovnih centara kako bi se zadovoljile potrebe na Tier 2 razini ALICE eksperimenta. Cilj je bio provesti istraživanje za višestruko veći broj zadataka i resurse znatno većeg kapaciteta u odnosu na dosadašnja istraživanja koja su provođena na manjem broju zadataka i resursima manjih kapaciteta. Predloženi i razvijeni model obrade za ALICE Monte Carlo produkciju temelji se na centraliziranom softverski definiranom pristupu upravljanja korištenjem heterogenih resursa. Analizirani su i predloženi algoritmi za dodjelu zadataka heterogenim virtualnim resursima. Predloženi algoritmi temelje se na odabranoj metaheuristici evolucijskih strategija (engl. Evolution Strategies) koja dosad nije korištena u ovoj domeni, a to su algoritam evolucijskih strategija, algoritam evolucijskih strategija s broker politikom kod koje prioritet izvođenja imaju najdulji zadaci (engl. Longest Job First) i algoritam evolucijskih strategija s broker politikom koja prioritet izvođenja daje najkraćim zadacima (engl. Shortest Job First). Model računalnog oblaka implementiran je pomoću CloudSim simulatora otvorenog koda. Zahtjevi ALICE Monte Carlo produkcijskih poslova uneseni su u simulaciju modela u obliku kreiranog radnog opterećenja u standardnom formatu radnog opterećenja (engl. Standard Workload Format, SWF) prilagođenom za rad u odabranom simulatoru. Analizirani su rezultati izvedbe simulacije referentne implementacije pod različitim opterećenjima i uspoređeni su s genetskim algoritmom (engl. Genetic Algorithm) iz iste skupine algoritama. Dobiveni rezultati pokazuju višestruka poboljšanja. Ovdje predložen model obrade podataka omogućava centralizirano softversko upravljanje heterogenom infrastrukturom u oblaku, optimizira mjerene metrike, poboljšava korištenje resursa i postiže skalabilnost sustava.

Wearable devices equipped with inertial and magnetic sensors are increasingly being employed for the monitoring and assessment of physical activity. The affordability, accuracy, and portability of such devices enable real-time monitoring of human movement during exercise, without requiring direct supervision by a trained professional. Furthermore, the data collected by these devices can be analyzed over time, providing valuable insights into an individual's patterns of physical activity and progress. While the initial interest focused mainly on aerobic exercises, research has recently begun to focus on strength exercises as well. The challenges researchers are facing with are primarily related to the minimization of the number and position of wearable devices on individual human body segments and the development of algorithms that will provide appropriate assessment and feedback. Addressing existing limitations and aiming for simplicity, generalizability and reliability, this thesis presents a comprehensive procedure for quantitative and qualitative monitoring of exercise performance, along with a measurement method for assessing human movement variability and movement execution metrics. The proposed method enables the detection, counting, and segmentation of repetitive movements during strength exercises without requiring prior knowledge of the specific exercise being performed. Additionally, the newly developed movement execution metric can identify incorrect repetitions during strength training exercises and distinguishing between experienced and inexperienced individuals. The scientific contributions are: Measurement method of human movement variability during exercise using sensor nodes with inertial and magnetic sensors, Movement execution metrics during exercise based on measurements using sensor nodes with inertial and magnetic sensors, Procedure for quantitative and qualitative monitoring of exercise performance using sensor nodes with inertial and magnetic sensors and measurement of heart rate. Nosivi uređaji opremljeni inercijskim i magnetskim senzorima sve se više koriste za praćenje i procjenu fizičke aktivnosti. Pristupačnost, točnost i prenosivost takvih uređaja omogućuju praćenje ljudskih pokreta tijekom tjelovježbe u stvarnom vremenu, bez potrebe za izravnim nadzorom stručnjaka. Nadalje, tako prikupljeni podaci mogu se analizirati tijekom vremena, pružajući vrijedne uvide u obrasce fizičke aktivnosti i napredak pojedinca. Iako je početni interes istraživačkih grupa uglavnom bio usmjeren na aerobne vježbe, istraživanja su se nedavno počela usmjeravati i na vježbe snage. Izazovi s kojima se istraživači suočavaju primarno su vezani uz minimiziranje broja i položaja nosivih uređaja na pojedinim dijelovima tijela te razvoj algoritama koji će pružiti odgovarajuću procjenu i povratne informacije. Imajući na umu postojeća ograničenja i težeći jednostavnosti, generaliziranosti i pouzdanosti, ova disertacija predstavlja sveobuhvatan postupak za kvantitativno i kvalitativno praćenje izvedbe vježbanja, zajedno s mjernom metodom za procjenu varijabilnosti ljudskog pokreta i metrikom izvođenja pokreta. Predložena metoda omogućuje detekciju, brojanje i segmentaciju ponavljajućih pokreta tijekom vježbi snage, bez potrebe za prethodnim poznavanjem specifične vježbe koja se izvodi. Također, novo razvijena metrika može identificirati neispravna ponavljanja tijekom vježbi snage te razlikovati iskusne i neiskusne pojedince. Ostvareni znanstveni doprinosi su: Metoda mjerenja varijabilnosti ljudskog pokreta tijekom tjelovježbe uporabom senzorskih čvorova s inercijskim i magnetskim senzorima, Metrika izvođenja pokreta tijekom tjelovježbe temeljena na mjerenjima senzorskim čvorovima s inercijskim i magnetskim senzorima, Postupak kvantitativnog i kvalitativnog praćenja uspješnosti vježbanja uporabom senzorskih čvorova s inercijskim i magnetskim senzorima i mjerenjem frekvencije srčanog ritma.

Suvremeni telekom operateri će morati promijeniti pristup poslovanju, organizaciju, razvoj proizvoda i usluga, pristup korisnicima i mnoga druga važna pitanja ako žele biti konkurentni na telekomunikacijskim tržištima u narednim godinama i desetljećima. Oni će imati ključne uloge u ovom razdoblju nazvanom era Industrije 4.0. Postoji mnogo različitih modela analize za telekom operatore, ali svi imaju parcijalni pristup analizi i niti jedan od njih ne daje cjelovitu sliku analize telekoma. Cjeloviti tehno-ekonomski (CTE) model za analizu potencijala telekom operatora je nov i originalan model za analizu i značajno će pomoći u procesima njihove transformacije. Ovaj model omogućit će brzu i jednostavnu analizu potencijala telekom operatora, ali i pojedinih njegovih dijelova, neovisno o ostalim dijelovima tvrtke. Unatoč tome što model lako definira ulazne podatke i što je brz u primjeni analize, daje precizne i matematički definirane rezultate iz kojih se može vidjeti procjena potencijala telekom operatora ili neovisna procjena njegovih pojedinih segmenata. Glavne prednosti ovog modela su jednostavnost, brzina analize telekom operatora, pouzdanost rezultata i njegova modularnost, odnosno neovisna procjena pojedinih dijelova. Takav je model nužan telekom operatorima za brzu i pouzdanu procjenu potencijala, analizu, modeliranje i lakšu prilagodbu novih proizvoda i usluga. Riječ je o jedinstvenom modelu sa znanstvenom pozadinom i teorijskim postavkama, koji pruža praktičnu primjenu na telekomunikacijskom tržištu. Modern telecom operators will have to change their business approach, organization models, products development and services development, customer approach and many other important issues if they want to be competitive on the telecommunication markets in the following years and decades. They will have key roles in this period named the Industry 4.0 era. There are many different models of analysis for telecom operators, but they all have a partial approach to analysis and none of them gives a complete picture of the analysis of telecom operators. The Comprehensive Techno-Economic (CTE) model for the analysis of telecom operator potentials is a new and original model for analysis. It will significantly help telecom operators in their transformation processes. This model will enable a quick and easy analysis of the potential of telecom operators, but also of individual parts of it, regardless of other parts of the company. Despite the fact that the model is easy to define the input data and that it is fast in applying the analysis, it gives reliable and mathematically defined results of the potential of telecom operators or independent assessment of their segments. The main advantages of this model are simplicity, speed of telecom analysis, reliability of results and its modularity, i.e., independent evaluation of individual segments. This model is necessary for telecom operators to achieve fast and reliable potential assessment, analysis, modeling and the easier adaptation of new products and services. This is a unique model with a scientific background and theoretical settings and it enables practical application on the telecommunications market.

Efficient implementation of electric filters is important because it reduces hardware complexity and decreases its power consumption. The most demanding component from both of these aspects is multiplier. Consequently, efficient systems are often designed to minimize number of multipliers or to simplify their structures. The former approach leads to sparse systems. The latter is common in the design of spatial filters, in which high operating frequencies impose additional requirements such as the constraint for low dynamic range ratio of excitation coefficients. Numeric optimization is a common tool used in such designs. However, optimization problems appearing in these cases are nonconvex and therefore difficult to solve. In this dissertation, methods for the design of both sparse systems and systems with constrained dynamic range ratio are considered. In particular, a method for the design of sparse FIR filters constrained in peak-error sense is presented. The method is based on signomial programming which utilizes the lp norm with 0 < p < 1. It is applied in the design of linear phase filters and filters without phase specifications. Furthermore, a method for the design of sparse linear phase FIR filters based on global optimization is described. This method utilizes a branch and bound algorithm with efficient tree pruning. It is suitable for the design of filters constrained in peak- or quadratic-error sense. In the area of spatial filter design, the global methods for nonsparse and sparse design of pencil beams with constrained dynamic range ratio are proposed. The sparse design extends the nonsparse method by allowing the coefficients to take zero values thus improving the obtained radiation patterns. Učinkovita implementacija električkih filtara je važna jer smanjuje potrebne sklopovske resurse i potrošnju energije. Najzahtjevnija komponenta s oba ova aspekta je množilo. Posljedično, učinkoviti sustavi se često dizajniraju s ciljem minimizacije broja množila ili pojednostavljenja njihovih struktura. Prvi pristup vodi na dizajn rijetkih sustava. Drugi je prisutan u dizajnu prostornih filtara gdje visoke radne frekvencije nameću dodatne zahtjeve kao što je zahtjev za niskim dinamičkim rasponom pobudnih koeficijenata. Numerička optimizacija je čest alat za dizajn ovakvih sustava. Nažalost, optimizacijski problemi koji se pritom javljaju su često nekonveksni i stoga teško rješivi. U ovoj disertaciji istražene su metode za dizajn rijetkih sustava i sustava s ograničenim dinamičkim rasponom koeficijenata. Konkretno, predložena je metoda za dizajn rijetkih FIR filtara čija amplitudna karakteristika minimalno odstupa od željene. Metoda se temelji na signomijalnom programiranju koja uključuje lp normu s 0 < p < 1, a primijenjena je na dizajn filtara s linearnom fazom i filtara bez zahtjeva na faznu karakteristiku. Nadalje, razvijena je metoda za dizajn rijetkih FIR filtara pomoću globalne optimizacije. Metoda se temelji na grananju i ograničavanju s učinkovitim odsijecanjem grana stabla. Primijenjena je na filtre čija je amplitudna karakteristika ograničena u smislu maksimalnog ili kvadratnog odstupanja od željene karakteristike. U području dizajna prostornih filtara, predložena je metoda za dizajn rijetkih i nerijetkih antenskih nizova s uskim snopom zračenja i ograničenim dinamičkim rasponom koeficijenata. Rijetki dizajn proširuje nerijetki tako da dopušta pojedinim koeficijentima zauzimanje nulte vrijednosti, što rezultira boljim dijagramima zračenja.

Kako bi se minimizirali gubitci zbog vrtložnih struja, lamelirana transformatorska jezgra se sastoji od velikog broja vrlo tankih feromagnetskih lamela međusobno odvojenih višestruko tanjim slojevima električne izolacije. Očekivano, konačna rješenja za elektromagnetska polja u tako heterogenom području su vrlo oscilatorne, samo po dijelovima kontinuirane prostorne funkcije. Kako bi aproksimacija rješenja bila valjana, izravna primjena metode konačnih elemenata na takav problem zahtijeva iznimno gustu mrežu konačnih elemenata, tj. prekomjeran broj konačnih elemenata iz perspektive raspoloživih računalnih resursa. Prema tome, nužno je provesti neki vid homogenizacije fizikalne slike. U slučaju jezgre otvorenog tipa, situacija se dodatno komplicira zbog neravnomjerne razdiobe elektromagnetskih polja u volumenu jezgre. Značajan dio magnetskog toka prodire u jezgru okomito na plohe lamela, a tako inducirane vrtložne struje poprimaju značajne iznose. Kako bi se smanjio njihov iznos, jezgra otvorenog tipa se obično izvodi od više vitkijih dijelova, tj. laminacijskih paketa, međusobno odvojenih dodatnim slojevima električne izolacije. Sa stajalište računalnih simulacija, takvi slojevi izolacije kvare brzinu konvergencije, a njihova homogenizacija nije moguća korištenjem standardnih metoda. Specifičnosti fizikalne slike kod jezgre otvorenog tipa moguće je uspješno uzeti u obzir korištenjem magnetskog vektorskog potencijala A i strujnog vektorskog potencijala T. U skladu s tim, najprije je izvedena Galerkinova slaba A,T – A-formulacija temeljena na izravnoj primjeni metode konačnih elemenata. Kako bi se znatno smanjila potrebna gustoća diskretizacije domene, unutar svakog pojedinog konačnog elementa su uvedene prikladne aproksimacije fizikalne slike u vidu raspregnutih, međusobno neovisnih elektromagnetskih polja. Na osnovu uvedenih aproksimacija, izvedena je modificirana slaba formulacija koja dopušta znatno veće dimenzije konačnog elementa, ali vrijedi samo lokalno, unutar volumena promatranog konačnog elementa. No, homogenizacijom takve modificirane formulacije dobiva se Galerkinova A,Tx – A-formulacija koja je valjana globalno, unutar cijele problemske domene. Kao posljedica postupka homogenizacije, u području jezgre se koriste nadomjesne homogene i anizotropne značajke materijala. Konačna A,Tx – A-formulacija omogućuje simultano modeliranje sveukupnih vrtložnih struja u jezgri. Dodatno, za razliku od početne slabe A,T – A-formulacije, homogenizirana slaba A,Tx – A-formulacija pokazuje odličnu brzinu konvergencije. Pored analitičkog postupka homogenizacije, razvijen je i numerički pristup homogenizaciji temeljen na 2D formulaciji problema vrtložnih struja kojim se uspješno homogeniziraju i rubni efekti vrtložnih struja, tj. promjene smjera vrtložnih struja koje se javljaju na rubovima lamela. Napravljena je usporedba računalne simulacije temeljene na A,Tx – A-formulaciji s računalnom simulacijom temeljenom na A,φ – A-formulaciji. Dobiveno je dobro slaganje rezultata računalnih simulacija, s tim da simulacija temeljena na A,Tx – A-formulaciji pokazuje bržu konvergenciju. Dodatno, razvijena metoda je primijenjena za izračun gubitaka u realnoj jezgri otvorenoga tipa za koju su prethodno provedena mjerenja ukupnih gubitaka. Rezultati simulacije se dobro slažu s rezultatima dobivenim putem mjerenja, što dodatno potvrđuje valjanost razvijene metode. In order to minimize eddy current losses, the laminated transformer core consists of a large number of very thin ferromagnetic laminations separated from each other by much thinner layers of electrical insulation. As expected, the final solutions for electromagnetic fields in such a heterogeneous region are highly oscillatory, piecewise continuous spatial functions. In order for the approximation of the solution to be valid, the direct application of the finite element method to such a problem requires an extremely dense finite element mesh, i.e., an excessive number of finite elements from the perspective of available computing resources. Therefore, it is necessary to carry out some kind of homogenization of the equations. In the case of an open-type core, the situation is further complicated by the uneven distribution of electromagnetic fields in the core volume. A significant part of the magnetic flux penetrates into the core perpendicular to the surfaces of the laminations, and the eddy currents thus induced have significant values. In order to reduce their amount, the open-type core is usually made of several slimmer parts, i.e. lamination packages, separated from each other by additional layers of electrical insulation. From the point of view of computer simulations, such layers of isolation reduce the speed of convergence, and their homogenization is not possible using standard methods. The specificities of the open-type core can be successfully taken into account using the magnetic vector potential A and the current vector potential T. Accordingly, Galerkin's weak A,T – A-formulation based on the direct application of the finite element method was first derived. In order to significantly reduce the required density of domain discretization, appropriate approximations in the form of uncoupled, mutually independent electromagnetic fields were introduced within each individual finite element. On the basis of the introduced approximations, a modified weak formulation was derived that allows significantly larger dimensions of the finite element, but is valid only locally, within the volume of the observed finite element. However, the homogenization of such a modified formulation gives the Galerkin A,Tx – A-formulation, which is valid globally, within the entire problem domain. As a consequence of the homogenization process, new homogeneous and anisotropic material features are used in the core region. The final A,Tx – A-formulation enables the simultaneous modeling of all eddy currents in the core. Additionally, unlike the initial weak A,T – A-formulation, the homogenized weak A,Tx – A-formulation shows an excellent convergence rate. In addition to the analytical homogenization procedure, a numerical approach to homogenization was developed based on the 2D formulation of the eddy current problem, which takes into account the edge effects of eddy currents, i.e., changes in the direction of eddy currents that occur at the edges of lamin ations. A comparison was made between the simulation based on the A,Tx – A-formulation and the simulation based on the A,φ – A-formulation . A good agreement of the simulation results was obtained, and the simulation based on the A,Tx – A-formulation shows a better convergence rate. Additionally, the developed method was applied to the calculation of losses in a real open-type core for which total losses were previously measured. The simulation results are in good agreement with the results obtained through measurements, which additionally confirms the validity of the developed method.

The research aims to leverage radio frequency (RF) technology for contactless sleep parameters measurement and their analysis using machine learning algorithms and signal processing techniques. The developed platform for sleep monitoring with UWB (Sleep-UWB Platform) was validated through the overnight clinical study in a University Psychiatric Hospital "Vrapče". Analysis of the obtained data collected with the Sleep-UWB Platform was compared with the reference polysomnography measurements. Unlike the common UWB radar approach, this research presents a non-radar solution, where movements and breathing patterns are extracted from the channel impulse response (CIR) of commercial low-cost UWB transceivers. The analysis of the breathing patterns, body postural transition, and overall sleep parameters analysis were investigated. A statistical model was developed and includes information about the position, snoring, sleep stage and the anthropometric measures taken before and after sleep, as well as the subjective questionnaires and behavioral characteristics of the patients on the day of the measurements, covering the lack of information given by AHI index. The goodness-of-fit analysis shows that this approach can bring new insights and improvements in the monitoring of respiratory events. Based on the above mentioned, a scientific contribution has been made in the following manner: • Method for measurement of sleep breathing patterns and sleep postural transitions based on ultra-wideband communication channel impulse response measurement; • Method for feature extraction of breathing patterns and postural transitions during sleep based on ultra-wideband communication channel impulse response measurement; • Method for analysis of sleep parameters from extracted features of breathing patterns and postural transitions. Cilj istraživanja je iskoristiti radiofrekvencijsku (RF) tehnologiju, točnije ultraširokopojasnu (UWB) tehnologiju, za beskontaktno mjerenje parametara spavanja i njihovu analizu korištenjem algoritama strojnog učenja i obradbe signala. Razvijena je UWB platforma za praćenje parametara spavanja (Sleep-UWB Platform) te je validirana provođenjem cjelonoćnog kliničkog istraživanja u Klinici za psihijatriju "Vrapče", Zavoda za psihofiziologiju i organski uvjetovane psihičke poremećaje, Centra za poremećaje spavanja i budnosti. Analiza dobivenih podataka prikupljenih platformom Sleep-UWB Platform uspoređena je s referentnim mjerenjima polisomnografijom (PSG). Za razliku od uobičajenog UWB radarskog pristupa, ovo istraživanje predstavlja ne-radarsko rješenje, gdje se pokreti i disanje izdvajaju iz impulsnog odziva kanala (CIR) komercijalnih UWB primopredajnika. Istražena je analiza obrazaca disanja i promjena položaja tijela tijekom spavanja i sveukupna analiza parametara spavanja. Rezultati istraživanja pokazali su da se predloženim pristupom postižu bolji rezultati u odnosu na postojeće slične metode. Razvijen je statistički model koji uključuje podatke o položaju i fazi spavanja, hrkanju te antropometrijske mjere, kao i podatke dobivene subjektivnim upitnicima i karakteristike ponašanja pacijenata na dan mjerenja, pokrivajući nedostatak informacija dan AHI indeksom. Analiza dobrote prilagodbe modela pokazuje da ovaj pristup može donijeti nove uvide i poboljšanja u praćenju respiratornih događaja. Iz prethodno navedenog ostvaren je znanstveni doprinos kako slijedi: • Metoda mjerenja obrazaca disanja i promjena položaja tijela tijekom spavanja zasnovana na mjerenju impulsnoga odziva ultraširokopojasnoga komunikacijskog kanala, • Metoda izlučivanja značajki obrazaca disanja i značajki promjena položaja tijela tijekom spavanja zasnovana na mjerenju impulsnoga odziva ultraširokopojasnoga komunikacijskog kanala, • Metoda analize parametara spavanja iz izlučenih značajki obrazaca disanja i promjena položaja tijela.

Prostorno-vremenski podatci mogu se vizualizirati na mnogo različitih načina. Za razliku od uobičajenih statičkih prikaza, dinamičke mrežne stranice omogućuju bolje upravljanje prostorno-vremenskim podatcima i lakše otkrivanje novih spoznaja. Napredni prikazi omogućavaju lakše otkrivanje statusa i trenda, pronalaženje ekstremnih vrijednosti u okolišnim podatcima i općenito njihovu bolju iskorištenost. U ovom su radu razvijene i implementirane nove metode prikaza i upravljanja podatcima u mrežnom okolišu. Doprinosi ove disertacije očituju se u tri nove metode: 1) prostorno-vremenske povezane prikaze za uočavanje statusa i trenda u podatcima, 2) prikazi podataka s automatskih mjernih sustava koji omogućuju lakši pronalazak zanimljivih izmjerenih pojava i 3) efikasniji prikaz rezultata numeričkih modela za lakši uvid u detalje i usporedbe s izmjerenim vrijednostima. Za svaku metodu definiran je model arhitekture i implementacije te je pokazano kako nova metoda doprinosi upravljanju podatcima. Implementacija se oslanja na relacijsku bazu podataka, JavaScript i vizualizacije sa klijentske strane. Prikazano je kako nove metode prikaza omogućavaju lakše izvlačenje sadržanih znanja iz podataka koristeći učinkovitu razmjenu podataka između klijenta i servera. Spatio-temporal data, which includes environmental data, can be visualized in many different ways. Dynamic web pages can facilitate data management and the discovery of new knowledge better than static representations. Improved representations make it easier to identify status and trends and find extreme values in environmental data. In this thesis, new methods for data representation and management in web environment were developed and implemented. The contributions of this dissertation manifest in three new methods: 1) spatiotemporal linked representations for observing status and trends in data, 2) representations of data from automatic measuring systems that make finding interesting measured phenomena easier, and 3) improved visualization of numerical model results for easier comparison with measured values and insight into details. For each method, architecture and implementation models are defined and the contribution of the new method to data management is shown. The implementation is based on a relational database, JavaScript, and client-side visualizations. It is shown how these representation methods enable easier extraction of the contained knowledge from data using efficient data exchange between client and server.

Razdvajanje frekvencije i neuskladenost impedancije posljedice su promjene faktora magnetske veze izmedu predajne i prijemne zavojnice odnosno promjene poloˇzaja prijemne zavojnice u odnosu na predajnu zavojnicu u konvencionalnom rezonantnom induktivnom sustavu za beˇziˇcni prijenos energije. Problemi razdvajanja frekvencije i neuskladenosti impedancije u rezonantnom induktivnom prijenosu energije mogu se ublaˇziti generiranjem homogenog magnetskog polja u ve´cini povrˇsine ravnine punjenja. Homogenost magnetskog polja postiˇze se optimiranjem geometrije predajne zavojnice i optimiranjem raspodjele struje izmedu namotaja iste zavojnice. U ovoj disertaciji provodi se postupak odredivanja strukture dvoslojne pravokutne predajne 3D zavojnice za generiranje homogenog magnetskog polja iscrpnim pretraˇzivanjem. Mjerenja magnetskog polja na prototipu takve zavojnice potvrduju homogenu regiju koja ˇcini 55.3% povrˇsine ravnine napajanja. Nakon ˇsto je odredena struktura predajne zavojnice uvodi se pretpostavka da ´ce raspodjela struje izmedu namotaja predajne zavojnice uz optimizaciju geometrije sada poznate strukture pridonijeti poboljˇsanju svojstava predajne zavojnice. Optimizacija se provodi u dva koraka: optimizacija 2D modela i optimizacija 3D modela. Razlog tome je smanjenje prostora pretraˇzivanja 147 B. SAZETAK ˇ odnosno dimenzije problema koji se nastoji rijeˇsiti optimizacijom 3D modela. Optimalne vrijednosti varijabli odluka iz optimizacije 2D modela (geometrijske varijable drugog sloja zavojnice, omjer struja) koriste se kao konstante u optimizaciji 3D modela. Optimizacija 2D modela 3D predajne zavojnice obavlja se u kosimulaciji Python-FEMM uz pomo´c MIDACO alata i genetskog algoritma te diferencijalne evolucije iz pygmo biblioteke. Najbolja rjeˇsenja fitnes funkcije ostvarena su diferencijalnom evolucijom. Nuˇzna je i optimizacija u Ansys Maxwell-u zbog toga ˇsto je to programski paket koji ima opciju simulacije, a takoder i optimizacije 3D modela s pomo´cu alata Optimetrics tool. Odabire se konstanta udaljenost prijenosa od 30 mm. Tako optimirana zavojnica ima bolja svojstva u odnosu na prvotnu. Homogena regija ˇcini 67.78% povrˇsine ravnine napajanja, dubina zavojnice smanjena je za 28.8% te je postignuta ve´ca srednja vrijednost jakosti magnetskog polja u ravnini napajanja. Frequency splitting and impedance mismatching are happening due to magnetic coupling factor variations. i.e. due to relative shift in receiving coil position with respect to transmitting coil in a conventional resonant inductive wireless power transfer system. Frequency splitting and impedance mismatching can be mitigated by generating homogeneous magnetic field intensity at the majority of the charging plane surface. Homogeneity of the magnetic field intensity are achieved by optimization of the transmitting coil geometry and current ratio among the coil turns. In this PhD thesis, the process of determining geometry of the 3D rectangular coil with two layers using exhaustive search is carried out. Measurements that were carried out with prototype of such optimized 3D coil verified that homogeneous region occupies at least 55.3% of the charging plane surface. Then, after the 3D coil structure is determined, a hypothesis is introduced. It is supposed that different current ratio among turns of such coil along with geometry optimization will contribute to achieve better features of known 3D coil structure. Optimization is carried out in two steps: optimization of the 2D model and optimization of the 3D model. Main reason to split optimization is to reduce the search space, i.e. dimension of the problem that is trying to be solved. Optimal va149 B. ABSTRACT lues of decision variables (variables that represent coil second layer geometry and current ratio) obtained from 2D model optimization are used as constants in optimization of 3D model. 2D model optimizations are executed in Python-FEMM co-simulation along with MIDACO tool. Furthermore, genetic algorithm and differential evolution from pygmo library are also applied in solving 2D model optimization. Best fitness function value is acheived using differential evolution strategy. Optimization in Ansys Maxwell is necesarry because of its feature that 3D models can be simulated and also optimized using Optimetrics Tool of Ansys Maxwell. Transfer distance in both 2D and 3D models is set to be 30 mm. Such optimized coil own better features compared to coil obtained by exhaustive search. Homogeneous region now occupies 67.78% of charging plane surface, coil depth is reduced by 28.8% and also the higher value of average magnetic field intensity at charging plane is achieved.

U ovom doktorskom prikazana je primjena naprednih metoda upravljanja i zaštite distributivnih mreža. Zbog razvoja i učestalih promjena u distribucijskoj mreži, potrebno je unaprijediti postojeće metode. Cilj ovoga istraživanja bio je razviti metode i algoritme koji imaju sposobnost automatizirati procese vođenja i zaštite u mreži na optimalan način i time skratiti vrijeme bez napajanja. Primjenom metoda inteligentnog pretraživanja razvijena je metoda adaptivne zaštite u okviru naprednih mreža. Predložen je novi algoritam za rješenje problema optimalne koordinacije i udešenje adaptivne zaštite, te algoritam za optimalnu rekonfiguracije mreže. Na osnovu rezultata algoritma optimalne koordinacije zaštite i algoritma optimalne rekonfiguracije mreže, predložen je model dinamičke adaptivnosti zaštite i algoritam samo-obnavljajuće mreže za zaštitu i vođenje srednjenaponskih mreža u stvarnom vremenu. Na izrađenom laboratorijskom modelu napredne mreže napravljena je simulacije primjene adaptivne zaštite i samoobnavljajuće mreže u različitim pogonskim uvjetima. Primjenom adaptivne zaštite povećava se selektivnost zaštite i raspoloživost sustava, te da se smanjuje vrijeme trajanje i učestalost prekida, i vrijeme isklopa zaštite. In this doctoral thesis, the application of advanced methods of management and protection of distribution networks was investigated. Due to the development and frequent changes in the distribution network, it is necessary to improve the existing methods. The aim of this research is to develop methods and algorithms that have the ability to automate the processes of management and protection in the network in an optimal way and thus shorten the time without power. The application of intelligent search methods will develop the method of adaptive protection within advanced networks. A new algorithm for solving the problem of optimal coordination and adjustment of adaptive protection, and an algorithm for optimal network reconfiguration will be proposed. Based on the results of the algorithm of optimal coordination of protection and the algorithm of optimal reconfiguration of the network, the model of dynamic adaptability of protection and the algorithm of self-healing network for protection and control of medium voltage networks in real time will be proposed. Simulations of the application of adaptive protection and self-healing network in different operating conditions will be made on the developed laboratory model of the advanced network. The application of adaptive protection increases the selectivity of protection and the availability of the system, and to reduce the duration and frequency of interruptions and the time of disconnection of protection.

Utjecaj visokog udjela energetskih pretvarača na dinamiku elektroenergetskog sustava relativno je neistraženo područje. Povećani udio obnovljivih izvora energije priključenih preko pretvarača smanjuje konstantu tromosti elektroenergetskog sustava. Istraživanje je usmjereno prema vrednovanju utjecaja povećane penetracije energetskih pretvarača na dinamičko vladanje frekvencije elektroenergetskog sustava. Osnovni cilj ovog istraživanja je kvantificiranje utjecaja povećane integracije fotonaponskih elektrana (FNE) na stabilnost frekvencije sustava te razvoj algoritma za njihovo sudjelovanje u regulaciji frekvencije. Istraživanje je posebice usmjereno na glavnu značajku regulacije frekvencije (maksimalno odstupanje frekvencije od nazivne vrijednosti) te metodu određivanja potrebnog iznosa rezerve snage s kojom FNE trebaju sudjelovati u regulaciji frekvencije. Da bi se odredila dostatna rezerva snage FNE-a za sudjelovanje u regulaciji frekvencije elektroenergetskog sustava, u disertaciji je razvijena aproksimacijska metoda frekvencijskog odziva elektroenergetskog sustava temeljena na paraboličnoj funkciji. Nakon određivanja potrebnog iznosa rezerve snage, razvijen je upravljački algoritam za sudjelovanje fotonaponske elektrane u regulaciji frekvencije elektroenergetskog sustava u slučaju lančanog poremećaja. Navedeni algoritam upravljanja omogućuje FNE višekratno korištenje rezerve snage u kratkom vremenskom intervalu što u slučaju uzastopnih poremećaja čini FNE sposobnom pružiti doprinos tijekom svakog pojedinog poremećaja. Ovime se značajno povećava stabilnost frekvencije elektroenergetskoga sustava. Cjelokupni model razvijen u disertaciji mogao bi se koristiti kao osnova u simulatorima frekvencije i u planiranju voznog reda elektrana u velikim elektroenergetskim sustavima s visokim udjelom energetskih pretvarača. The impact of a high share of power converters on the power system dynamics is a relatively unexplored research area. The increased share of renewable energy sources connected via converters reduces the inertia constant of the power system. The research is aimed at evaluating the impact of the increased penetration of power converters on the system frequency dynamics. The main goal is to quantify the impact of increased integration of photovoltaic power plants on system frequency stability and to develop an algorithm for their participation in frequency control. The research is particularly focused on the main feature of frequency regulation (maximum frequency deviation from the nominal value) and the method of determination of the required minimum photovoltaic power plant power reserve to maintain the system frequency stability. In order to determine the sufficient power reserve of the photovoltaic power plant for participation in the frequency control, an approximation method of the system frequency response was developed. After determining the required amount of power reserve, a control algorithm was developed for the participation of the photovoltaic power plant for power system frequency support in case of cascading events. The control algorithm aims to enable rapid recovery of the photovoltaic power plant power reserve which makes the photovoltaic system capable of providing support to the cascading disturbance events. This significantly improves the frequency stability of the power system. The model developed in the thesis could be used as a basis in frequency response simulations and in scheduling power plants in large power systems with a high proportion of power converters.

Most robotic manipulators, and especially those designed with autonomous operation in mind, consist of a series of joints that rotate about a single axis, also known as revolute joints. These mechanisms give robotic manipulators the degrees of freedom and versatility similar to that of the human arm, which they are designed to outperform. However, this results in a geometry of motion or kinematics that makes all aspects of robotic manipulation challenging from a computational perspective. A major part of this challenge lies in the fact that computing joint configurations adhering to a specific set of constraints (i.e., gripper pose) is a non-trivial problem. The procedure of finding feasible joint configurations and the mathematical problem associated with it are known as inverse kinematics - a core part of motion planning, trajectory optimization, calibration and other important challenges in successfully performing robotic manipulation. In recent years, the overall decrease of computation time required to perceive and process environmental and proprioceptive information has helped realize the potential of robotic manipulation in dynamic environments. Concurrently, a new standard in manipulator design has emerged, where additional degrees of freedom are added in order to increase their overall dexterity and capacity for motion. These two developments have vastly increased the requirements for inverse kinematics algorithms, which are now expected to deal with infinite solution spaces and difficult, nonlinear constraints. On the other hand, the addition of degrees of freedom in recent robot designs has enabled algorithms to search for locally optimal configurations with respect to some performance criteria in an infinitely large solution space. This property has motivated approaches that leverage non-Euclidean geometries to replace conventional constraints and optimization criteria, thereby overcoming computational bottlenecks and common failure modes. The contributions presented in this thesis propose three such approaches, that aim to develop new ways of looking at the problems associated with inverse kinematics through the use of geometric representations that are not widely utilized in robotic manipulation. Većina robotskih manipulatora, posebno onih dizajniranih za autonomno obavljanje zadataka, sastoji se od niza zglobova koji se okreću oko jedne osi rotacije. Ovi mehanizmi omogućuju robotskim manipulatorima stupnjeve slobode slične onima ljudske ruke, što je ujedno i cilj pri njihovom projektiranju. Međutim, ovi mehanizmi rezultiraju geometrijom gibanja ili kinematikom, koja čini sve aspekte robotske manipulacije izazovnima iz računske perspektive. Glavni dio ovog izazova leži u činjenici da je izračun konfiguracija zglobova koje se pridržavaju određenog skupa ograničenja (npr. položaja izvršnog člana) netrivijalan problem. Postupak pronalaska izvedivih konfiguracija zglobova i matematički problem asociran s tim postupkom poznati su kao inverzna kinematika, te čine ključni dio planiranja kretanja, optimizacije putanje, kalibracije i drugih važnih izazova pri uspješnom izvođenju robotske manipulacije. Nedavno smanjenje vremena izračuna potrebnog za opažanje i obradu okolišnih i proprioceptivnih informacija pomoglo je ostvarenju potencijala robotske manipulacije u dinamičnim okruženjima. Istodobno se postavlja novi standard u dizajnu manipulatora koji uključuje dodavanje redundantnih stupnjeva slobode sa ciljem povećanja sposobnosti kretanja. Ova su dva čimbenika uvelike povećala zahtjeve za algoritme inverzne kinematike, od kojih se sada očekuje operacija pri beskonačnim prostorima rješenja i teškim, nelinearnim ograničenjima. S druge strane, redundantni stupnjevi slobode omogućuju algoritmima potragu za lokalno optimalnim konfiguracijama s obzirom na specifične kriterije izvedbe u beskonačno velikom prostoru rješenja. Ovo svojstvo je motiviralo pristupe koji koriste neeuklidsku geometriju za svladavanje konvencionalnih ograničenja i kriterija optimizacije, čime se prevladavaju računska uska grla i razni uobičajeni problemi pri optimizaciji. Doprinosi obrazloženi u ovoj disertaciji predlažu tri takva geometrijska pristupa. Njihov je cilj razvoj novih načina gledanja na probleme povezane s inverznom kinematikom korištenjem geometrijskih prikaza koji još nemaju široku upotrebu u robotskoj manipulaciji.

Low back pain (LBP) is a global health problem phenomenon. Most patients are categorized as non-specific, thus requiring an individualized approach which still poses a major challenge. Myoelectric recordings from two pairs of lumbar muscle sites were collected during an isometric trunk extension exercise. Ninety-one subjects were included in the study; 29 patients with non-specific chronic LBP (CLBP), 25 patients with radiculopathy (RLBP), and 37 control healthy subjects (HS). Six best-performing time-domain raw features were employed to model contextual secondary feature groups. Neuromuscular LBP characteristics were described with coordination, co-activation, trends, and frequency-based fatigue measures. The initial large secondary feature set was significantly reduced by employing neighborhood component analysis (NCA), alongside preserving features in the original interpretable domain. Support vector machines (SVM) and k-nearest neighbor (kNN) consistently resulted in high classification accuracy, among the 23 classifiers exploited. Based on thirteen NCA components, from CLBP vs. HS model, CLBP subgrouping was performed by exploiting Hierarchical clustering analysis (HCA), resulting in seven CLBP subgroups. The formal procedure for assigning new CLBP patients, to one of the subgroups, was enabled by introducing a new distance metric CLBP Pattern Distance (CPD). As a final step, for each CLBP subgroup, a corresponding medical interpretation was proposed, thus enabling a direct connection between mathematical procedure and the deeper inference from the medical perspective to account for the complex multifactorial interaction among different neuromotor control, biomechanical, or even psychological aspects. Križobolja (LBP) je prepoznata kao globalni zdravstveni problem. Većina pacijenata se kategorizira kao nespecifična, stoga je nužan individualizirani pristup koji još uvijek predstavlja veliki izazov. Elektromiografske snimke, s dva para pozicija lumbalnih mišića, prikupljene su tijekom izometrijske vježbe ekstenzije trupa. U istraživanje je bio uključen devedeset i jedan ispitanik; 29 bolesnika s nespecifičnim kroničnim LBP-om (CLBP), 25 bolesnika s radikulopatijom (RLBP) te 37 kontrolnih zdravih ispitanika (HS). Analizom dobivene najučinkovitije jednostavne značajke iz vremenske domene, njih šest, upotrijebljene su za modeliranje grupa kontekstualnih sekundarnih značajki. Neuromuskularne karakteristike LBP-a su opisane kroz grupe značajki koordinacije, koaktivacije, trendova i mjera umora temeljenih na frekvencijskoj analizi. Inicijalno veliki skup sekundarnih značajki je značajno reduciran tehnikom analize komponenti metodom susjedstva (NCA), uz očuvanje značajki u izvornoj interpretabilnoj domeni. Metoda potpornih vektora (SVM) i klasifikatori temeljeni na k-najbližem susjedstvu (kNN) konzistentno su rezultirali visokim klasifikacijskim točnostima, iz skupa 23 upotrijebljena klasifikatora. Koristeći trinaest NCA komponenti, dobivenih iz modela diferencijacije CLBP u odnosu na HS, provedeno je podgrupiranje CLBP pacijenata koristeći metodu grupiranja hijerarhijskom analizom, što je rezultiralo s ukupno sedam CLBP podskupina. Formalni postupak za dodjelu novih pacijenata s CLBP-om u jednu od sedam podskupina, osiguran je uvođenjem nove metrike udaljenosti, CLBP Pattern Distance (CPD). Kao posljednji korak, za svaku podskupinu CLBP-a predloženo je odgovarajuće medicinsko tumačenje, čime je omogućena izravna veza između matematičkog postupka i dubljeg zaključivanja iz medicinske perspektive, a kako bi se objasnila složena multifaktorijalna interakcija između različitih aspekata neuromotoričkih kontrole, biomehaničkih ili čak psiholoških aspekata.

Ova disertacija rezultat je istraživanja u području upravljanja sinkronim strojevima bez mjernog člana brzine vrtnje za potrebe uključenja pretvarača dok je stroj u vrtnji, što se naziva leteći start. Potreba za letećim startom postoji kod vjetroagregata, pumpi i ventilatora. Izvedba takvih sustava bez mjernog člana brzine vrtnje dovodi do povećanja robusnosti te smanjenja troškova investicije. U disertaciji je dan pregled literature za algoritme estimacije brzine vrtnje i položaja rotora, s posebnim osvrtom na algoritme koji se primjenjuju za leteći start. Izvedeni su matematički modeli sinkronog stroja s umetnutim permanentnim magnetima te sinkronoga reluktantnoga stroja potpomognut s permanentnim magnetima. Opisano je upravljanje sinkronim strojevima tijekom trajnog načina rada, radna područja, vektorsko upravljanje orijentacijom magnetskog toka i vektorska modulacija širine impulsa. Izvorni znanstveni doprinos ove disertacije vezan je uz leteći start sinkronoga stroja s permanentnim magnetima i leteći start sinkronoga reluktantnoga stroja primjenom isprekidanoga načina rada energetskoga pretvarača, a sastoji se od: Matematički model sinkronoga stroja s permanentnim magnetima za isprekidani način rada energetskoga pretvarača u svrhu kontinuirane estimacije brzine vrtnje i položaja rotora. Matematički model sinkronoga reluktantnoga stroja za isprekidani način rada energetskog pretvarača u svrhu kontinuirane estimacije brzine vrtnje i položaja rotora. Metoda upravljanja isprekidanim načinom rada energetskog pretvarača tijekom perioda sinkronizacije pretvarača i sinkronoga stroja u vrtnji. Predloženi algoritmi estimacije brzine vrtnje i položaja rotora kao i algoritam upravljanja strujom kratkoga spoja implementirani su u digitalni sustav i eksperimentalno provjereni na laboratorijskom postavu. Eksperimentalna ispitivanja potvrđuju učinkovitost implementiranih algoritama, prvenstveno mogućnost kontinuirane estimacije brzine vrtnje dok se ona mijenja te održavanje struje tijekom letećeg starta na zadanom iznosu. This dissertation is the result of research in the sensorless control of synchronous machines to switch on the inverter while the machine is rotating, which is called a flying start. The need for a flying start exists in wind turbines, pumps, and fans. Sensorless control increases robustness and reduces investment costs. The dissertation reviews the literature on sensorless algorithms, i.e., rotor position and speed estimation, with particular reference to flying start algorithms. Mathematical models of a permanent magnet synchronous machine and a permanent magnet-assisted synchronous reluctance machine were derived. Control of synchronous machines in continuous operation, operating ranges, field-oriented vector control, and space vector pulse-width modulation are described. The original scientific contribution of this dissertation is related to the flying start of a synchronous machine using the discontinuous mode of operation of the power converter and consists of the following: Mathematical model of a permanent magnet synchronous machine for the discontinuous mode of operation of the power converter for continuous rotor speed estimation. Mathematical model of synchronous reluctance machine for the discontinuous mode of operation of power converter for continuous rotor speed estimation. Method for controlling the discontinuous mode of operation of the power converter during the flying start. The proposed rotor speed and position estimation algorithms and the short-circuit current control algorithm were implemented in a digital control system and experimentally verified in a laboratory setup. Experimental tests confirm the effectiveness of the implemented algorithms, primarily the possibility of continuous estimation of the rotor speed while it changes and maintaining the current during the flying start at the reference value.

U doktorskom radu obrađena je problematika vezana za optimizaciju parametara PSS stabilizatora (engl. Power System Stabilizer) i AVR regulatora (eng. Automatic Voltage Regulator) kroz upotrebu modeliranja elektroenergetskog sustava (EES-a) i simuliranja rada sustava u specifičnim operativnim stanjima. Operativna stanja su vezana za ispitivanje kutne stabilnosti i uključuju realne pogonske situacije kao što su privremeni ispad interkonekcijskog voda, kratkotrajni ispad opterećenja, tropolni kratki spoj ili uklop dodatnog tereta. Kroz simulaciju se koristi višeciljna metaheuristička optimizacija sa svrhom nalaženja optimalnih parametara za koje sustav treba reagirati sa što manjim oscilacijama odziva. Predviđeni algoritmi su algoritmi starije i novije generacije metaheurističke inteligencije: VRPSO (eng. Velocity Relaxed Particle Swarm Optimization), ALO (eng. Ant Lion Optimizer), SSA (eng. Salp Swarm Algorithm) i SA (eng. Simulated Annealing). Ciljne funkcije su vezane za integral umnoška vremena i odstupanja promatranih veličina, te pokazatelje kvalitete prijelaznog procesa. U višeciljnoj optimizaciji tih dviju funkcija koristi se analiza Pareto optimalnosti dva stanja. This dissertation discusses problems and proposes methods regarding parameter optimization of PSS (Power System Stabilizer) and AVR (Automatic Voltage Regulator) devices in power system. This problem is inspected by using power system modelling and simulation of various operating conditions. These simulations are a part of rotor angle stability inspection, and they present real-world power system contingency problems such as: interconnection transmission line outage, short term load outage, short circuit, additional load injection, etc. Multiobjective swarm optimization is applied with purpose of finding parameters optima for which a system should respond with oscillations as minimal as possible. Algorithms predicted for use in such optimization are VRPSO (velocity relaxed particle swarm optimization), ALO (Ant Lion Optimizer), SSA (Salp Swarm Algorithm) and SA (Simulated Annealing). Used objective functions include ITAE (integral of time multiplied with absolute error) and time domain transitional process quality indicators, which are analyzed through Pareto optimality.

Agregatori fleksibilnosti potrošnje predstavljaju nove dionike na tržištu energije koji su zaduženi za agregiranje fragmentiranih izvora fleksibilnosti i njihovu valorizaciju na tržištu. Ovaj rad razmatra mogućnost valorizacije fleksibilnosti potrošnje u cilju ostvarivanja ušteda energije. Za tu potrebu razvijen je model koji s relativno malo prikupljenih podataka daje mogućnost ocjene učinka takvog programa. Određivanje dostupne fleksibilnosti u kućanstvima otežava nedostupnost povijesnih podataka o potrošnji uređaja prikladne granularnosti i preciznosti. Metoda se zasniva na ekstrapolaciji iz minimalnog dostupnog skupa podataka, analizira ključne parametre koji utječu na potrošnju, primjenjuje regresijske modele te omogućava kvantifikaciju tako dobivene fleksibilnosti. Razvijeni model je prilagođen tipičnim obrascima korištenja kućanstava i potrošnje u mediteranskim zemljama gdje je dostupnost daljinskog grijanja uobičajeno manjkava. Rezultati koji su dobiveni iz modela gdje se prioritiziraju energetske uštede uz minimalni utjecaj na komfor korisnika, povoljni su za ona kućanstva čije zgrade pokazuju bolja energetska svojstva. U okviru ovog rada predloženi su i indikatori koji bi u stvarnom okruženju mogli olakšati kvantifikaciju fleksibilnosti u okviru standardiziranih procesa za praćenje i verifikaciju. Local flexibility aggregators can be a source of innovative aggregation solutions at the local level. The unleashed flexibility potential of households could create new revenue streams for local flexibility aggregators. This thesis considers the possibility of valorisation of short-term flexibility activations in a form of energy savings. A model has been developed for this purpose, which with minimal available collected dataset offers the possibility of evaluating the effect of such a program. The assessment of available flexibility within household’s could be arduous due to the lack of historical energy consumption data of the flexibility assets with the appropriate granularity and precision. The method is based on the extrapolation from the minimal available data set, analyses key parameters that affects consumption, applies regression models, and allows the quantification of the obtained flexibility. The developed model is adapted to typical consumption and behavioural patterns of households in Mediterranean countries where the availability of district heating network may be scarce. The results obtained from the model, where energy savings are prioritized with minimal impact on user comfort, show favourable results for those households whose buildings show better energy properties. As part of this research, indicators that could facilitate the quantification of flexibility within the standardized monitoring and verification processes in the real environment are proposed.

Za potrebe rada prikupljen je veliki uzorak terenskih mjerenja visokofrekvencijskih elektromagnetskih polja na 4252 lokacije od 25512 mjernih točaka u Republici Hrvatskoj u periodu od 2016. do 2018. godine. Istraživanje je temeljeno na najvećem uzorku mjerenja visokofrekvencijskih elektromagnetskih polja javnih izvora u frekvencijskom rasponu od 30 MHz do 3 GHz na području Republike Hrvatske. Mjerenja su napravljena prema kompleksnoj metodi mjerenja koja se sastoji od sveukupnog skupa znanja i prakse prema aktualnim međunarodnim standardiziranim metodama, zakonima, pravilnicima, iskustvima i dokumentaciji u okviru rada akreditiranog laboratorija. Na temelju obrađenih rezultata mjerenja napravljena je klasifikacija izvora visokofrekvencijskih elektromagnetskih polja temeljena na proračunu kumulativnih kvocijenata izloženosti s ciljem poboljšanja zaštite ljudi i okoliša od mogućeg utjecaja visokofrekvencijskih elektromagnetskih polja. Klasifikacija svih lokacija izvora je napravljena na temelju procjene rizika u odnosu na granice najviše dopuštene vrijednosti kumulativnih kvocijenata izloženosti postavljene od strane Međunarodne komisije za zaštitu od neionizirajućeg zračenja (engl. International Commission on Non-Ionizing Radiation Protection, ICNIRP). Ustanovljeno je da najveći dio lokacija ima vrlo male ili male rizike od utjecaja visokofrekvencijskih elektromagnetskih polja dok dio lokacija umjerenih rizika zahtjeva posebnu pozornost. Svega nekoliko lokacija u uzorku prelaze granice najviše dopuštene vrijednosti kumulativnih kvocijenata izloženosti. Na temelju tih rezultata unaprijeđena je metodologija sustava nadzora visokofrekvencijskih elektromagnetskih polja u svrhu zaštite ljudi i okoliša u okviru društvene odgovornosti u području visokofrekvencijskih elektromagnetskih polja. Istraživanjem je ustanovljeno da procjene visokofrekvencijskih elektromagnetskih polja dodatnih izvora nisu realne te da bi trebale biti komplementarne mjerenjima. U tu svrhu, kako bi se verificirali rezultati mjerenja i kako bi se zaključilo da su rezultati procjena prihvatljivi, u radu su izložena četiri algoritma kojima je potvrđeno da procjenama treba prethoditi nulto mjerenje razina visokofrekvencijskih elektromagnetskih polja prije instalacije novih izvora. U svrhu racionalizacije metodologije nadzora, tome može koristiti originalni i izvorni doprinos supstitucije nultog mjerenja uz pomoć prvog mjerenja i točnog poznavanja karakteristika novoinstaliranih izvora neionizirajućih visokofrekvencijskih elektromagnetskih polja. Research presented in this dissertation is based on a large in-situ measurement sample of high-frequency electromagnetic field measurements performed on 4252 locations during three years’ time in Republic of Croatia in the vicinity of electromagnetic field sources. On each location six measurements of high-frequency electromagnetic fields from public sources in frequency range from 30 MHz to 3 GHz were carried out, which gives a total of 25512 measuring points. Measurements implemented requirements according to complex measurement method which consists of an advancing set of knowledge and best practice integrated with relevant international standardized methods, laws, ordinances, professional experiences and procedures documented in the framework of the accredited testing laboratory. All locations in the sample are classified in relation to high-frequency electromagnetic fields exposure and based on cumulative exposure quotients calculations with a purpose to improve human and environment protection from possible high-frequency electromagnetic field effects. Classification of all locations is based on risk assessment in respect to the limit of maximum permitted cumulative exposure quotient value set by the ICNIRP (International Commission on Non-Ionizing Radiation Protection). It has been established that predominant risk-based classification groups of exposure to high-frequency electromagnetic fields sources are very small risk and small risk locations while a part of moderate risk locations require special attention. Five locations in a sample belong to significant (unacceptable) risk classification group. On the basis of these results a new monitoring high-frequency electromagnetic field methodology is proposed as a part of social responsibility. On the basis of the research, it has been established that, for the purpose of installing additional electromagnetic field sources, high-frequency electromagnetic field assessment is not rational and it should be complementary with measurements. Therefore, this research presented four algorithms, which give confirmation that exposure assessment should be done on the basis of zero-measurement high-frequency electromagnetic fields before any new source installation, to verify measurement results and to conclude whether assessment results are acceptable. For the purpose of monitoring methodology rationalisation, zero-measurement substitution, given in this research, is used. Zero-measurement substitution is obtained from first-measurement with knowledge and understanding of newly installed high-frequency electromagnetic field sources characteristics.

U ovoj doktorskoj disertaciji predstavljena su tri nova regulacijska sustava s izmjenjivačem kvazi Z-tipa napajanim iz fotonaponskog izvora. Prvi sustav je bez baterija te izmjenjivač radi u spoju s mrežom, dok su druga dva sustava s baterijama a izmjenjivač radi u jednom slučaju u spoju s mrežom, a u drugom u otočnom načinu rada. Za upravljanje izmjenjivačem kvazi Z-tipa korištena je nova pulsno-širinska modulacija, s dodanim trećim harmonikom i implementiranim mrtvim vremenom, kod koje je početak prostrijelnog stanja sinkroniziran s početkom nultog stanja i koja rezultira značajnim smanjenjem poluvodičkih gubitaka. Razvijena su dva nova algoritma za izračun poluvodičkih gubitaka izmjenjivača kvazi Z-tipa upravljanog novom pulsno-širinskom modulacijom, čija je točnost eksperimentalno provjerena. Prvi novi regulacijski sustav s izmjenjivačem kvazi Z-tipa napajanim iz fotonaponskog izvora i spojenim na električnu mrežu osigurava praćenje točke maksimalne snage bez oscilacija i bez mjerenja struje fotonaponskog izvora. Simulacijska analiza ovog sustava provedena je za novi dinamički model fotonaponskog izvora s dvije diode i s uračunatim efektom kapaciteta osiromašenog područja i difuzijskog kapaciteta svake od dioda. Drugi novi regulacijski sustav s baterijama potpomognutim izmjenjivačem kvazi Z-tipa u otočnom načinu rada za traženje točke maksimalne snage fotonaponskog izvora umjesto struje izvora koristi struju baterija, dok treći novi regulacijski sustav s baterijama potpomognutim izmjenjivačem kvazi Z-tipa, u mrežnom načinu rada, za istu svrhu koristi d-komponentu vektora struje mreže. Za sintezu oba regulacijska sustava s baterijama primijenjene su eksperimentalno potvrđene prijenosne funkcije dobivene na temelju novog lineariziranog usrednjenog matematičkog modela sustava. This doctoral thesis presents three novel control systems of a photovoltaic-fed quasi-Z-source inverter. The first system does not contain batteries and the inverter is grid-tied, whereas the other two contain batteries and operate both in the stand-alone mode in one case and in grid-tied mode in the other. The quasi-Z-source inverter is controlled by means of a novel pulse-width modulation with injected third harmonic and implemented dead-time, where the start of a shoot-through state is synchronized with the start of a zero-switching state and which results in a significant reduction of semiconductor losses. Two novel semiconductor loss-calculation algorithms were developed for the quasi-Z-source inverter with the novel pulse-width modulation and their accuracy was experimentally evaluated. The first proposed control system of the photovoltaic-fed, grid-tied quasi-Z-source inverter ensures the maximum power point tracking without oscillations and without measurement of the photovoltaic source current. A simulation analysis of this control system was carried with the novel dynamic two-diode model of the photovoltaic source, which accounts for diffusion and junction capacitances of both diodes. The second proposed control system of the battery-assisted, stand-alone quasi-Z-source inverter utilizes the battery current instead of the photovoltaic source current for the maximum power point tracking of the photovoltaic source, whereas the third proposed control system of the battery-assisted, grid-tied quasi-Z-source inverter utilizes the d-axis grid current for the same purpose. The design of both these control systems was carried out by means of the experimentally validated transfer functions, which were obtained based on the novel small-signal model of the respective system.

Električni strojevi projektiraju se optimirano s maksimalnim iskorištenjem materijala. Posljedica toga je radno područje stroja u nelinearnom području magnetske karakteristike željeza i velikim strujnim opterećenjima. U tom radnom području induktiviteti u stroju izrazito su promjenjivi s radnim točkama opterećenja stroja, te u takvim uvjetima određivanje granica dozvoljenog radnog područja velikih proizvodnih jedinica složen je problem. Prezentiran je niz postojećih objavljenih načina za izračun utjecaja nelinearnosti na odnose elektromagnetskih veličina u stroju. U traženju najjednostavnijeg i najefikasnijeg načina opisivanja stacionarnih radnih stanja sinkronih strojeva koji bi obuhvatio magnetsko zasićenje (magnetic saturation) i efekt poprečne magnetizacije (cross-magnetization effect), modeli sa strujno ovisnim funkcijama ulančenih tokova (u daljnjem tekstu Ψ(I) funkcije) pokazali su se najpraktičnijima. U slučaju opisivanja ulančenih tokova polinomnim Ψ(I) funkcijama osnovna prednost u postavljanju modela je ta da se identifikacija koeficijenata složenog nelinearnog sustava svodi na rješavanje linearnog sustava jednadžbi. Jednom identificiran model nadalje se jednostavno koristi pomocu polinomnih Ψ(I) funkcija u svim pogonskim stanjima na kojima je model identificiran. Međutim, u daljnjem testiranju polinomnih Ψ(I) funkcija uocena je njihova nezgodna karakteristika s kojom znatno gube točnost odmah izvan područja u kojem su identificirane, premda su vrlo točne unutar tog područja. ˇ Osnovni problem pri postavljanju modela stacionarnih stanja sinkronog generatora čija namjena bi bila izračunavanje granica pogonske karte je u tome što traženi model treba čim točnije opisivati ponašanje stroja u rubnom području oko granica pogonske karte i izvan dozvoljenog područja, a u tom području u praksi na velikim generatorima u komercijalnoj upotrebi nije dozvoljeno niti moguće napraviti mjerenja na kojima bi se taj model identificirao. Zbog registriranih nepovoljnih karakteristika polinomnih Ψ(I) funkcija istaknuta je potreba za pronalaženjem takvog modela koji bi na osnovi poznatog ponašanja stroja u dozvoljenom području rada čim točnije opisivao ponašanje stroja u nedozvoljenom području rada, odnosno izvan granica pogonske karte. Prezentiran je model sigmoidnih Ψ(I) funkcija, čiji oblik odgovara izgledu nelinearne krivulje magnetiziranja aktivnih dijelova sinkronog generatora, koja je odraz magnetskog zasićenja i drugih nelinearnih pojava, poput efekta poprečne magnetizacije. Takvom funkcijom omogućava se bolje predviđanje nelinearnih magnetskih pojava u stroju, izvan dozvoljenog područja u kojemu je model identificiran. In an electric power system, the most efficient and cost-effective way of voltage regulation is the one that provides corrective measures closest to the part of the network where the consumption is, i.e., the cause of voltage condition deterioration and where the deviations from the nominal voltage are significant. This way, the unnecessary flows of reactive power are reduced, which produces active losses, overwhelms the network, and thus reduces the transmission capacity for the primary line task, which is the transmission of active power. Therefore, in a classic strategy of the electric power system control, synchronous generators have been used as sources of active power, while the rest of the equipment, such as capacitor banks, thyristor controlled capacitors (TSC), inductances (TSR), and synchronous compensators were used to produce reactive power and to regulate voltage in the network. Due to their abundance and distributed arrangement in all parts of the network, the most appropriate solution is to use the existing production units for this purpose. Due to their capability of working in inductive and capacitive areas with continuous, quick, and precise regulation, they present the best solution for regulating voltage conditions in an electric power system. Historically, the purpose of a power plant was focused on producing active power, while the production of reactive power was not considered as a relevant option. Because of this, the operating points of the generator in real conditions could be kept conservatively in a safe area, far from the permitted reactive power border area. Synchronous generator manufacturers provide capability diagrams of generators with fixed limits based on calculated data from the classic vector-phasor diagram with constant reactances and are verified once as a part of warranty tests during machine handovers. Users of the units were not motivated to precisely determine safe limits by the reactive power coordinate, which considers complex electromagnetic effects in the machine that can be confirmed with measurements. Due to this, the possible generator production capacity was idle in large zones of the generator capability diagram, while it could safely be utilized and contribute to the stability of voltage conditions of the electric power system, avoiding voltage breakdown or reducing over-voltages.

U ukupnim gubicima hidrogeneratora s istaknutim polovima, ventilacijski gubici sudjeluju s približno 10% do 25%, a u iznimnim slučajevima mogu dosegnuti i iznos od 30%. Pouzdan proračun tih gubitaka izazovan je zadatak za svakog projektanta. Istovremeno, neistraženost utjecaja različitih elemenata konstrukcije na strujanje zraka kroz stroj i nastanak ventilacijskih gubitaka dovela je do zanemarivanja aerodinamičkih problema pri konstruiranju ovih, u pravilu, velikih strojeva. Ovo istraživanje je provedeno s ciljem da se rasvijetle utjecaji pojedinih konstrukcijskih elemenata generatora na ventilacijske gubitke i stvore preduvjeti za povećanje pouzdanosti izračuna ventilacijskih gubitaka brzim analitičkim alatima koji se redovito koriste pri projektiranju. Na početku rada, provedena je statistička analiza izmjerenih ventilacijskih gubitaka za 27 izvedenih generatora te je predložen empirijski model za preliminarni izračun ventilacijskih gubitaka u ranoj fazi projektiranja strojeva koji se bazira na osnovnim geometrijskim značajkama rotora. Zatim je izrađeno šest detaljnih CFD modela generatora nad kojima je provedena detaljna aerodinamička analiza strujanja rashladnog zraka uz pomoć komercijalnog programa za računalnu dinamiku fluida, ANSYS Fluent®. Validacija odabranog pristupa modeliranja strujanja zraka kroz generatore provedena je usporedbom rezultata izmjerenih gubitaka s rezultatima simulacija. Kako bi se pojasnili određeni efekti uočeni na modelima izvedenih stanja generatora, za pojedine generatore izrađeni su dodatni modeli s izmijenjenom konstrukcijom, tako da su simulacije provedene za sveukupno 9 različitih izvedbi generatora. Analiza strujanja zraka kroz modele, te gubitaka koje pojedini elementi konstrukcije generiraju, pokazali su vrlo kompleksnu ovisnost o izvedbi različitih dijelova konstrukcije generatora. Rezultat su neželjeni efekti, kao što su vrtlozi, koji smanjuju očekivano djelovanje ugrađenih tlačnih elemenata čija je svrha ostvariti cirkulaciju rashladnog zraka kroz generator. Zanemarivanje tih efekata pri izradi analitičkih modela za izračun strujanja zraka kroz generator, može dovesti do krivog izračuna raspodjele zraka kroz generator i posljedično, do krivog izračuna ventilacijskih gubitaka. Isto tako, neizbalansirani otpori povratnih putova za zrak na pogonskoj i slobodnoj strani stroja rezultiraju neželjenim recirkulacijama, posebno ako se ostave otvorenima razni tehnološki i konstrukcijski raspori u rotoru. Analiza utjecaja glavnih elemenata rotora na generiranje ventilacijskih gubitaka pokazala je da na tlačne elemente ugrađene na rotor otpada prosječno oko 80% ukupnog iznosa. Preostalih 20% su parazitni gubici koje uglavnom stvaraju istaci na čeonim plohama rotora. Od onih 80% koje stvaraju tlačni elementi nešto više od polovice stvaraju polovi. Provedena je analiza mogućnosti izračuna karakteristika tlačnih elemenata rotora pojednostavljenim CFD modelima te su rezultati uspoređeni s karakteristikama koje se sada koriste u modelima. Pokazano je da se takvim jednostavnim modelima mogu dobiti kvalitetne podloge za izračun ventilacijskih mreža. Na temelju rezultata simulacija, izračunati su koeficijenti otpora vijenca ukruta i matica koje se često ugrađuju na čelo rotora. Predloženi su empirijski izrazi za izračun koeficijenta otpora takvih elemenata rotora u ovisnosti o glavnoj dimenziji i razmaku između susjednih ukruta/matica. Predloženi izraz uspoređen je s podacima iz literature. Spoznaje koje su proizašle iz ovog istraživanja pružaju dobru podlogu za razvoj unaprijeđenih alata za analitičke izračune ventilacije i ventilacijskih gubitaka. Isto tako, trebale bi poslužiti kao smjernice pri konstruiranju generatora sa smanjenim ventilacijskim gubicima u odnosu na one koji se danas grade. Na kraju rada su predložene izmjene u postojećim modelima ventilacijskih mreža kako bi se povećala pouzdanost rezultata izračuna. Introduction According to [1], with approximately 1300 TW of installed capacity and about 4200 TWh of annual production, hydropower is currently the most important renewable energy source in the world. Salient-pole generators are key components in hydropower conversion. Although very efficient machines (for large units the efficiency exceeds 98%) the inevitable losses that occur in the machine (mostly electromagnetic in nature) need to be carried away from machine to keep the temperature of the active parts within the allowable limits. This is usually achieved by means of the circulating air that collects excessive heat and transfers it to the cooling water in air-water heat exchangers. Air circulation is achieved by mounting an axial or a centrifugal fan on the rotor and/or by using existing rotor structural elements that act as primitive fans, i.e., rotor spider, interpolar space and radial ducts in the rotor rim. The inevitable consequence of air circulation through the generator are aerodynamic losses, which are also called windage losses. According to [2], windage losses are defined as: "the power taken from the generator shaft (without friction in the bearings) when the rotor rotates at a synchronous speed, at a certain flow of coolant (air)". Windage losses participate in the total losses of salient poles generators with approximately 10% to 25%. For old, high-speed and/or aerodynamically non-optimized generators, this value can reach up to 30%. Windage losses importance is illustrated by the fact that, due to the synchronous speed of rotation, their amount is constant regardless of the load at which the machine operates. The motivation for this research is unreliable models used to calculate windage losses. Prerequisite for windage loss calculations is solving the airflow networks (also called lumped parameter models), that are used to calculate the distribution of the air flow through the generator. The calculation of windage losses is based on the assumption that the total windage losses consist of the sum of friction losses of the rotor side faces and rotor end faces and the power taken by the rotor pressure elements from the turbine shaft to achieve the required cooling air flow through the generator. The aim of this research was to investigate aerodynamic phenomena within salient poles generators, to distinguish the influence of layout and geometric characteristics of the most important structural elements on generating windage losses and to determine new correlations for calculating resistance coefficients of most influential structural elements depending on basic geometric features. Furthermore, the research was aimed at defining guidelines for the aerodynamic design of those parts of the generator that are unavoidable in construction and actively participate in the generation of windage losses. Materials and methods This research is based on a statistical analysis of the basic geometric and aerodynamic parameters of the 27 manufactured and installed generators and on a subsequent detailed aerodynamic analysis of the cooling air flow through a CFD models of subset of six selected generators. Analysis of air flow through the generator was performed using the commercial CFD software ANSYS Fluent®. Detailed CFD models have been made for each selected generator. Fully predictive approach was used to model the flow within the generator meaning no inlet or outlet boundary conditions were needed. The basic idea behind this principle is to incorporate a closed air circle in the model while the airflow was driven only by the rotor rotation. Thus, the airflow created through the generator is a result of the solution and not of the boundary conditions. The RANS k-ω SST turbulence model was used in all simulations. Its implementation in Fluent® utilized the y+ insensitive wall treatment and as long as enough prism layers covered the boundary layer high quality numerical results could be achieved [3]. Since weak stator-rotor interaction was expected, rotation of the rotor domain was modelled using multiple reference frame method. The rotational periodic boundary condition on the side faces of the domain and “no slip” boundary condition on all walls were applied. The non-conformal interface approach [3] was used to connect the rotor domain with the stator. Special attention was paid to the quality of the mesh and the resulting minimal orthogonal quality of all meshes was higher than 0.14. To evaluate and quantify the numerical uncertainty arising from the discretization of the domain, the grid convergence index (GCI) method [4] was performed for validation case 3 on two generator examples with three different density meshes. Three validation cases were used to validate the selected simulation approach. First two cases were simple examples from the literature: smooth rotating disc [5] and rotating disc with protrusion [6]. Validation showed good agreement with published data (within ±3,5% for the first case, and ±5,0% for the validation second case). The third validation case was a subset of the 6 generators tested on-site. A measurements of the windage losses were performed during the comissioning of the generators according to the IEC 60034-2-2:2012 standard. Agreement of ±6,5% between calculated and measured windage losses were achieved. Results and discussion Basic statistical analysis of measured windage losses on a set of 27 built and tested generators showed that the average relative windage losses (reduced to the rated power of the generator) is 0.293% with a standard deviation of 0.1157%. Slightly less than 80% of the considered generators have relative windage losses in the range between 0,15% and 0,40%. No visible trend was observed when windage losses were correlated to the rated power of the generator. On the other hand, a linear trend is visible between measured windage losses and expressions ω3·D5 and ω3·D4·L which denote the measure of rotor end and side faces losses. Achieved coefficients of determination were R2=0,903 and R2=0,908, respectively. In order to improve the fit of the data, least squared fit to expression a·ωb·Dc·Ld was proposed, where a, b, c and d are fitting coefficients. Achieved coefficient of determination was R2=0,974, which represents an acceptable empirical model for first estimation of the windage losses in the early stage of the design process. The overall analysis of the calculated windage losses for selected set of six generators showed that an average of 80% of the losses can be attributed to the pressure rise elements used to establish airflow through the generator. Of these, more than 50% belongs to interpolar space, which makes them main single source of windage losses in salient poles generators. It is worth noting that the highest share of the poles in generation of windage losses is observed in ventilation system with axial fans (80%), and the lowest in radial ventilation without fans (30%). On average, 20% of the total windage losses is associated with end faces of the generator. These losses are called parasitic losses and are associated with the roughness of the end faces. For rough rotors, with large protrusions, they can reach more than 30%. A detailed analysis of the air flow through the generator models indicated significant interactions that occur between the different structural elements installed on the rotor. Two main issues detected are existance of different air recirculations that are main source of unreliability of existing calculation models and distribution of tangential velocity profile within the generator. Some airway details have been clarified leading to proposed changes to existing ventilation networks which are expected to increase the reliability of the ventilation calculation model. Based on the calculated aerodynamic forces on stiffeners and nuts mounted on end faces of the rotor, empirical model for calculation of the drag coefficient is proposed. In order to improve existing models for calculation of characteristics of the rotor pressure rise elements (rotor spider, radial channels in rotor rim and interpolar space) it has been shown that simplified CFD models of these rotor structures can lead to more reliable characteristics. Due to the simple geometry of these rotor elements, the construction of such models can be automated and integrated into new, improved models for calculating ventilation. Finally, changes in the existing models for the calculation of ventilation and windage losses are proposed, which are expected to lead to more reliable models for their calculation. Scientific contribution 1. An empirical model for the preliminary calculation of windage losses in the early design phase of a salient poles generators is proposed. 2. By comparing the results of measurements and numerical simulations on six generator models, it was determined that using a detailed 3d CFD generator model it is possible to calculate windage losses and airflow within engineering acceptable limits of ± 6.5%. 3. Design elements of construction that significantly affect the creation of windage losses in salient poles generators have been determined, and the mechanisms of their action in the air throughflow the generator have been clarified. 4. An empirical expressions for the calculation of the drag coefficients for stiffeners and nuts often mounted on the rotor end faces are proposed. 5. Guidelines to redefine existing airflow networks in order to reduce deviations of the calculation of ventilation and windage losses from those obtained by existing models are proposed. Conclusion The conducted research confirmed the initial hypotheses and achieved the research goals. Analysis of airflow through different models of the generators revealed the roles of main rotor components in generation of windage losses. A new correlation for estimation of the windage losses in the early stage of the design process is proposed. Interactions between different rotor components and their influence on the formation of vortices that affect the energy conversion pressure elements used to establish cooling airflow through the generator are explained. The air paths through the generator are explained and based on that, changes in the existing models are proposed in order to obtain more reliable models for calculating the air distribution through the generator and thus a more reliable calculation of windage losses.

The energy sector is characterized by a growing share of variable renewable energy sources, primarily solar and wind power plants, and the spread of electrification in the heating, transport and industry sectors. Such trends require an increase in flexibility in the power system, which can be achieved by developing smart grids and new business models. The dissertation explores the potential of local energy markets to mitigate unfavourable consumption/production patterns and contribute to maintaining voltages and currents within limits. Furthermore, if local energy trading can increase the competitiveness of local flexibility providers and improve social welfare, meaning it could serve as a mean to enhance the integration of renewable energy sources in electrical distribution networks through price incentives for supply/demand balancing. The original scientific contribution was achieved as follows: 1. Short-term model for near real-time local electricity market over a distribution network; 2. Method for determining the effects of local energy trading on voltage profiles and power flows over electrical distribution networks; and 3. Method for assessing the impacts of the local electricity market on market participants. Energetski sektor karakterizira rastući udio varijabilnih obnovljivih izvora energije, prvenstveno sunčanih i vjetrenih elektrana, te širenje elektrifikacije u sektorima grijanja, prometa i industrije. Ovakvi trendovi zahtijevaju povećanje fleksibilnosti u elektroenergetskom sustavu što se može postići razvojem naprednih mreža i novih poslovnih modela. Disertacija istražuje potencijal lokalnih energetskih tržišta za ublažavanje nepovoljnih obrazaca potrošnje i proizvodnje te doprinos održavanju napona i struja unutar granica u distribucijskim mrežama. Nadalje, modelira se i analizira mogućnost lokalnog trgovanja za povećanje konkurentnosti lokalnih pružatelja usluga fleksibilnosti i poboljšanje društvenog blagostanja, odnosno njihov potencijal za poboljšanje integracije obnovljivih izvora energije u elektroenergetske distribucijske mreže putem cjenovnih poticaja za uravnoteženje ponude i potražnje. Izvorni znanstveni doprinos postignut je na sljedeći način: 1. Kratkoročni model za lokalno tržište električne energije u gotovo stvarnom vremenu na razini distribucijske mreže; 2. Metoda za određivanje utjecaja lokalnog trgovanja električnom energijom na profile napona i tokove snaga u lokalnoj distribucijskoj mreži; i 3. Metoda za procjenu učinka lokalnih tržišta električnom energijom na tržišne sudionike.

Cilj ove doktorske disertacije je izraditi te osposobiti sustav za simuliranje i prototipiranje uredaja iz podruˇcja elektroenergetike ili elektrostrojarstva te na njemu implementirati modele elektromotornog pogona s razliˇcitim algoritmima upravljanja asinkronog stroja. Sustav takoder mora biti omogu´cen za rad u kosimulaciji s razliˇcitim programskim okruˇzenjima i alatima iz podruˇcja mekog raˇcunarstva. Sustav je realiziran pomo´cu stvarno-vremenskog simulatora te je iskoriˇsten za prototipiranje frekvencijskog pretvaraˇca za napajanje asinkronog stroja. Predstavljena je napredna metoda za regulaciju brzine vrtnje asinkronim strojem, a u kosimulaciji s meta-heuristiˇckim optimizatorom, sustav je iskoriˇsten za optimiranje parametara neizrazitog regulatora brzine vrtnje asinkronog stroja. U prvom poglavlju, predstavljeni su matematiˇcki modeli svih elemenata elektromotornog pogona koji su bitni za uspjeˇsno simuliranje dinamike elektromotornih pogona pomo´cu stvarno-vremenskog simulatora. U drugom poglavlju, prikazani su dijelovi sustava za simuliranje i prototipiranje upravljaˇckih algoritama elektromotornih pogona. Glavni dijelovi ovog sustava su stvarno-vremenski simulator s proˇsirenjem za digitalne i analogne ulaze i izlaze te energetski dio koji se sastoji od filtera energetskih modula za izradu frekvencijskog pretvaraˇca. Pomo´cu sustava je izvrˇsena simulacija elektromotornog pogona sa skalarno upravljanim asinkronim strojem te stvarna skalarna regulacija asinkronog stroja. Simulirani i stvarni odzivi su potom usporedeni. U tre´cem poglavlju, predstavljeni su konvencionalne i napredne metode upravljanja brzinom vrtnje asinkronog stroja. Od naprednih metoda izdvojeno je prediktivno upravljanje strujom i momentom asinkronog stroja na temelju modela. Prediktivno upravljanje strujom asinkronog stroja je pobliˇze analizirano te prilagodeno za rad s konstantnom frekvencijom uklapanja frekvencijskog pretvaraˇca. Glavni doprinos pri razvoju ove tehnike upravljanja je dokaz stabilnosti klasiˇcnom metodom analiziranja svojstvenih vrijednosti i ˇcinjenica da nema potrebe za raˇcunanjem teˇzinskih faktora koji inaˇce osiguravaju stabilnost sustava. Razvijena metoda je usporedena sa srodnom metodom upravljanja te je vidljivo poboljˇsanje u spektru elektromagnetskog momenta asinkronog stroja. U ˇcetvrtom poglavlju izvrˇsena su dodatna poboljˇsanja metode prediktivnog upravljanja strujom, gdje je PI regulator brzine vrtnje zamijenjen s neizrazitim regulatorom. Parametri neizrazitog regulatora su optimirani pomo´cu meta-heuristiˇckog optimizatora. Optimizacija je izvrˇsena kosimulacijskim pristupom na naˇcin da stvarno-vremenski simulator izvrˇsava simulaciju elektromotornog pogona s neizrazitim regulatorom, a optimizator prima rezultate na temelju kojih generira optimalne parametre. Odzivi elektromotornog pogona upravljanog s optimiranim neizrazitim regulatorom su usporedeni s odzivima dobivenim srodnim metodama te je vidljivo poboljˇsanje odziva momenta i brzine vrtnje. The goal of this PhD thesis is to develop a system for simulation and prototyping of devices from the fields of electrical or electromechanical engineering and use it for implementation of electric drive models that utilize different structures for induction machine control. The system should also allow for cosimulation with different software environments and tools from the field of soft computing. This system is realized using the real-time simulator and it is used for induction machine frequency converter prototyping. Advanced method for induction machine speed control is presented. System is also used for optimization of induction machine fuzzy speed controller utilizing meta-heuristic optimizer through cosimulation process. First chapter presents mathematical models of all electric drive elements that are relevant for successful dynamical simulation using the real-time simulator. Chapter two presents parts of the rapid control prototyping system used for electric drives. Main parts of this system are real-time simulator with the extension for digital and analog inputs and outputs, and power part that consists of passive filters and energy modules for frequency converter realization. Using the system, a model in the loop simulation of scalar controlled electric drive is executed and then scalar control is tested on the real machine. Simulated and real responses are then compared. The third chapter presents conventional and advanced induction machine control methods with the emphasis on model predictive control methods, specifically model predictive torque, and model predictive current control. Model predictive current control is closely analyzed and adapted for operation with constant modulation frequency. Main contribution regarding this algorithm is proof of stability using the classic eigenvalue analysis, which makes weighing factor matrix that is usually used to ensure stability obsolete. Proposed method is compared with another method from the family of model predictive control methods and the improvement in electromagnetic torque spectrum is clearly visible. Chapter four represents additional improvements of induction machine predictive current control where speed PI controller gets replaced with fuzzy speed controller. Parameters of the fuzzy speed controller are optimized using meta-heuristic optimizer. Optimization is executed through cosimulation where real-time simulator executes electric drive simulation, and optimizer receives the results based on which it generates new set of parameters. Induction machine responses controlled by fuzzy controller are compared with the responses obtained by several different methods which shows improvement in electric drive speed and torque responses.

Ovaj doktorski rad predstavlja istraživanje bezsenzorskog upravljanja vjetrogeneratorom s permanentnim magnetima temeljeno na opserveru u kliznom načinu rada. Prvi dio disertacije donosi pregled najčešće korištenih topologija vjetroagregata, fizikalne osnove pretvorbe energije vjetra, matematički model sinkronog generatora s permanentnim magnetima te standardne algoritme upravljanja AC-DC-AC pretvaračem. Nakon toga, dan je uvod u teoriju kliznih režima. Zatim je opisan laboratorijski postav na kojem je izvršena eksperimentalna verifikacija predloženih algoritama. Iza opisa laboratorijskog postava, predstavljen je znanstveni doprinos disertacije koji se dijeli na tri dijela: 1) metoda za bezsenzorsko upravljanje sinkronim vjetrogeneratorom s permanentnim magnetima temeljena na opserveru u kliznom načinu rada, estimiranom rotirajućem koordinatnom sustavu i modelu aktivnog magnetskog toka vjetrogeneratora, 2) metoda za leteći start sinkronog vjetrogeneratora s permanentnim magnetima temeljena na opserveru u kliznom načinu rada prilagođenom isprekidanom radu energetskog pretvarača i 3) metoda za adaptaciju induktiviteta u poprečnoj osi rotirajućeg koordinatnog sustava primjenom aproksimativne funkcije s ciljem smanjenja estimacijske greške opservera u kliznom načinu rada. Razvijene metode eksperimentalno su potvrđene. Laboratorijski postav sastoji se od sinkronog generatora s permanentnom magnetima, pretvarača i asinkronog stroja. Upravljački algoritmi za pretvarač napisani su u programskom jeziku C i implementirani u digitalni upravljački sustav temeljen na procesoru Delfino C28343. This dissertation presents a study of sensorless control of a permanent magnet wind generator based on a sliding mode observer. The first part gives an overview of the most common topologies of wind energy conversion systems, the physical principles of wind energy conversion, a mathematical model of a synchronous generator with permanent magnets, and the standard algorithms for controlling AC-DC-AC converters. This is followed by an introduction to sliding mode theory. Then, the laboratory setup in which the proposed algorithms were verified is described. Finally, the scientific contribution of the thesis is presented, which is divided into three parts: 1) a method for sensorless control of a permanent magnet wind generator based on a sliding mode observer modeled in an estimated rotating coordinate system using a model of the active magnetic flux of the rotor, 2) a method for sensorless flying start of a permanent magnet wind generator based on a sliding mode observer adapted to the discontinuous operation of the converter, and 3) a method for reducing the estimation error of the sliding mode observer based on the approximation function for fitting the q-axis inductance. The developed methods were experimentally confirmed. The laboratory setup consists of a permanent magnet wind generator, a converter and an asynchronous machine. The control algorithms for the converter were written in the C programming language and implemented in a digital control system based on the Delfino C28343 processor.

Pētījumā atklāta elektrotehnikas speciālista, Rīgas Politehnikuma (RP) / Rīgas Politehniskā institūta (RPI) mācībspēka (1884– 1918), profesora (1906) Nikolaja Ozmidova (Nikolai Ozmidoff; Николай Озмидов; 1850–1938) daudzpusīgā darbība dažādos amatos – uz dzelzceļa Šveicē un Vācijā, ģeodēzijā, mērniecībā, elektrotehnikā Rīgā, kā arī pedagoģiskā un zinātniskā darbība RP / RPI un Ivanovovozņesenskas Politehniskajā institūtā, Krievijā. Autore veidojusi rakstu, izmantojot arhīva dokumentus, zinātnisko literatūru un publikācijās presē.The research covers the multi-faceted activities of Nikolai Ozmidoff (Николай Озмидов; 1850–1938), a specialist in electrical engineering, member of acade­mic staff (1884–1918) of Riga Polytechnicum (RP) / Riga Polytechnic Institute (RPI), professor (1906), in various positions – working on the railway in Swit­zerland and Germany, in geodesy, surveying, electrical engineering in Riga, as well as his pedagogical and scientific activities at RP / RPI and Ivanovo-Vozne­sensk Polytechnic Institute in Russia. Developing the article, the author used archival documents, scientific literature, and publications in the press.

U sklopu doktorskog rada postignut je model višeagentskog sustava za kontekstno usmjeravanje flote električnih vozila s ciljem maksimizacije preferenci korisnika koji se može koristiti za poboljšanje održivosti u transportnim sustavima i u korištenju gradskih procesa i resursa. Višeagentski model sastoji se od modela višeagentskog sustava za usmjeravanje električnih vozila koji uključuje postupke kontekstnog usmjeravanja flote električnih vozila uzimajući u obzir prostorne i vremenske uvjete kretanja vozila i obavljanja poslova, i s njima povezane funkcionalnosti električnih punionica i spremišta vozila. Postupci kontekstnog usmjeravanja flote električnih vozila uključuju korištenje modela umjetne iskoristivosti za predviđanje mjere iskoristivosti električnih punionica i resursa odredišta na kojima električna vozila obavljaju poslove. Ostvareni model višeagentskog sustava za usmjeravanje električnih vozila korišten je za ostvarivanje platforme za simulaciju transportnog sustava flote električnih vozila koja omogućuje evaluaciju postupaka usmjeravanja pri promjeni kontekstnih uvjeta te usporednu analizu postupaka usmjeravanja u definiranim kontekstnim uvjetima. Znanstveni doprinos sastoji se od modela višeagentskog sustava za usmjeravanje električnih vozila koji se sastoji od pokretnih agenata - električna vozila te stacionarnih agenata - električne punionice i spremište vozila, postupaka kontekstnog usmjeravanje flote električnih vozila, te višeagentske platforme za simulaciju transportnog sustava flote s ciljem provođenja usporednih analiza. This dissertation presents a model of a multi-agent system for context-aware routing of electric vehicle fleets with the goal of maximizing user preferences, which can be used to improve sustainability in transportation systems and in the use of urban processes and resources. The multi-agent model consists of a model of a multi-agent system for electric vehicle routing, which includes methods for context-aware routing of electric vehicle fleet considering spatial and temporal conditions of vehicle movement and performance, as well as related functionality of electric charging stations and vehicle depots. The methods for context-aware routing of the electric vehicle fleet include the use of artificial utilization models to predict the utilization of charging stations and destination resources where the electric vehicles are operated. The implemented multi-agent system for electric vehicle routing was used to create a platform for simulating the electric vehicle fleet transportation system, enabling the evaluation and comparative analysis of routing methods under changing contextual conditions. The scientific contribution consists of a model of a multi-agent system for electric vehicle routing consisting of mobile agents - electric vehicles - and stationary agents - electric charging stations and vehicle depot, procedures for context-aware routing of electric vehicle fleets, and a multi-agent platform for simulating the fleet transportation system with the aim of comparative analysis.

Neregularna ustaljena stanja neželjena su pojava u radu istosmjernih pretvarača. Uzrokuju povećanje valovitosti napona i struja pretvarača i akustičku i strukturnu buku. Kvaziperiodično ustaljeno stanje nastalo bifurkacijom srazom granica uzrokuje značajno povećanje vršne i efektivne vrijednosti struje prigušnice naponski upravljanog istosmjernog uzlaznog pretvarača. To povećava gubitke pretvarača, prvenstveno prigušnice i strujno opterećenje pretvaračkih komponenata pa može doći do kvara pretvarača. Zato je takav način rada potrebno izbjeći. To se postiže matematičkim modeliranjem pretvarača kojim će se predvidjeti bifurkaciju srazom granica i štetne posljedice te pojave te omogućiti odabir parametara pretvarača, izvora i trošila da se izbjegne bifurkacija srazom granica koja uzrokuje kvaziperiodično ustaljeno stanje. Matematički model pretvarača definiran je tako da je simuliran utjecaj svakog od parametara modela na pojavu neregularnih ustaljenih stanja. Tako je utvrđeno koji parametri pretvarača imaju presudnu ulogu u pojavi neregularnih ustaljenih stanja. Matematičkim modelom pretvarača postignuto je predviđanje Hopfovih bifurkacija i bifurkacija srazom granica. Te bifurkacije uzrokuju slijedove ustaljenih stanja do sada nezabilježene u literaturi. Ustaljena stanja dobivena simulacijama identificirana su temeljem valnih oblika, bifurkacijskih dijagrama, trajektorija i Poincaréovih slika. Bifurkacijski dijagrami simulirani su tako da se za idući bifurkacijski parametar koriste početni uvjeti postignuti pri prethodnom bifurkacijskom parametru. Tako je postignuto predviđanje histereze prve Hopfove bifurkacije. Pri pojavi bifurkacija na fizičkom modelu pretvarača uočene su histereze ustaljenih stanja u području Hopfove bifurkacije i bifurkacije srazom granica. Bifurkacijom srazom granica nastaje značajno povećanje struje prigušnice pa je histereza bifurkacijske točke prve bifurkacije srazom granica od tehničkog značaja. Uz to, pretvarač iskazuje multistabilnost u cijelom području ulaznih napona za koje se pojavljuju kvaziperiodična ustaljena stanja. Ustaljena stanja dobivena mjerenjem identificirana su iz valnih oblika, trajektorija i harmonijskog sastava. U vremenskom području opisano je kako nastaje bifurkacija srazom granica. Utvrđena je pravilnost u promjeni ustaljenih stanja temeljem odnosa upravljačkih signala i signala nosioca. Na osnovi toga pokazano je izbjegavanje bifurkacije srazom granica kad pretvarač radi u ustaljenom stanju osnovne periode. The nonstability steady-states of the voltage controlled boost converter are an unwanted phenomena. These nonstability steady-states cause increased ripple values as well as acoustic and structural noise. In the voltage mode controlled DC-DC boost converter a significant increase of effective currents and peak currents is caused by quasiperiodic steady-state obtained by border collision bifurcation. Furthermore, the converter efficiency is decreased, and component stresses are increased. The quasiperiodic steady-state obtained by border collision bifurcation can lead to converter failure. Therefore, it is important to predict the bifurcation and avoid it, by proper adjustment of the converter parameters. In order to do this, it is necessary to build a mathematical model of the converter that predicts these nonstability steady-states. The mathematical model of the converter is built by changing each of the parameters of the model separately. In that way it is determined which of the converter's parameters lead to the emergence of the nonstability steady-states. The chosen mathematical model successfully predicts border collision bifurcation and Hopf bifurcation. These bifurcations are the cause of a new phenomena and new steady-states not previously known to power electronic engineers. The steady-states obtained by the simulations are identified by waveforms, trajectories, Poincaré maps and bifurcation diagrams. The initial values at each input voltage of bifurcation diagrams are calculated using previous values of input voltages. In this way the hysteresis of Hopf bifurcation was successfully predicted. The hysteresis of Hopf bifurcation and border collision bifurcation was identified in the boost converter. The converter also exhibits multistability in the whole range of its quasiperiodic behaviour. The border collision bifurcation significantly increases the inductor current. Hence, the border collision bifurcation and its hysteresis have significant technical relevance. The steady-states obtained by measurements were identified by waveforms, trajectories and harmonic analysis. In the time domain, it is described how the border collision bifurcation occurs. The regularity in the change of steady-states was determined on the basis of the relationship between control signals and carrier signals. The avoidance of border collision bifurcations when the converter is operating in period-one steady-state was demonstrated.

Izraz kardiovaskularne bolesti (KVB) odnosi se na brojne funkcionalne abnormalnosti srca i krvožilnog sustava. KVB uzrokuju gotovo jednu trećinu (33%) smrtnosti u suvremenom svijetu, što predstavlja najveći udio u odnosu na sve druge bolesti. Rana dijagnoza i odgovarajuće liječenje kardiovaskularnih bolesti mogu značajno smanjiti smrtnost i poboljšati kvalitetu pacijentova života. Postavljanje dijagnoze temelji se na cjelokupnoj slici kardiovaskularnog sustava, uključujući anatomiju i fiziologiju srca. Dijagnostički proces obično se sastoji od dva glavna dijela. Prvi dio odnosi se na prikuplanje slika srca pomoću medicinskih uređaja. Razvijene su brojne invazivne i neinvazivne tehnike medicinskog snimanja koje omogućuju uvid u anatomiju i funkcionalnost srca. Drugi dio dijagnostičkog procesa je kvantifikacija i interpretacija prethodno dobivenih slika pomoću naprednih metoda obrade slike. Razvoj učinkovitih metoda za obradu medicinskih slika je složen zadatak, s obzirom da podrazumijeva obradu ogromne količine visokodimenzionalnih podataka. Napredak u razvoju algoritama obrade slike, računalnog vida i umjetne inteligencije, kao i dostupnost grafičkih procesorskih jedinica (GPU-a), značajno su olakšale i ubrzale razvoj takvih metoda. Segmentacija medicinskih slika ima važnu ulogu u procjeni, dijagnozi te postavljanju prognoze različitih kardiovaskularnih bolesti. Opsežna istraživanja i kliničke primjene pokazale su da računalna tomografija (CT) i magnetska rezonanca (MRI), kao osnovne tehnike prikupljanja medicinskih slika, imaju izrazito važnu ulogu u procjeni kardiovaskularnih bolesti. Njima je omogućeno kvantificiranje bolesti, mjerenje volumena kao i analizira morfologije različitih organa. Prema tome, segmentaciju srca i srčanih struktura predstavlja osnovu za širok spektar kliničkih primjena. Primjerice, često se koristi se za modeliranje i analizu anatomije i funkcionalnosti kao i za lokalizaciju različitih patologija. Izrada trodimenzionalnog (3D) modela srca specifičnog za pojedinog pacijenta predstavlja izrazit potencijal za poboljšanje kirurškog planiranja za pacijente s urođenom srčanom manom. Kako bi se takvi 3D modeli mogli izraditi, potrebno je imati segmentirane različite srčane strukture, uključujući pojedine srčane komore, epikardijalnu površinu, aortu kao i pojedine žile kardiovaskularnog sustava. Segmentacija lijeve i desne klijetke ima izrazito važnu ulogu u kvantitativnoj analizi globalnih i regionalnih informacija, odnosno pokazatelja rada srca, poput xviii volumena na kraju dijastole (VKD), volumena na kraju sistole (VKS), frakcije izbacivanja (FI), debljine stijenke ili mase. Primjerice, ventrikularna hipertrofija uzrokovana je abnormalnim povećanjem srčanog mišića koji okružuje lijevu ili desnu klijetku. Prema tome, segmentacija cijelog srca i srčanih komora iz volumetrijskih medicinskih slika igraju bitnu ulogu u procjeni cjelokupnog kardiovaskularnog zdravlja. Nadalje, radiolozi često trebaju ocrtati aortu kako bi dobili njezinu morfologiju, što je bitno za otkrivanje i dijagnosticiranje aneurizme aorte. Ručna segmentacija srca i srčanih struktura je vremenski veoma zahtijevan posao, podložan subjektivnosti. Prema tome, razvoj točnih i robusnih automatskih algoritama za segmentaciju je neophodan za primjenu u kliničkoj praksi. Duboko učenje predstavlja najsuvremeniju metodu za različite zadatke obrade slike poput raspoznavanja, segmentacije i klasifikacije. Metode dubokog učenja temelje se na umjetnim neuronskim mrežama. Najčešće upotrebljena vrsta neuronske mreže su konvolucijske neuronske mreže (CNN). FCNs predstavljaju specifičnu vrstu CNN-a bez potpuno povezanog sloja, kojima se obrađuje cijela slika te nije potrebno korištenje patcheva. Razvijene su različite varijante FCN-a, od kojih su najznačajnije varijante koje koriste koder-dekoder arhitekture. U biomedicinskoj obradi slika, za segmentaciju, najčešće se koristi U-Net arhitektura neruonske mreže kao i njezina odgovarajuća 3D verzija. U-Net arhitektura ima snažnu reprezentativnu snagu te je u mogućnosti zabilježiti značajke niskih razina što je izrazito važno prilikom treniranja mreže sa malom količinom podataka. Iako U-Net ima snažnu reprezentativnu snagu, dugoročni odnosi između značajki su slabi zbog upotrebe konvolucijskih operacija. Prema tome, potrebno je razvijati naprednije mehanizme kao i dodatne blokove koji će biti u mogućnosti ispraviti nedostatke U-Net arhitekture. Tehnike i blokovi poput veza za preskakivanje ili dubokog nadzora, omogućuju izgradnju dubljih arhitektura neuronskih mreža koje pružaju apstraktnije rezultate učenja te postižu veću točnost prilikom segmentacije medicinskih slika. S obzirom da povećanje broja slojeva osigurava veći prostor parametara koji omogućuje učenje apstraktnijih značajki, dublje arhitekture neuronskih mreža pružaju apstraktnije učenje koje rezultira boljim performanse i većom točnost u zadacima medicinske segmentacije. Unatoč tome, kako se dubina mreže povećava, informacije o gradijentu prolaze kroz mnogo slojeva te mogu nestati ili nakupiti velike pogreške do trenutka kada gradijet dosegne kraj mreže. To dovodi do uobičajenih prepreka treninga dubokih arhitektura neuronskih mreža kao što su problem nestajajućih gradijenta, ekstenzivnog rasta parametara, kao i smanjenja točnosti, što dovodi do računalno zahtjevnih modela. U ovoj doktorskoj disertaciji, predložen je niz metoda dubokog učenja za automatsku segmentaciju srca i srčanih komora. Fokus disertacije je na poboljšanju metoda dubokog učenja za segmentaciju cijeloga srca, lijeve i desne klijetke i miokarda kao i aneurizme abdominalne aorte. S xix obzirom na karakteristične probleme koji se javljanju prilikom dizajniranja metoda dubokog učenja za segmentaciju medicinskih slika, poput problema visoke dimenzionalnosti slika koje rezultiraju treniranim modelima s velikim brojem parametara kao i nedostatkom anotiranih podataka za treniranje, cilj ove disertacije je ublažiti navedene izazove predlaganjem novih i robusnih arhitektura neuronskih mreža koje smanjuju broj korištenih parametara, ali zadržavaju izrazito visoku točnost krajnjih rezultata segmentacije. Prvi i najvažniji znanstveni doprinos predstavlja nova struktura povezivanja rezidualnih jedinica, koju nazivamo rezidualna jedinica za spajanje značajki (FM-Pre-ResNet). FM-Pre-ResNet struktura povezivanja rezidualnih jedinica dodaje konvolucijski sloj na vrh i na dno već postojećih prethodno aktivirajućih rezidualnih jedinica. Pri tome, gornji sloj uravnotežuje parametre dviju grana rezidualne jedinice, dok donji sloj smanjuje dimenzije kanala. Na ovaj način predložena struktura povezivanja rezidualnih jedinica omogućuje kreiranje značajno dubljih modela uz održavanje iste ili čak manje količine parametara u odnosu na originale rezidualne jedinice. Nakon toga, u drugom znanstvenom doprinosu, predložena je nova 3D arhitektura neuronske mreže bazirana na koder-dekoder arhitekturi koja uspješno integrira FM-Pre-ResNet jedinice s varijacijskim autokoderima (VAE) za segmentaciju srca i srčanih komora iz CT i MRI slika. Metoda se sastoji od tri osnovna dijela. U prvom dijelu, prethodno predložene FM-Pre-ResNet jedinice koriste se za učenje nisko-dimenzionalnog prikaza ulaza u fazi kodiranja. U drugom dijelu, VAE rekonstruira ulaznu sliku iz nisko-dimenzionalnog latentnog prostora, osiguravajući da su sve težine modela snažno regulirane, kako bi se izbjegnula neželjena pojava pretreniranja. VAE dio koristi se samo tijekom treniranja mreže. Konačno, u trećoj fazi dekodiranja ponovno su integrirane FM-Pre-ResNet jedinice pomoću kojih se stvaraju konačne segmentacije srca. Predložena nova arhitektura evaluirana je na testnom skupu podataka koji se sastoji od 40 različitih pacijenata dostupnih kroz MICCAI Multi-Modality Whole Segmentation Challenge (MM-WHS) izazov. Naša metoda ostvarila je prosječni DSC, JI, SD i HD za cijelo srce od 90,39%, 82,24%, 1.1093 i 15,3621 na CT snimkama, odnosno 89,50%, 80,44%, 1,8599, 25,6558 na MRI snimkama. Predloženi pristup ostvario je približno slične rezultate kao i najsuvremenije metode za segmentaciju cijelog srca na CT slikama dok su rezultati na MRI slikama bolji od rezultata prethodno objavljenih najsuvremenijih metoda. Treći znanstveni doprinos, predstavlja novu automatsku metodu za segmentaciju miokarda (MiO), lijeve (LK) i desne klijetke (DK) iz cineMRI slika. Predstavljena je nova arhitekturu koja integrira SERes blokove u 3D U-net arhitekturu (3D SERes-U-Net). SERes blokovi upotrebljavaju operacije stiskanja i uzbude u rezidualne jedinice. Sposobnost ponovne kalibracije značajki operacija stiskanja i uzbude povećava reprezentativnu snagu mreže, dok ponovna upotreba značajki xx koristi učinkovito učenje o značajkama, što poboljšava performanse segmentacije. Predloženu metodu evaluirali smo na testnom skupu podataka MICCAI Automated Cardiac Diagnosis Challenge (ACDC). Naša predložena metoda za segmentaciju pomoću 3D SERes-U-Net ostvarila je prosječni DSC za LK, DK i MiO na kraju dijastole od 95%, 90%, 83%. Slično, prosječni DSC za LK, DK i MiO na kraju sistole je 86%, 83%, 85%. Dodatno, izračunati su volumeni LK, DK i MiO na temelju kojih su dalje računate značajne kliničke metrike te su uspoređeni rezultati s referentnim rezultatima. Navedeno uključuje kliničke metrike, odnosno pokazatelje funkcionalnosti srca, uključujući volumen lijeve klijetke na kraju dijastole (VLKKD), volumen lijeve klijetke na kraju sistole (VLKKS), frakciju izbacivanja lijeve klijetke (FILK), volumen desne klijetke na kraju dijastole (VDKKD), volumen desne klijetke na krajnjoj sistoli (VDKKS), frakciju izbacivanja desne klijetke (FIDK), volumen miokarda na krajnjoj sistoli (VMiOKS) kao i masu miokarda na kraju dijastole (MiOKD). Bland-Altman analiza pokazuje visoki koeficijent korelacije od R = 0,99 za VLKKD i VLKKD, dok je R = 0,95 za FILK. Korelacije VDKKD, VDKKS i FIDK su R = 0,97, R = 0,93, R = 0,69. Konačno, R = 0,96 za VMiOKS i R = 0,95 za MiOKD dodatno pokazuju snagu točnosti i preciznosti naše predložene metode. Konačno, četvrti znanstveni doprinos predstavlja novi automatski pristup za segmentaciju aneurizme abdominalne aorte (AAA). 3D UNet arhitektura modificirana je uvođenjem rezidualnih jedinica u koder dijelu kao i mehanizmom dubokog nadzora u dekoder dijelu. Kako bi se povećala točnost rezultata, mreža je trenirana i validirana na 19 preoperativnih AAA CTA volumena različitih pacijenata primjenom 4-ostrukog pristupa unakrsne provjere valjanosti. Naša metoda postiže DSC rezultat od 91,03% za segmentaciju aneurizme abdominalne aorte. Tijekom rada na ovoj doktorskoj disertaciji, objavljeno je 5 radova u časopisima (od čega 3 kao prvi autor), 10 radova objavljeno je na međunarodnim konferencijama (od čega 5 kao prvi autor) te 1 rad kao dio knjige (ko-autor). The term cardiovascular disease (CVD) refers to numerous dysfunctions of the heart and circulatory system. Cardiovascular disease accounts for nearly one-third (33%) of all deaths in the modern world, which is the highest proportion of all diseases. Early diagnosis and appropriate treatment can significantly reduce mortality and improve quality of life. The diagnosis of heart disease is based on the complete cardiovascular picture, including anatomy and physiology. The diagnostic process usually consists of two main parts. The first part refers to obtaining images of the heart using imaging devices. Numerous invasive and noninvasive imaging techniques have been developed to characterize the anatomy and functionality of the heart. The second part of the diagnostic process is the quantification and interpretation of the images using advanced image processing methods. Developing efficient medical image processing and analysis methods is a complex task, mainly because it involves processing large amounts of high-dimensional data. Advances in the development of image processing, computer vision, and artificial intelligence, as well as the widespread availability of powerful graphical processing units (GPUs), have made this challenging task manageable. Medical image segmentation plays an important role in the assessment, diagnosis, and prognosis of various cardiovascular diseases. Extensive research and clinical applications have shown that computed tomography (CT) and magnetic resonance imaging (MRI) play an important role in the noninvasive assessment of cardiovascular disease. They help quantify disease, measure the volume of structures, and analyze organ morphology. Therefore, segmentation of whole heart is an important step for a variety of clinical applications. For example, it is used for modeling and analyzing the anatomy and function of the heart and for localizing pathologies. The creation of a patient-specific 3D heart model holds excellent potential for improving surgical planning for patients with congenital heart defects. It requires delineation of all cardiac structures, including heart chambers, epicardial surface, entire blood pool, and great vessels. Segmentation of the left and right ventricles plays a critical role in quantitative analysis of global and regional information, i.e., indicators of cardiac function, such as enddiastolic volume (EDV), end-systolic volume (ESV), ejection fraction (EF), wall thickness, and mass. For example, ventricular hypertrophy x is caused by abnormal enlargement of the myocardium surrounding the left or right ventricle. Therefore, segmentation of the whole heart and heart chambers from volumetric medical images plays an essential role in cardiac assessment. In addition, radiologists often need to delineate the aorta to obtain its morphology, which is essential for the detection and diagnosis of aortic aneurysms. Manual segmentation of cardiac structures is a time-consuming process that depends on observer variability. Therefore, the development of accurate and robust automatic segmentation algorithms is critical for clinical practice. Deep learning has emerged as a state-of-the-art method for various image processing tasks such as recognition, segmentation, and classification. Deep learning methods are based on deep artificial neural networks. The most common type of deep neural network is convolutional neural networks (CNNs). Fully convolutional neural networks (FCNs) are a special type of CNNs that do not have a fully connected layer and are trained and applied to the entire image so that no patch selection is required. Several variants of FCNs have been proposed to transfer features from the encoder to the decoder to increase segmentation accuracy. The most widely used FCNs for biomedical image segmentation are the U-net architecture and its corresponding three-dimensional counterpart, the 3D U-net architecture. The ability of U-Net architecture to capture low-level features makes them very useful in scenarios with a small amount of training data. Although it has strong representational power, long-range relationships are weak due to the inherent localization of convolutional operations, so more advanced mechanisms and building blocks are required. Techniques and building blocks such as residual connections and deep supervision enable the construction of deeper architectures that provide more abstract learning results and higher accuracy for medical segmentation tasks. The increment in the number of layers provides larger parameter space enabling learning of more abstract features. Therefore, deeper architectures could provide more abstract learning that results in better performance and higher accuracy in medical segmentation tasks. Nevertheless, when the depth of CNN increases, information about the gradient passes through many layers, and it can vanish or accumulate large errors by the time it reaches the end of the network. This leads to common obstacles of training deep neural network architectures such as appearance of vanishing gradients, accuracy degradation, and extensive parameter growth, which results in computationally intensive models. In this Thesis, we propose a set of deep learning methods for automatic heart and heart chambers segmentation. We focus on improving deep learning segmentation methods for the whole heart, both ventricles, myocardium, and abdominal aortic aneurysm. Several unique challenges and issues arise in developing deep learning methods for medical image segmentation and analysis. For example, the high image dimensionality leads to trained models with a high number of xi parameters, and the lack of expert annotation makes the models more susceptible to overfitting. Therefore, we aim to alleviate these challenges by proposing new and robust CNNs that reduce the number of parameters so that they can be trained with smaller training sets and are less prone to overfitting. One of the most important scientific contributions of this work is the novel connectivity structure of residual units, which we call the feature merge residual unit (FM-Pre-ResNet). The FM-Pre-ResNet unit attaches two convolution layers at the top and at the bottom of the pre-activation residual block. The top layer balances the parameters of the two branches, while the bottom layer reduces the channel dimension. The proposed connectivity allows the construction of notably deeper models while maintaining the same or smaller number of parameters than the pre-activation residual units. Following that, the second scientific contribution is a novel threedimensional (3D) encoder-decoder architecture that successfully integrates FM-Pre-ResNet units and is additionally guided with variational autoencoders (VAE) for the task of whole heart segmentation from CT and MRI images. The architecture includes three stages. First, in an encoding stage, FM-Pre-ResNet units learn a low-dimensional representation of the input. Second, in the VAE stage, an input image is reduced to a low-dimensional latent space and reconstructs itself to provide a strong regularization of all model weights. This ensures that all model weights are strongly regularized while avoiding overfitting the training data. Third, the decoding stage creates the final whole heart segmentation. We evaluate our method on the 40 test subjects of the MICCAI Multi-Modality Whole Heart Segmentation (MM-WHS) Challenge. Our method achieves an average Dice score (DSC), Jaccard index (JI), surface distance (SD), and Hausdorff distance (HD) for WHS of 90.39%, 82.24%, 1.1093, and 15.3621 on CT images and 89.50%, 80.44%, 1.8599, 25.6558 on MRI images, respectively. The proposed approach obtains highly comparable DSC to the state-of-the-art for whole heart segmentation tasks on CT images while outperforming the current state-of-the-art on the MRI images. The third scientific contribution is a new automatic method for left ventricle (LV), right ventricle (RV), and myocardium (Myo) segmentation and quantification from cine-MRI images. We introduce a new architecture that incorporates SERes blocks into 3D U-net architecture (3D SERes-U-Net). The SERes blocks incorporate squeezeand- excitation operations into residual learning. The adaptive feature recalibration ability of squeeze-and-excitation operations boosts the network’s representational power while feature reuse utilizes effective feature learning, which improves segmentation performance. We evaluate the proposed method on the MICCAI Automated Cardiac Diagnosis Challenge (ACDC) testing dataset. Our method obtains an average DSC for LV, RV, and Myo at end-diastole of 95%, 90%, 83%, respectively. Similarly, we obtain an average DSC for LV, RV, and xii Myo at end-systole of 86%, 83%, 85%, respectively. Additionally, we calculate significant clinical metrics, i.e., indicators of hearts’ function, including volume of the left ventricle at end-diastole (LVEDV), the volume of the left ventricle at end-systole (LVESV), left ventricles’ ejection fraction (LVEF), the volume of the right ventricle at end-diastole (RVEDV), volume of the right ventricle at end-systole (RVESV), right ventricles’ ejection fraction (RVEF), myocardium volume at end-systole (MyoLVES), and myocardium mass at end-diastole (MyoMED). The Bland-Altman analysis shows a high correlation coefficient of R=0.99 for LVEDV and LVESV, while R=0.95 for LVEF. Correlations of RVEDV, EVESV and RVEF are R=0.97, R=0.93, R=0.69, respectively. Finally, R=0.96 for MyoLVES and R=0.95 for MyoMED further show our proposed methods’ strength of accuracy and precision. Finally, the fourth scientific contribution includes a new automatic approach for robust and reproducible abdominal aortic aneurysm (AAA) segmentation. The 3D U-Net network is adapted by introducing residual units in the contracting pathway and a deep supervision mechanism in the expanding pathway. We conduct an ablation study to demonstrate the effect of the addition of residual units and deep supervision for this particular clinical application. To increase the robustness of the results, networks are trained, validated, and evaluated on 19 pre-operative CTA volumes from different patients using a 4-fold cross-validation approach. Our pipeline achieves a Dice score of 91.03% for AAA segmentation. The work conducted during this Thesis resulted in 5 journal publications (of which 3 as the first author), 10 papers are published at international conferences (of which 5 as the first author), and 1 publication in book chapters (as co-author).

U disertaciji je razvijen i prikazan unaprijedeni dvoslojni adaptivni sustav upravljanja energijom za kratkoroˇcno planiranje pogona mikromreˇze. Dvoslojni sustav upravljanja energijom sastoji se od dva medusobno povezana optimizacijska problema (sloja). Viˇsi sloj je realiziran kao mjeˇsovito cjelobrojni linearni optimizacijski problem implementiran u okviru strategije pomiˇcnog horizonta s korektivnim djelovanjem i ciljem maksimizacije dobiti mikromreˇze. Niˇzi sloj je realiziran kao nelinearni optimizacijski problem u kosimulacijskom okruˇzenju koji uvaˇzava elektriˇcka ograniˇcenja elektroenergetske mreˇze, a ima tehniˇcki cilj minimizacije odstupanja od planiranih vrijednosti viˇsim slojem (minimizacija korektivnih mjera) u gotovo stvarnom vremenu. Sustav upravljanja energijom ima implementiran unaprijedeni model baterijskog sustava pohrane energije koji se sastoji od kombiniranja uˇcinkovitosti dvosmjernog energetskog pretvaraˇca ovisne o optere´cenju, energetskog modela punjenja baterije, modeliranja cikliˇckog i kalendarskog starenja baterije i valorizacije pohranjene energije u bateriji. Ulazni podaci u optimizacijske probleme predvidaju se uz pomo´c razliˇcitih metoda, ovisno o sloju sustava upravljanja energijom i veliˇcini koja se predvida. Koriˇstene metode podrazumijevaju modele numeriˇckog predvidanja vremena i umjetne neuronske mreˇze s dugom kratkoroˇcnom memorijom. Unaprijedeni dvoslojni adaptivni sustav upravljanja energijom primijenjen je na planiranje kratkoroˇcnog pogona testne mikromreˇze na osnovu kojeg su doneseni generalni zakljuˇcci da toˇcnost prognoza koje sluˇze kao ulazni podaci uvelike utjeˇce na optimalnost pogona mikromre ˇze. Instalirana snaga vlastitih izvora elektriˇcne energije u mikromreˇzi uvelike pridonosi stupnju samodostatnosti opskrbe mikromreˇze i smanjenju pogonskih troˇskova. Razina sloˇzenosti modela baterijskog sustava pohrane energije omogu´cuje vodenje razliˇcitih strategija pogona mikromreˇze kao kompromis izmedu smanjenja pogonskih troˇskova i produljenja ˇzivotnog vijeka opreme. An advanced double-layer adaptive energy management system for short-term microgrid operation planning is developed and presented in the thesis. The double-layer energy management system consists of two interconnected optimization problems (layers). The higher layer is realized as a mixed integer linear optimization problem implemented within the rolling horizon strategy with reactive scheduling and the objective function of maximizing the profit of the microgrid. The lower layer is realized as a nonlinear optimization problem in a cosimulation environment that integrates the electrical limitations of the power grid, and has the technical objective of minimizing deviations from the planned values of the higher layer (minimisation of corrective measures) in near real time. The energy management system has implemented an advanced model of battery energy storage system consisting of combining the load-dependent efficiency of a bidirectional power converter unit, battery energy charging model, modeling cycle and calendar aging of the battery and valorization of stored energy in the battery. The input data to the optimization problems are predicted using different methods, depending on the layer of the energy management system and the size that is predicted. The methods used include numerical weather prediction models and artificial neural networks with long short-term memory. An advanced double-layer adaptive energy management system was applied to the shortterm operation planning of the test microgrid, based on which general conclusions were made that the accuracy of forecasts that serve as input data greatly affects the optimality of microgrid operation. The installed power of microgrid sources of electricity greatly contributes to the degree of self-sufficiency of the microgrid and the reduction of operating costs. The level of complexity (reality) of the battery energy storage system model enables the microgrid operator to manage different operation strategies as a compromise between reducing operating costs and extending equipment life.

Suvremeni trendovi u dizajniranju električnih strojeva usmjereni su na povećanje efikasnosti rada, odnosno na usavršavanje tehnologije proizvodnje električnih strojeva. To posljedično dovodi do smanjenja volumena i smanjenja toplinskih vremenski konstanti električnog stroja. Kako se električnim strojevima u suvremenim industrijskim pogonima nameće i dinamičniji režimi rada to dovodi do pojave bržeg zagrijavanja motora pri preopterećenju što je posebno izraženo kod teških elektromotornih pogona (EMP). Visokonaponski (VN) asinkroni kavezni motori (AKM) dimenzioniraju se za točno određeni teret velike snage te sa vrlo malim dozvoljenim preopterećenjem za primjenu u teškim EMP. Postoji veliki broj različitih faktora koji utiču na rad VN AKM, a odstupanje svakog od njih može uzrokovati njegovo oštećenje. Razvijeni su razni oblici zaštita motora u suvremenim industrijskim pogonima sa ciljem sprečavanja nastanka kvara ili smanjenja materijalne štete procesa proizvodnje. Štete zaustavljanja procesa proizvodnje pogonjeni sa VN AKM u pravilu su višestruko veće od štete uništenja motora uslijed kvara, pa se nameću zahtjevi procjene stanja i produženja rada motora u stanju kvara dok se ne osigura zamjenski pogoni. Najčešće korištene zaštite u praksi su korištenjem numeričkih releja s obzirom na jednostavniju funkcionalnost, nižu cijenu i robusniju izradu. U ovoj disertaciji prezentirano je korištenje numeričkih releja kao već postojećeg kontrolnog uređaja zaštite VN AKM kojemu možemo aktivirati i dodatnu prediktivnu zaštitnu ulogu u cilju osiguravanja uvjeta za procjenu stanja teških elektromotornih pogona sa VN AKM. Time se postiže povećanje pouzdanosti rada postrojenja uz zanemarivo ulaganje u postojeću infrastrukturu upravljanja i zaštite. Prezentirane su teorijske postavke dinamičkih režima rada teških EMP sa VN AKM, utjecaji napojne mreže, kao i ekspertna znanja dobivena istraživanjima kvarova te prijelaznih režima rada. Dat je prikaz postojećih dijagnostičkih metoda za lociranje kvarova sa posebnim osvrtom na on-line metode, te primjene zaštitnih numeričkih releja u procjeni stanja teških EMP pogonjenih sa VN AKM. Modern trends in the design of electrical machines aim to increase efficiency, i.e. to improve the technology of electrical machine production. This in turn leads to a reduction in the volume and a reduction in the thermal time constants of the electrical machine. As the electrical machines in modern industrial plants are operated more and more dynamically, this leads to faster overheating of the motor in the event of an overload, which is particularly pronounced in heavy electric motor drives (EMD). High voltage (HV) Squirrel Cage Induction Motors (SCIM) are designed for particularly high loads and have very low permissible overloads for use in heavy duty applications. There are a number of different factors that affect the operation of HV SCIM and a deviation from any of these factors can result in damage to the motor. In modern industrial plants, various forms of motor protection have been developed to prevent the occurrence of failures or to reduce the damage to the production process. The damage caused by stopping the production process driven by HV SCIM is usually many times more expensive than the damage caused by destroying the motor due to a failure. Therefore, it's necessary to assess the condition of the failed motor and extend its operation until replacement drives are available. The most common protective measures used in practise are the use of numerical relays due to their simpler operation, lower cost and more robust construction. The dissertation presents the use of numerical relays as an existing control device for HV SCIM protection, to which we can activate an additional predictive protection function to create conditions for state evaluation of heavy electric motor drives with HV SCIM. This increases the reliability of the system with negligible investment in the existing management and protection infrastructure. Theoretical settings of dynamic modes of heavy EMD driven by HV SCIM, influences of the power supply network and expert knowledge from research on failures and transient operating modes are presented. Current diagnostic methods for fault location are presented, in particular on-line methods and the application of numerical protection relays in the condition assessment of heavy EMD driven by HV SCIM..

Doktorski rad pod naslovom Stohastičko višekriterijsko optimiranje industrijske mikromreže rezultat je dugogodišnjeg znanstvenog istraživanja na polju industrijskih kogeneracijskih postrojenja. Znanstveni doprinosi rada su: 1) Definicija funkcije istovremenog trajanja potrošnje električne energije i toplinske energije, 2) Generički model kogeneracijskog postrojenja korištenjem funkcije istovremenog trajanja potrošnje električne energije i toplinske energije i 3) Model za stohastičko višekriterijsko optimiranje industrijske mikromreže korištenjem funkcije istovremenog trajanja potrošnje električne energije i toplinske energije. Rad ima osam poglavlja. Prvo poglavlje daje osvrt na uzroke i ciljeve reforme EES-a te uvodi koncept naprednih mreža kao odgovor na izazove budućeg razvoja EES-a. U drugom poglavlju rada ukratko je opisana problematika planiranja pogona. Treće poglavlje detaljnije obrađuje temu industrijskih mikromreža. U četvrtom poglavlju je predstavljen novi pristup planiranja industrijske mikromreže uvođenjem pojma funkcije istovremenog trajanja potrošnje električne i toplinske energije kao i generičkog model kogeneracijskog postrojenja. Peto poglavlje opisuje kreiranje višekriterijskog modela optimizacije industrijske mikromreže. Zadatak optimizacije je vrednovati moguće opcije sudjelovanja industrijske mikromreže na tržištu električne energije. Predložene su četiri opcije. U šestom poglavlju detaljno je opisan dijagram toka višekriterijskog stohastičkog optimizacijskog postupka. Napravljen je računalni program za demonstraciju, a rezultati su prikazani na kraju poglavlja. U sedmom dodatnim opisom predmeta istraživanja. The doctoral thesis entitled Stochastic multicriterial optimization of an industrial microgrid is the result of long-term scientific research in the field of industrial cogeneration. The scientific contributions of the thesis are: 1) Definition of the function of the simultaneous duration of electricity and thermal energy consumption, 2) Generic model of a cogeneration plant using the function of the simultaneous duration of electricity and thermal energy consumption and 3) Model for stochastic multicriterial optimization of the industrial microgrid using the function of the simultaneous duration of electricity consumption energy and thermal energy loads. The work has eight chapters. The first chapter overviews the causes and goals of the electric power system (EPS) reform and introduces the concept of advanced networks. In the second chapter the microgrid operation planning issues are briefly described. The third chapter deals with the topic of industrial microgrids in more detail. In the fourth chapter, a new approach to industrial microgrid planning is presented by introducing the concept of the function of the simultaneous duration of electricity and thermal energy loads, as well as a generic model of a cogeneration plant. The fifth chapter describes the creation of a multicriterial industrial microgrid optimization model. The task of optimization is to evaluate options for industrial microgrid participation in the electricity market. Four options are proposed. In the sixth chapter, the flowchart of the optimization procedure is described in detail. A computer program for demonstration was created, the results are presented at the end of the chapter. In the seventh chapter, concluding considerations are presented. The eighth chapter is an appendix where additional description of the research subject is presented.

Izraz kardiovaskularne bolesti (KVB) odnosi se na brojne funkcionalne abnormalnosti srca i krvožilnog sustava. KVB uzrokuju gotovo jednu trećinu (33%) smrtnosti u suvremenom svijetu, što predstavlja najveći udio u odnosu na sve druge bolesti. Rana dijagnoza i odgovarajuće liječenje kardiovaskularnih bolesti mogu značajno smanjiti smrtnost I poboljšati kvalitetu pacijentova života. Postavljanje dijagnoze temelji se na cjelokupnoj slici kardiovaskularnog sustava, uključujući anatomiju i fiziologiju srca. Dijagnostički proces obično se sastoji od dva glavna dijela. Prvi dio odnosi se na prikupljanje slika srca pomoću medicinskih uređaja. Razvijene su brojne invazivne i neinvazivne tehnike medicinskog snimanja koje omogućuju uvid u anatomiju i funkcionalnost srca. Drugi dio dijagnostičkog procesa je kvantifikacija i interpretacija prethodno dobivenih slika pomoću naprednih metoda obrade slike. Razvoj učinkovitih metoda za obradu medicinskih slika je složen zadatak, s obzirom da podrazumijeva obradu ogromne količine visokodimenzionalnih podataka. Napredak u razvoju algoritama obrade slike, računalnog vida i umjetne inteligencije, kao i dostupnost grafičkih procesorskih jedinica (GPU-a), značajno su olakšale i ubrzale razvoj takvih metoda. Segmentacija medicinskih slika ima važnu ulogu u procjeni, dijagnozi te postavljanju prognoze različitih kardiovaskularnih bolesti. Opsežna istraživanja i kliničke primjene pokazale su da računalna tomografija (CT) i magnetska rezonanca (MRI), kao osnovne tehnike prikupljanja medicinskih slika, imaju izrazito važnu ulogu u procjeni kardiovaskularnih bolesti. Njima je omogućeno kvantificiranje bolesti, mjerenje volumena kao i analizira morfologije različitih organa. Prema tome, segmentaciju srca i srčanih struktura predstavlja osnovu za širok spektar kliničkih primjena. Primjerice, često se koristi se za modeliranje i analizu anatomije i funkcionalnosti kao i za lokalizaciju različitih patologija. Izrada trodimenzionalnog (3D) modela srca specifičnog za pojedinog pacijenta predstavlja izrazit potencijal za poboljšanje kirurškog planiranja za pacijente s urođenom srčanom manom. Kako bi se takvi 3D modeli mogli izraditi, potrebno je imati segmentirane različite srčane strukture, uključujući pojedine srčane komore, epikardijalnu površinu, aortu kao i pojedine žile kardiovaskularnog sustava. Segmentacija lijeve i desne klijetke ima izrazito važnu ulogu u kvantitativnoj analizi globalnih i regionalnih informacija, odnosno pokazatelja rada srca, poput volumena na kraju dijastole (VKD), volumena na kraju sistole (VKS), frakcije izbacivanja (FI), debljine stijenke ili mase. Primjerice, ventrikularna hipertrofija uzrokovana je abnormalnim povećanjem srčanog mišića koji okružuje lijevu ili desnu klijetku. Prema tome, segmentacija cijelog srca i srčanih komora iz volumetrijskih medicinskih slika igraju bitnu ulogu u procjeni cjelokupnog kardiovaskularnog zdravlja. Nadalje, radiolozi često trebaju ocrtati aortu kako bi dobili njezinu morfologiju, što je bitno za otkrivanje i dijagnosticiranje aneurizme aorte. Ručna segmentacija srca i srčanih struktura je vremenski veoma zahtijevan posao, podložan subjektivnosti. Prema tome, razvoj točnih i robusnih automatskih algoritama za segmentaciju je neophodan za primjenu u kliničkoj praksi. Duboko učenje predstavlja najsuvremeniju metodu za različite zadatke obrade slike poput raspoznavanja, segmentacije i klasifikacije. Metode dubokog učenja temelje se na umjetnim neuronskim mrežama. Najčešće upotrebljena vrsta neuronske mreže su konvolucijske neuronske mreže (CNN). FCNs predstavljaju specifičnu vrstu CNN-a bez potpuno povezanog sloja, kojima se obrađuje cijela slika te nije potrebno korištenje patcheva. Razvijene su različite varijante FCN-a, od kojih su najznačajnije varijante koje koriste koder-dekoder arhitekture. U biomedicinskoj obradi slika, za segmentaciju, najčešće se koristi U-Net arhitektura neruonske mreže kao i njezina odgovarajuća 3D verzija. U-Net arhitektura ima snažnu reprezentativnu snagu te je u mogućnosti zabilježiti značajke niskih razina što je izrazito važno prilikom treniranja mreže sa malom količinom podataka. Iako U-Net ima snažnu reprezentativnu snagu, dugoročni odnosi između značajki su slabi zbog upotrebe konvolucijskih operacija. Prema tome, potrebno je razvijati naprednije mehanizme kao i dodatne blokove koji će biti u mogućnosti ispraviti nedostatke U-Net arhitekture. Tehnike i blokovi poput veza za preskakivanje ili dubokog nadzora, omogućuju izgradnju dubljih arhitektura neuronskih mreža koje pružaju apstraktnije rezultate učenja te postižu veću točnost prilikom segmentacije medicinskih slika. S obzirom da povećanje broja slojeva osigurava veći prostor parametara koji omogućuje učenje apstraktnijih značajki, dublje arhitekture neuronskih mreža pružaju apstraktnije učenje koje rezultira boljim performanse i većom točnost u zadacima medicinske segmentacije. Unatoč tome, kako se dubina mreže povećava, informacije o gradijentu prolaze kroz mnogo slojeva te mogu nestati ili nakupiti velike pogreške do trenutka kada gradijet dosegne kraj mreže. To dovodi do uobičajenih prepreka treninga dubokih arhitektura neuronskih mreža kao što su problem nestajajućih gradijenta, ekstenzivnog rasta parametara, kao i smanjenja točnosti, što dovodi do računalno zahtjevnih modela. U ovoj doktorskoj disertaciji, predložen je niz metoda dubokog učenja za automatsku segmentaciju srca i srčanih komora. Fokus disertacije je na poboljšanju metoda dubokog učenja za segmentaciju cijeloga srca, lijeve i desne klijetke i miokarda kao i aneurizme abdominalne aorte. S obzirom na karakteristične probleme koji se javljanju prilikom dizajniranja metoda dubokog učenja za segmentaciju medicinskih slika, poput problema visoke dimenzionalnosti slika koje rezultiraju treniranim modelima s velikim brojem parametara kao i nedostatkom anotiranih podataka za treniranje, cilj ove disertacije je ublažiti navedene izazove predlaganjem novih i robusnih arhitektura neuronskih mreža koje smanjuju broj korištenih parametara, ali zadržavaju izrazito visoku točnost krajnjih rezultata segmentacije. Prvi i najvažniji znanstveni doprinos predstavlja nova struktura povezivanja rezidualnih jedinica, koju nazivamo rezidualna jedinica za spajanje značajki (FM-Pre-ResNet). FM-Pre-ResNet struktura povezivanja rezidualnih jedinica dodaje konvolucijski sloj na vrh i na dno već postojećih prethodno aktivirajućih rezidualnih jedinica. Pri tome, gornji sloj uravnotežuje parametre dviju grana rezidualne jedinice, dok donji sloj smanjuje dimenzije kanala. Na ovaj način predložena struktura povezivanja rezidualnih jedinica omogućuje kreiranje značajno dubljih modela uz održavanje iste ili čak manje količine parametara u odnosu na originale rezidualne jedinice. Nakon toga, u drugom znanstvenom doprinosu, predložena je nova 3D arhitektura neuronske mreže bazirana na koder-dekoder arhitekturi koja uspješno integrira FM-Pre-ResNet jedinice s varijacijskim autokoderima (VAE) za segmentaciju srca i srčanih komora iz CT i MRI slika. Metoda se sastoji od tri osnovna dijela. U prvom dijelu, prethodno predložene FM-Pre-ResNet jedinice koriste se za učenje nisko-dimenzionalnog prikaza ulaza u fazi kodiranja. U drugom dijelu, VAE rekonstruira ulaznu sliku iz nisko-dimenzionalnog latentnog prostora, osiguravajući da su sve težine modela snažno regulirane, kako bi se izbjegnula neželjena pojava pretreniranja. VAE dio koristi se samo tijekom treniranja mreže. Konačno, u trećoj fazi dekodiranja ponovno su integrirane FM-Pre-ResNet jedinice pomoću kojih se stvaraju konačne segmentacije srca. Predložena nova arhitektura evaluirana je na testnom skupu podataka koji se sastoji od 40 različitih pacijenata dostupnih kroz MICCAI Multi-Modality Whole Segmentation Challenge (MM-WHS) izazov. Naša metoda ostvarila je prosječni DSC, JI, SD i HD za cijelo srce od 90,39%, 82,24%, 1.1093 i 15,3621 na CT snimkama, odnosno 89,50%, 80,44%, 1,8599, 25,6558 na MRI snimkama. Predloženi pristup ostvario je približno slične rezultate kao i najsuvremenije metode za segmentaciju cijelog srca na CT slikama dok su rezultati na MRI slikama bolji od rezultata prethodno objavljenih najsuvremenijih metoda. Treći znanstveni doprinos, predstavlja novu automatsku metodu za segmentaciju miokarda (MiO), lijeve (LK) i desne klijetke (DK) iz cineMRI slika. Predstavljena je nova arhitekturu koja integrira SERes blokove u 3D U-net arhitekturu (3D SERes-U-Net). SERes blokovi upotrebljavaju operacije stiskanja i uzbude u rezidualne jedinice. Sposobnost ponovne kalibracije značajki operacija stiskanja i uzbude povećava reprezentativnu snagu mreže, dok ponovna upotreba značajki koristi učinkovito učenje o značajkama, što poboljšava performanse segmentacije. Predloženu metodu evaluirali smo na testnom skupu podataka MICCAI Automated Cardiac Diagnosis Challenge (ACDC). Naša predložena metoda za segmentaciju pomoću 3D SERes-U-Net ostvarila je prosječni DSC za LK, DK i MiO na kraju dijastole od 95%, 90%, 83%. Slično, prosječni DSC za LK, DK i MiO na kraju sistole je 86%, 83%, 85%. Dodatno, izračunati su volumeni LK, DK i MiO na temelju kojih su dalje računate značajne kliničke metrike te su uspoređeni rezultati s referentnim rezultatima. Navedeno uključuje kliničke metrike, odnosno pokazatelje funkcionalnosti srca, uključujući volumen lijeve klijetke na kraju dijastole (VLKKD), volumen lijeve klijetke na kraju sistole (VLKKS), frakciju izbacivanja lijeve klijetke (FILK), volumen desne klijetke na kraju dijastole (VDKKD), volumen desne klijetke na krajnjoj sistoli (VDKKS), frakciju izbacivanja desne klijetke (FIDK), volumen miokarda na krajnjoj sistoli (VMiOKS) kao i masu miokarda na kraju dijastole (MiOKD). Bland-Altman analiza pokazuje visoki koeficijent korelacije od R = 0,99 za VLKKD i VLKKD, dok je R = 0,95 za FILK. Korelacije VDKKD, VDKKS i FIDK su R = 0,97, R = 0,93, R = 0,69. Konačno, R = 0,96 za VMiOKS i R = 0,95 za MiOKD dodatno pokazuju snagu točnosti i preciznosti naše predložene metode. Konačno, četvrti znanstveni doprinos predstavlja novi automatski pristup za segmentaciju aneurizme abdominalne aorte (AAA). 3D U-Net arhitektura modificirana je uvođenjem rezidualnih jedinica u koder dijelu kao i mehanizmom dubokog nadzora u dekoder dijelu. Kako bi se povećala točnost rezultata, mreža je trenirana i validirana na 19 preoperativnih AAA CTA volumena različitih pacijenata primjenom 4-ostrukog pristupa unakrsne provjere valjanosti. Naša metoda postiže DSC rezultat od 91,03% za segmentaciju aneurizme abdominalne aorte. Tijekom rada na ovoj doktorskoj disertaciji, objavljeno je 5 radova u časopisima (od čega 3 kao prvi autor), 10 radova objavljeno je na međunarodnim konferencijama (od čega 5 kao prvi autor) te 1 rad kao dio knjige (ko-autor). The term cardiovascular disease (CVD) refers to numerous dysfunctions of the heart and circulatory system. Cardiovascular disease accounts for nearly one-third (33%) of all deaths in the modern world, which is the highest proportion of all diseases. Early diagnosis and appropriate treatment can significantly reduce mortality and improve quality of life. The diagnosis of heart disease is based on the complete cardiovascular picture, including anatomy and physiology. The diagnostic process usually consists of two main parts. The first part refers to obtaining images of the heart using imaging devices. Numerous invasive and non-invasive imaging techniques have been developed to characterize the anatomy and functionality of the heart. The second part of the diagnostic process is the quantification and interpretation of the images using advanced image processing methods. Developing efficient medical image processing and analysis methods is a complex task, mainly because it involves processing large amounts of high-dimensional data. Advances in the development of image processing, computer vision, and artificial intelligence, as well as the widespread availability of powerful graphical processing units (GPUs), have made this challenging task manageable. Medical image segmentation plays an important role in the as-assessment, diagnosis, and prognosis of various cardiovascular diseases. Extensive research and clinical applications have shown that computed tomography (CT) and magnetic resonance imaging (MRI) play an important role in the non-invasive assessment of cardiovascular disease. They help quantify disease, measure the volume of structures, and analyze organ morphology. Therefore, segmentation of whole heart is an important step for a variety of clinical applications. For example, itis used for modelling and analysing the anatomy and function of the heart and for localizing pathologies. The creation of a patient-specific3D heart model holds excellent potential for improving surgical planning for patients with congenital heart defects. It requires delineation of all cardiac structures, including heart chambers, epicardial surface, entire blood pool, and great vessels. Segmentation of the left and right ventricles plays a critical role in quantitative analysis of global and regional information, i.e., indicators of cardiac function, such as end-diastolic volume (EDV), end-systolic volume (ESV), ejection fraction(EF), wall thickness, and mass. For example, ventricular hypertrophy is caused by abnormal enlargement of the myocardium surrounding the left or right ventricle. Therefore, segmentation of the whole heart and heart chambers from volumetric medical images plays an essential role in cardiac assessment. In addition, radiologists often need to delineate the aorta to obtain its morphology, which is essential for the detection and diagnosis of aortic aneurysms. Manual segmentation of cardiac structures is a time-consuming process that depends on observer variability. Therefore, the development of accurate and robust automatic segmentation algorithms is critical for clinical practice. Deep learning has emerged as a state-of-the-art method for various image processing tasks such as recognition, segmentation, and classification. Deep learning methods are based on deep artificial neural networks. The most common type of deep neural network is convolutional neural networks (CNNs). Fully convolutional neural networks (FCNs) are a special type of CNNs that do not have a fully connected layer and are trained and applied to the entire image so that no patch selection is required. Several variants of FCNs have been pro-posed to transfer features from the encoder to the decoder to increase segmentation accuracy. The most widely used FCNs for biomedical image segmentation are the U-net architecture and its corresponding three-dimensional counterpart, the 3D U-net architecture. The ability of U-Net architecture to capture low-level features makes them very useful in scenarios with a small amount of training data. Although it has strong representational power, long-range relationships are weak due to the inherent localization of convolutional operations, so more advanced mechanisms and building blocks are required. Techniques and building blocks such as residual connections and deep supervision enable the construction of deeper architectures that provide more abstract learning results and higher accuracy for medical segmentation tasks. The increment in the number of layers provides larger parameter space enabling learning of more abstract features. Therefore, deeper architectures could provide more abstract learning that results in better performance and higher accuracy in medical segmentation tasks. Nevertheless, when the depth of CNN increases, information about the gradient passes through many layers, and it can vanish or accumulate large errors by the time it reaches the end of the network. This leads to common obstacles of training deep neural network architectures such as appearance of vanishing gradients, accuracy degradation, and extensive parameter growth, which results in computationally intensive models. In this Thesis, we propose a set of deep learning methods for automatic heart and heart chambers segmentation. We focus on improving deep learning segmentation methods for the whole heart, both ventricles, myocardium, and abdominal aortic aneurysm. Several unique challenges and issues arise in developing deep learning methods for medical image segmentation and analysis. For example, the high im-age dimensionality leads to trained models with a high number of parameters, and the lack of expert annotation makes the models more susceptible to overfitting. Therefore, we aim to alleviate these challenges by proposing new and robust CNNs that reduce the number of parameters so that they can be trained with smaller training sets and are less prone to overfitting. One of the most important scientific contributions of this work is the novel connectivity structure of residual units, which we call the feature merge residual unit (FM-Pre-ResNet). The FM-Pre-ResNet unit attaches two convolution layers at the top and at the bottom of the pre-activation residual block. The top layer balances the parameters of the two branches, while the bottom layer reduces the channel dimension. The proposed connectivity allows the construction of notably deeper models while maintaining the same or smaller number of parameters than the pre-activation residual units. Following that, the second scientific contribution is a novel three-dimensional (3D) encoder-decoder architecture that successfully integrates FM-Pre-ResNet units and is additionally guided with variational autoencoders (VAE) for the task of whole heart segmentation from CT and MRI images. The architecture includes three stages. First, in an encoding stage, FM-Pre-ResNet units learn a low-dimensional representation of the input. Second, in the VAE stage, an input image is reduced to a low-dimensional latent space and reconstructs itself to provide a strong regularization of all model weights. This ensures that all model weights are strongly regularized while avoiding overfitting the training data. Third, the decoding stage creates the final whole heart segmentation. We evaluate our method on the 40 test subjects of the MICCAI Multi-Modality Whole Heart Segmentation (MM-WHS)Challenge. Our method achieves an average Dice score (DSC), Jaccard index (JI), surface distance (SD), and Hausdorff distance (HD) for WHS of 90.39%, 82.24%, 1.1093, and 15.3621 on CT images and 89.50%,80.44%, 1.8599, 25.6558 on MRI images, respectively. The proposed approach obtains highly comparable DSC to the state-of-the-art for whole heart segmentation tasks on CT images while outperforming the current state-of-the-art on the MRI images. The third scientific contribution is a new automatic method for left ventricle (LV), right ventricle (RV), and myocardium (Myo) seg-mentation and quantification from cine-MRI images. We introduce a new architecture that incorporates SERes blocks into 3D U-net architecture (3D SERes-U-Net). The SERes blocks incorporate squeeze-and-excitation operations into residual learning. The adaptive feature recalibration ability of squeeze-and-excitation operations boosts the network’s representational power while feature reuse utilizes effective feature learning, which improves segmentation performance. We evaluate the proposed method on the MICCAI Automated Cardiac Diagnosis Challenge (ACDC) testing dataset. Our method obtains an average DSC for LV, RV, and Myo at end-diastole of 95%, 90%, 83%,respectively. Similarly, we obtain an average DSC for LV, RV, and Myo at end-systole of 86%, 83%, 85%, respectively. Additionally, we calculate significant clinical metrics, i.e., indicators of hearts’ function, including volume of the left ventricle at end-diastole (LVEDV), the volume of the left ventricle at end-systole (LVESV), left ventricles’ ejection fraction (LVEF), the volume of the right ventricle at end-diastole(RVEDV), volume of the right ventricle at end-systole (RVESV), right ventricles’ ejection fraction (RVEF), myocardium volume at end-systole(MyoLVES), and myocardium mass at end-diastole (MyoMED). The Bland-Altman analysis shows a high correlation coefficient of R=0.99for LVEDV and LVESV, while R=0.95 for LVEF. Correlations ofRVEDV, EVESV and RVEF are R=0.97, R=0.93, R=0.69, respectively. Finally, R=0.96 for MyoLVES and R=0.95 for MyoMED further show our proposed methods’ strength of accuracy and precision. Finally, the fourth scientific contribution includes a new automatic approach for robust and reproducible abdominal aortic aneurysm(AAA) segmentation. The 3D U-Net network is adapted by introducing residual units in the contracting pathway and a deep supervision mechanism in the expanding pathway. We conduct an ablation study to demonstrate the effect of the addition of residual units and deep supervision for this particular clinical application. To increase the robustness of the results, networks are trained, validated, and evaluated on 19 pre-operative CTA volumes from different patients using a 4-foldcross-validation approach. Our pipeline achieves a Dice score of 91.03%for AAA segmentation. The work conducted during this Thesis resulted in 5 journal publications (of which 3 as the first author), 10 papers are published at international conferences (of which 5 as the first author), and 1publication in book chapters (as co-author).

Doktorski rad se bavi problemom umjeravanja prijenosnih mjernih etalona za mjerenje snage i efektivne vrijednosti izmjeničnoga signala korištenjem mikrokalorimetra. Mikrokalorimetar predstavlja instrument kojim se mjeri efikasnost umjeravanog etalona putem određivanja gubitaka snage uslijed toplinskih efekata. Prikazana izvedba mikrokalorimetra sastoji se od sljedećih elemenata: dvostrukog termostata; izolacijskog sloja u koaksijalnom vodu (optimizacija toplinskog protoka); koaksijalnog voda za prijenos signala; termoelemenata za mjerenje gubitaka snage u bolometarskom kućištu; istosmjernih komponenti potrebnih za održavanje bolometarskog elementa u ravnoteži. Određivanje korekcije gubitaka mikrokalorimetra provodi se mjerenjem gubitaka uslijed toplinskih efekata (gubici u konektorima i u kućištu bolometra). Cilj ovog rada je razvoj mikrokalorimetra za precizno mjerenje snage u frekvencijskom području od 500 MHz – 6 GHz namijenjenog za primjenu u pametnim mrežama. U prvom poglavlju (1. Uvod) je sažeto prikazana motivacija i ciljevi doktorskog rada, te je razrađen sadržaj rada. Također, navedeni su pojedini dijelovi rada koji su prethodili konačnoj realizaciji mikrokalorimetarskog mjerenja snage izmjeničnog signala temeljenoga na samouravnotežujućem mostu. U drugom poglavlju (2. Mikrokalorimetri i prijenosni etaloni) sažeto je prikazan razvoj mikrokalorimetara tijekom desetljeća s posebnim naglaskom na komponente od kojih se mikrokalorimetar sastoji. Također, dan je osvrt na najnovija istraživanja u području te put do pojave mikrokalorimetara s prekidom u vanjskom vodiču adijabatske linije. U trećem poglavlju (3. Komunikacija pametnih mreža i prikupljanje energije iz okoline) dan je uvod u komunikacije pametnih mreža gdje postoji primjena mikrokalorimetra u umjeravanju, s posebnim osvrtom na postojeće bežične komunikacijske mreže te frekvencijske raspone. U četvrtom poglavlju (4. Definicija električne snage) dana je definicija električne snage prema normi IEEE Std 1459TM - 2010. Navedena norma opisuje komponente snage te teoretske osnove mjerenja snage. vi U petom poglavlju (5. Precizno mjerenje snage i efektivne vrijednosti izmjeničnog signala) sažeto je prikazan mjerni postupak preciznog mjerenja snage korištenjem termalnih detektora snage s posebnim naglaskom na proceduru umjeravanja. Ukratko su opisani i termalni pretvornici te njihove mogućnosti za mjerenje snage na visokim frekvencijama. U šestom poglavlju (6. Mikrokalorimetarsko mjerenje snage na visokim frekvencijama) dan je opis procedure mjerenja snage na visokim frekvencijama korištenjem mikrokalorimetra. Ovo poglavlje predstavlja najvažnije poglavlje doktorskog rada jer su u njemu prikazani: simulacijski rezultati adijabatske linije, prototip razvijenog uređaja te električna i toplinska karakterizacija adijabatske linije. Također je opisan termalni detektor korišten za izradu prototipa kao i dvostruko metalno kućište termostata. Pored navedenog, dan je opis termostata, toplinska karakterizacija i temperaturna regulacija termostata te termopila. U sedmom poglavlju (7. Uravnoteženje mosnog spoja) opisan je postupak uravnoteženja mosnog spoja uz akviziciju mjernih podataka digitalnim multimetrima razlučivosti 8 ½ digita uz sinkronizaciju i upravljanje kroz LabVIEW sustav. Osmo poglavlje je zaključak. PhD thesis addresses the calibration of power and effective AC measurement standards for the measurement of power and RMS value of AC signals. Microcalorimeter represents an instrument for effective efficiency measurements of test standard by measuring losses due to heat transfer. Design of the microcalorimeter is based on modelling: double shielded housing; isolation section (heat transfer optimization); coaxial transmission line; elements for detection of losses in bolometer housing; DC components for the realization of bridge circuit in order to hold bolometer constant resistance; A correction factor of microcalorimeter is determined by measuring heat losses and heat effects (losses in connectors, in bolometer mount). The main purpose of PhD dissertation was to develop a microcalorimeter structure for accurate power measurement in the frequency range of 500 MHz - 6 GHz intended for use in smart grids. The first chapter (1. Introduction) summarizes the motivation and goals of the doctoral thesis and elaborates the content of the of dissertation. Also, individual parts of work that preceded the final realization of microcalorimetric measurements of AC signal power based on a selfbalancing bridge are. The second chapter (2. Microcalorimeters and transfer standards) summarizes the development of microcalorimeters over the decades with special emphasis on the components of which the microcalorimeter consists of. Also, a review of the latest research in the field and a chronological path to the appearance of microcalorimeters with an interruption in the outer conductor of the adiabatic line is given. The third chapter (3. Smart Grid communication and energy harvesting) introduces Smart Grid communication and energy harvesting systems, where exists the application of microcalorimeters, with special reference to existing wireless communication networks and frequency bands. v In the fourth chapter (4. Definition of electrical power) the definition of electrical power is given according to the standard IEEE Std-2010. This standard describes the components of power and theoretical basis of power measurement. The fifth chapter (5. Precise measurement of power and RMS value of the AC signal) summarizes the measurement procedure for precise power measurement using thermal power detectors with special emphasis on the calibration procedure. Thermal converters and their possibilities for measuring power at high frequencies are also briefly described. Chapter 6 (6. Microcalorimetric Power Measurement at High Frequencies) describes the procedure for measuring the power at high frequencies using a microcalorimeter. This chapter is the most important chapter of the doctoral thesis because it presents: simulation results of the adiabatic line, a prototype of the developed device and electrical and thermal characterization of the adiabatic line. Also, thermal detector is described which is used in prototype as well as a double metal thermostat housing. In addition to the above, a description of the thermal characterization and temperature regulation of the thermostat. In the seventh Chapter (7. Balancing of the bridge circuit) describes the process of balancing the bridge connection and the acquisition of measurement data by digital multimeters with a resolution of 8 ½ digital with synchronization and control through the LabVIEW system. The eight chapter are conclusions.

The transition of the coal-based energy system is a challenge that can be solved by considering the integration of variable renewable energy sources and synergic effects between electricity, heating, cooling, transport and industrial sectors. This thesis focuses mostly on the integration of variable renewables using power-to-heat technologies in district heating systems. As 70-80% of total energy consumption in buildings takes the form of heat consumed for space heating purposes, spatial identification is crucial for planning and designing sustainable district heating systems. Hence, this thesis studies the quantification and validation of heat demand distribution within a small municipality using a newly developed Bottom-up and Top-down heat mapping method. The Bottom-up mapping method is based on building features such as surface floor area, building height, building use, and the heated area share. In contrast, the top-down mapping method relies on energy balances and population distribution densities. The results are shown spatially in grids with high spatial resolution. The finding shows that current models overestimate the heat demand, while with the proposed method, the error between existing and proposed models is negligible. The bottom-up heat demand maps are further improved by considering domestic hot water besides space heating demand. Domestic hot water should be taken into account when assessing district heating potential, as up to 30% of total final energy consumption in buildings in developed countries takes the form of heat used to prepare domestic hot water. Hence, this thesis develops a spatial-temporal method for annual hot water demand in conjunction with space heating demand while technically and economically assessing the expansion potential of the district heating systems. The main findings show that the actual district heating can be increased four times when excluding domestic hot water and five times when considering both space heating and domestic hot water demand. The thesis also considers the heat saving potential in buildings by developing a robust information socket of the urban building stock. Such information is used for assessing the impact of energy efficiency measures on space heating demand savings and CO2 emission reduction potential in the existing buildings based on a Geographical Information System tool. The findings show that space heat demand saving potential for scenarios 1 and 2 compared to the reference scenario was 50 % and 68.5%, respectively. As the transport and heating sectors account for the largest energy consumers in energy systems, their electrification plays a major role in decarbonising the said sectors. Many solutions support the decarbonization of energy systems by increasing the integration of variable renewable energy and utilizing the synergic effect between electricity, transport and heating sectors. In that regard, this thesis emphasizes the importance of utilizing heat pumps for individual heating solutions and electric vehicles in the transport sector as the main sources for enhancing the energy system flexibility and emphasizing their consequences in thermal power plant operational capacities and efficiencies. The findings show that electrification of heating and transport sectors significantly impacts variable renewable integration and CO2 emission reduction; hence, the same poses major challenges for the nominal operation of thermal power plants. Besides using power-to-heat technologies for individual heating, the thesis also concentrates on increasing the share of variable renewable using power-to-heat technologies in district heating systems. The main goal is to identify the influence of using district heating systems coupled with the power-to-heat technologies based on the flexible operation of coal-based thermal power plants and limited electricity system interconnections on the maximum integration of variable renewables. Results show that wind and PV power plant capacities installed in the existing Kosovo energy system, when operating in an isolated mode, are 450 MW and 300 MW, respectively. Additional capacities around 800 MW for wind and 385 MW for PV can be further integrated into this isolated energy system with the contribution of power-to-heat technologies coupled with thermal energy storage in district heating with a fixed capacity. Finally, the previous findings were used to asses sustainable energy transition pathways for coal-based energy systems. The method shows how the scaling-up in variable renewable energy sources and sector coupling (electricity and heating) while maintaining high flexibility in thermal power plants can shed light on achieving sustainability in a coal-based energy system. Five different scenarios have been created. Significant differences in annualized technology and emission costs can be observed between scenarios. In addition, scenario three seems to have the least cost in comparison to other scenarios. The total CO2 emissions for projected scenarios 1, 2, 3, 4, 5 in 2030 accounted for 4.78, 5.28, 4.48, 3.97 and 4.95 MtCO2/year. In addition, the total annual costs for projected scenarios 1, 2, 3, 4, 5 in 2030 accounted for 2168, 1611, 1993, 2479 and 2817 Mil. EUR respectively. Rad ne sadrži sažetak na drugom jeziku.

Odabir učinkovitog algoritma za podudaranje znakovnih nizova u tekstu izazovan je proces. Na izvršavanje algoritma mogu utjecati razni čimbenici poput jačine procesora, veličine radne memorije računala ili mogućnosti programskog jezika. U ovome radu predstavljena je metodologija izgradnje domenskih modela vrednovanja algoritama za podudaranje znakovnih nizova u tekstu temeljena na entropiji uzorka, izražavajući učinkovitost algoritma korištenjem metrika neovisnih o hardverskoj i softverskoj platformi. Razvijena metodologija je usmjerena prema algoritmima za egzaktnu usporedbu znakova prilikom podudaranja uzorka u tekstu bez prethodne spoznaje strukture teksta. Metodologija rangira algoritme prema njihovoj učinkovitosti koristeći svojstva uzorka koji se pretražuje, a ne ovisi o implementaciji algoritma ni o računalnoj arhitekturi te stoga pruža neovisan način odabira algoritma za određenu domenu primjene. Metodologija izgradnje domenskih modela vrednovanja algoritama eksperimentalno je potvrđena na dvije domene, DNA domeni i domeni prirodnog jezika. U radu je prezentiran i novi algoritam za egzaktno podudaranje znakovnih nizova. Heuristička tehnika prolaska kroz uzorak pomoću neparnih i parnih pozicija znakova u kombinaciji s podacima dobivenim predobradom uzorka oblikovana je pravilima algoritma. Definirana heuristička pravila koriste se za preskakanje nepotrebnih usporedbi znakova uzorka. Točnost i pouzdanost algoritma je eksperimentalno vrednovana. Choosing an efficient algorithm for strings matching in a text is a challenging process. The algorithm execution can be affected by various factors such as the processor power, the size of the computer memory, or the programming language capabilities. This doctoral thesis presents the methodology for building domain models for strings matching algorithms evaluation based on pattern entropy, expressing the algorithm's efficiencyusing hardware and software platform independent metrics. The developed methodology is focused on algorithms for exact string comparison to match the pattern in the text without a prior knowledge of the text structure. The methodology ranks algorithms according to their efficiency using the properties of the pattern being searched, and does not depend on the implementation of the algorithm or the computer architecture and therefore provides an independent way of selecting an algorithm for a particular application domain. The methodology for building domain models for evaluating algorithms has been experimentally confirmed on two domains, the DNA domain, and the natural language domain. The doctoral thesis also presents a new algorithm for exact string matching. The heuristic technique of passing through a pattern using odd and evenpositions of characters, in combination with the data obtained by preprocessing the pattern, is shaped in the algorithm rules. Defined heuristic rules are used to skip unnecessary comparisons of pattern characters. The accuracy and reliability of the algorithm were experimentally evaluated.

Tržišne okolnosti i integracija obnovljivih izvora energije značajno su izmijenile proces planiranja prijenosne mreže. Porastom opterećenja i udjela obnovljivih izvora u ukupnim proizvodnim kapacitetima prijenosne mreže javlja se potreba za povećanjem kapaciteta prijenosa električne energije. Izgradnja novih dalekovoda kompleksan je proces koji uključuje izradu studija, ishođenje dozvola, rješavanje imovinskopravnih odnosa, provođenje javne nabave i izgradnje. Navedeni proces traje godinama i može značajno odstupati od inicijalno planiranih vremenskih rokova. Doktorski rad analizira mogućnosti povećanja kapaciteta postojećih dalekovoda uzimajući u obzir primjenjivost razmatranih rješenja u hrvatskom prijenosnom sustavu. Provedena istraživanja sastoje se od tri međusobno povezana dijela. Prvi dio odnosi se na razvoj metode za odabir mogućih dalekovoda koji nemaju dovoljne kapacitete prijenosa u sadašnjem ili budućem promatranom razdoblju. Pri određivanju pogodnih lokacija provodi se analiza sadašnjeg stanja prijenosne mreže i analiza koja uključuje razmatranje prijenosne mreže s obzirom na buduća planirana kretanja uvoza, izvoza, povećanja proizvodnje i opterećenja na prijenosnoj mreži. Za odabrane dalekovode razmatra se povećanje kapaciteta primjenom kompaktiranja nadzemnih vodova i ugradnjom visokotemperaturnih vodiča malog provjesa (engl. High Temperature Low Sag Conductors – HTLS). Za predložena rješenja razmotrit će se utjecaj na prijenosnu mrežu kroz utjecaj na postojeći kapacitet prijenosne mreže i izračun gubitaka. Za oba parametra razvijeni su modeli koji uzimaju u obzir realne uvjete i mjerenja iz pogonskog stanja prijenosne mreže. U drugom se dijelu na temelju koordinacije izolacije određuju potrebni stupanj izolacije za predloženi 400 kV kompaktirani dalekovod i potreba za ugradnjom odvodnika prenapona. Broj odvodnika prenapona određuje se na temelju zahtijevane razine prenaponske zaštite i tehno-ekonomske analize. Raspored odvodnika na dalekovodu određen je proračunom koordinacije izolacije gdje se za predloženi 400 kV kompaktirani dalekovod razmatraju sklopni prenaponi na temelju rezultata simulacija u programskom alatu EMTP. Uz prihvatljiv ekonomski trošak moguće je dodatno povećati raspoloživost predloženog 400 kV kompaktiranog dalekovoda. Treći dio rada odnosi se na provjeru dielektričnih svojstava predloženog kompaktiranog rješenja. Za 400 kV kompaktirani dalekovod napravljeni su izračuni električnih i magnetskih polja standardnim izračunima i metodom konačnih elemenata (Finite Element Method) u programskom okruženju ANSYS. Prikazana je raspodjela el. potencijala i el. polja na pojedinim elementima dalekovoda (stup, vodič, izolator i ovjesna oprema) te su napravljene analize u sklopu kojih se provjeravaju električni parametri odabranog kompaktiranog rješenja. Provedena je termička analiza za HTLS vodiče, ovjesnu opremu i izolator kako bi se utvrdilo temperaturno ponašanje pojedinih komponenti za različite scenarije. Provedena analiza uključuje izračune temperature vodiča IEEE standardom i temperaturnu raspodjelu na površini i središtu vodiča primjenom FEM metode. Za potpuno razumijevanje procesa generiranja i odvođenje topline napravljena je analiza koristeći CFD (Computational Fluid Dynamics) simulacije. CFD simulacijom kroz primjenu multifizičkog modela detaljnije se sagledava temperaturna raspodjela promatranih elemenata za navedene procese. Koristeći numeričke postupke FEM i CFD moguće je odrediti površinsku i prostornu temperaturu promatranih elemenata. Integration of renewable energy sources and market circumstances have significantly changed the transmission network planning process which leads to the need to increase capacity of transmission network. It is necessary to better use of the existing transmission grids in purpose to transferee wind and solar power to where it’s needed. Lack of capacity on key points of the transmission grid is also postpone new renewable energy projects from being developed and build. To achieve integration of renewables it is necessary to build new overhead lines (OHL) or increase capacity of existing ones. Building new OHL is an expensive and time-consuming process that highly depends by landowner, lawsuits and unfavorable regulatory rulings although renewables developers and their offtakers can afford it. Process of building new OHL takes years and may deviate significantly from the initially planned deadlines. New approach to increase transmission capacity is needed which include new technologies that can help to solve the problem. Possible solution that can be integrated into the grid with low cost and can have significant amounts in congestion managment are power flow control, dynamic line ratings and topology optimization. Each of mentioned technology doesn’t provide long term solution and is not applicable in each case of solving transmission congestion and additional solution must be considered. One of the way to increase OHL capacity is to increase operating current or voltage level on existing OHL. Increasing current on existing OHL as a consequence have increase in conductor temperature, melt or fail conductor strength and increase in sag. In that purpose new conductor was developed that can withstand mentioned demands (High Temperature Low Sag Conductors HTLS). HTLS conductor can withstand permanent operating temperatures up to 210 °C carrying higher power compared to conventional conductors. These conductors have a wide range of application possibilities when there is need for increase transmission capacity, clearance problems and restrictions to the use of new and higher towers. Big advantage of HTLS conductors is fast installation comparing to building/reconstructing OHL and without need to modify most of the existing towers. Increasing voltage level on existing OHL is considered in cases when there is need to increase transmission capacity where building new OHL is not possible. Voltage uprating on existing OHL demands investigation on electrical parameters and mechanical parameters of existing infrastructure OHL. Main characteristics of voltage uprating is smaller clearance, composite insulators, coordination insulation, overvoltages and main electrical characteristics (natural power, electrical gradient, corona etc.). For that reason (smaller clearance) and designee solution such OHL is named compact OHL. In some cases voltage uprating with compact designee requires modification and foundation reinforcement of existing tower or new tower designee and foundation. Main advantage of compact OHL is higher power transfer, better stability in operation, smaller impact on the environment, smaller right-of-way compared to conventional solution. However, the cost of the voltage uprating is high and requires additional considerations comparing to conventional methods in process of reconstructing or building OHL. Motivation for dissertation came from desire to consider the implementation HTLS and compact designee technologies in the transmission network when the main aim is to increase the capacity of the existing OHL. The aim of the research is to provide better insights into the behavior of HTLS conductors and compacted transmission lines in the transmission network taking into account the transmission network development plans and the influence of chosen solutions. Conducted analyzes consider increase in transmission capacity and impact of losses in the transmission network for different types for selected technologies. Recently transmission system operator become responsible for amount and cost of losses and new investment must be optimized from prospective that the lifetime of these investments is longer than 40 years. For any type of investments it is necessary to determine possibilities of applicability taking account of other limitations. For compact solution in this dissertation will be consider dielectric properties and electric fields on individual components of the transmission line and in the transmission line corridor. The application of HTLS conductors allows to increase the temperature of conductors and consequently suspension equipment and insulators. To determine radial temperature distribution in the HTLS conductor and to establish how the temperature expands thermal model of HTLS conductor will be proposed. For both approaches additional verification of electric and thermal parameters will be verified using FEM and CFD numerical programs. Doctoral thesis analyzed the possibilities of increasing the capacity of existing transmission lines with the main goal that proposed solution are applicable in the Croatian transmission system. Dissertation can be divided in three main parts that represent process from the beginning of determine which OHL doesn’t have satisfactory capacity, consider possible solution and true the additional research determine technical parameters of proposed designee. The first part of doctoral thesis presents overview of the parameters that influence where and when is increase in transmission capacity is needed. The process of determining the limitation of existing OHL in transmission network using the procedure in which it is necessary identify input data, perform additional calculations and analyzes considering future plans and in the end interpret obtained results. Presented process is algorithm that consider implementation compact designee and HTLS conductor with the aim to increase transmission network capacity. The algorithm presents steps in terms of a systematic procedure that involve the collection and data processing, calculations and decision-making criteria. The computational parts of the algorithm are based on power flow calculations for chosen scenarios conditions that consider development plans of transmission network. After determining the critical transmission OHL, the sensitivity analysis is carried to determine electrical parameters of OHL which need to be reconstructed in order to prevent possible congestion in the network. Conducted sensitivity analysis is perform considering increase capacity of each OHL and capacity of transmission network considering influence on losses. Determining the capacity of the transmission network are the main issues in the planning and management of the transmission network and at the same time an extremely important basis for planning the development of the transmission network. Due the several calculations of transmission network capacity methods Net Transfer Capacity and Total Transfer Capacity calculation are considered dominant and used in conducted considerations. Model for monitoring losses on each transmission lines is developed and presents an improvement of the existing process of determining Joule and corona losses on high-voltage transmission lines. From the available measurements and available input data, prediction of losses can be made for each OHL considering predicted power flows. The results of the losses assessment model have significant importance and is used in purpose of planning and procurement of losses leading to reduced costs for proposed solution. Furthermore, ability to manage electricity losses in the transmission system has practical significance, since they are an important technical and economic indicator of transmission system management. As was mentioned before compact designee have a smaller clearance and to achieve that applications of composite insulators as insulating cross-arms is needed. Due the smaller clearance temporary overvoltages can occur and their amounts need to be limited to acceptable level. For the proposed solution of 400 kV compacted transmission line calculation of insulation coordination is conducted according to IEC 60071-2. According to the results installation of surge arrester is needed where optimum installation locations is determine based on amount of temporary overvoltages and techno economical method that was presented. The third part of doctoral thesis presents calculations using FEM to evaluate electromagnetic behavior of individual transmission line components and thermal behavior of HTLS conductor comparing to existing conventional methods. Compact designee implies smaller clearance which causes increase in electrical filed on components in OHL tower. Although the design principles are similar to those for classic OHL designee, the margins are additional reduced, so special attention is made to understand how geometrical configuration influence on spreading electrical field on each elements of OHL. In that purpose 2D and 3D FEM model is developed to better understand difference between OHL designee. Conducted analysis consider distribution of electrical potential in tower OHL, electrical field on insulators, suspension equipment and potential gradient on conductors. In order to better understand the influence of conductor geometry on electric field different type of conductor was analyzed. The influence of compact designee on electrical and magnetic field in for OHL right of way OHL was also demonstrated. Conducted research gives overview of technical aspects that need to be consider before making final compact designee from dielectrically point of view. For HTLS conductors and insulator thermal analysis was performed using the FEM method to determine the temperature behavior compered to ACSR conductors. To determine conductor temperature IEEE 738 standard was applied for ACSR, ACCR and ACCC Drake conductors. For all three conductor influence of speed and angle of wind, insolation and air temperature on conductor current was performed. The results are used as input data for modeling 2D conductors in FEM analysis. For each of the three mentioned conductors temperature distribution was calculated using FEM Ansys. In addition to fully understand the process of generating and dissipating heat, an analysis was made using computational fluid dynamics (CFD) simulations. CFD simulation through the application of a multi-physical model examines in more detail the temperature distribution of conductor for processes of dissipating heat by velocity of air and heat radiation. Using the numerical procedures FEM and CFD it is possible to determine the surface and internal temperature of the observed elements and consider thermal limitation of conductors and insulator. Research in doctoral is relevant for transmission network development and revitalization power system. Scientific contributions of the dissertation are: algorithm for determining electrical parameters of compacted transmission lines and HTLS conductors with respect to the increase of transmission capacity and reduction of losses in the transmission system; model for selection conductor characteristics and configuration based on performed analyzes of electromechanical and thermal parameters for compacted transmission lines; multicriteria model for compact configuration selection with the aim of determining the availability of compacted transmission line.

The increasing share of renewable energy sources (RES) is one of the main goals in the transition towards a carbon-neutral power system. Due to their intermittent nature, RES require an additional level of flexibility to ensure stable and secure power system operation. In the past, the flexibility used to be provided from conventional power plants, such as coal power plants which are less flexible and open cycle gas turbine (OCGT) and natural gas-fired internal combustion engine (ICE) power plants with a high level of flexibility characterized with short minimum uptime and minimum downtime, short start-up time and low start-up cost. However, the enlarged need for flexibility in green energy transitions opens the doors for providing system services from the end-user's side. The role of final customers has been under significant changes. From completely passive entities, final customers are becoming active network participants involved in the real time (RT) power system operation. With the recent price decrease of low carbon technology, more final customers tend to invest in photovoltaic (PV) systems, household battery energy storage (BES), electric vehicles (EV), and household energy management systems (HEMS). To fully exploit their potential, European directives highlight the importance of providing transparent and easy-to-understand pricing mechanisms together with pricing comparison tools in order to ensure fair and the most economic solutions to all final customers. As some final customers, such as households, are too small to independently trade their flexibility on the market, a new market entity called aggregator gathers their flexibility potential and trades with their services on the market. Final customer's flexible behaviour can be remunerated through direct incentives, electricity bill reduction or even a discount for investing in low carbon technology and fixed annual direct payments for flexible management of different household appliances. The thesis investigates the impact of diverse pricing mechanisms on the behaviour of final customers, their consumption profile, and electricity bill in different European countries considering not only energy prices, but also transmission and distribution network charges, tax, VAT, and the pricing policy related to RES installed at final customer's side. Flat prices, two-tariff pricing, and RT dynamic pricing for prosumers are compared resulting in more flexible behaviour and electricity bill reduction under dynamic prices. In general, the current pricing structure encourages final customers with installed RES to shift their flexible consumption in the periods of high production, instead of selling excess to the supplier. Furthermore, the thesis deals with different forms of aggregation resulting not only in financial benefits to all entities involved but also reducing the negative impact of the intermittent nature of RES. The first type of aggregation investigated in the thesis is a formation of an innovative balancing group (BG) composed of the aggregator of flexible prosumers and a wind power plant (WPP) participating in the day ahead (DA) energy market and balancing market. The focus is put on joint market participation which reduces prosumers' electricity cost and increases WPP profit due to decreased balancing cost arising from flexible behaviour of final prosumers which serve as a buffer between balancing market and the intermittent and variable WPP's production. The second type of aggregation in the thesis is focused on an energy community. The energy community will participate in the power system transition by allowing the participation of citizens in the clean energy transition in diverse legal forms, such as associations, cooperatives, partnerships, non-profit organizations, and small/medium enterprises. Community members investigated in the thesis are equipped with PV, flexible electric heating, flexible household appliances, and have an opportunity of smart EV charging. The community is operated by the community manager (CM) who is in control of flexible household devices through HEMS. The CM tackles the issue of the stochastic nature of PV and households' consumption by providing flexibility incentives to community members in order to reduce the deviation from preannounced DA schedule. The energy community participates on the market as a single entity and the cost of community members is calculated ex-post, after energy delivery which makes the optimization algorithm simple to solve. Prices for the internal energy exchange between community members are calculated based on three pricing mechanisms: Bill Sharing Method Net (BSMN), MidMarket Rate Net (MMRN), and Supply Demand Ratio Net (SDRN) taking into account the net load of each community member, DA energy prices and flexibility incentives. Participation in the energy community together with the price calculation developed in the thesis which is conducted in a two-stage process guarantee lower electricity cost to each community member compared to their individual contract with the supplier. Ambiciozni ciljevi Europske unije teže smanjenju stakleničkih plinova za 55% do 2030. godine u usporedbi s emisijama iz devedestih godina prošlog stoljeća kako bi se smanjio globalni porast temperature na 1.5°C uz povećanje energetske učinkovitosti. Kako bi se to ostvarilo, potrebno je povećanje udjela obnovljivih izvora energije (OIE) što je jedan od glavnih ciljeva tranzicije prema niskougljičnom elektroenergetskom sustavu. Upravljanje elektroenergetskim sustavom s visokim udjelom OIE je složenije i zahtjevnije od upravljanja tradicionalnim elektroenergetskim sustavom. Zbog svoje isprekidane prirode, OIE zahtijevaju dodatnu razinu fleksibilnosti kako bi se osigurao stabilan i siguran rad elektroenergetskog sustava. U prošlosti su fleksibilnost pružale konvencionalne elektrane, kao što su manje fleskibilne elektrane na ugljen te plinske turbine otvorenog ciklusa (OCGT) i elektrane s unutarnjim izgaranjem na prirodni plin (ICE) s visokom razinom fleksibilnosti koju karakterizira kratko minimalno vrijeme prije ponovnog gašenja i paljenja, kratko vrijeme pokretanja i niski troškovi pokretanja. Unatoč svim izazovima koji se javljaju u tranziciji prema niskougljičnom elektroenergetskom sustavu, operator sustava dužan je na sigurnan i efikasnan način planirati i upravljati sustavom. Povećana potreba za fleksibilnošću u zelenoj energetskoj tranziciji otvara vrata za pružanje pomoćnih usluga od strane krajnjih korisnika čija se uloga značajno mijenja. Od pasivnih entiteta, krajnji kupci postaju aktivni mrežni sudionici uključeni u rad elektroenergetskog sustava. S nedavnim smanjenjem cijena niskougljičnih tehnologija, sve više krajnjih kupaca ulaže u fotonaponske (PV) sustave, baterijske spremnike, električna vozila (EV) i sustave za upravljanje energijom u kućanstvu (HEMS). Kako bi krajnji kupci u potpunosti iskoristili svoj potencijal, europske direktive ističu njihovu važnost te ih stavljaju u centar niskougljične tranzicije. Posebno je istaknuto da se svim krajnjim kupcima moraju omogućiti poštena i ekonomična rješenja pružajući im na tranparetan i lako razumljiv način različite mogućnosti odabira cijena električne energije, zajedno s alatima za usporedbu cijena, koji će omogućiti brzu i prikladnu promjenu opskrbljivača. Krajnji kupci trebaju imati točne informacije o svojem obračunu koje se temelje na stvarnoj potrošnji električne energije što im omogućuje bolju kontrolu vlastite potrošnje i troška električne energije. Zbog premalih snaga krajnjih kupaca, novi tržišni subjekt agregator okuplja krajnje korisnike i njihovu fleksibilnost te sudjeluje na različitim tržištima. Kako bi se agregatorima omogućilo posrednost između tržišta i krajnjih kupaca te kako bi se osiguralo da krajnji kupac ostvaruje odgovarajuću korist od njihove aktivnosti, svaki krajnji kupac treba biti obaviješten o izabranom modelu koji sadržava transparentna i poštena pravila. Kako bi se na pošten način potaknulo upravljanje potrošnjom, usluge koje nude agregatori trebaju biti definirane na svim tržištima, uključujući tržišta električnom energijom, tržišta pomoćnih usluga i tržišta kapaciteta. Ugovor o agregiranju svaki krajnji kupac može sklopiti neovisno o svom opskrbljivaču. Fleksibilno ponašanje krajnjeg kupca može se nagraditi izravnim poticajima, smanjenjem računa za električnu energiju ili čak smanjenom investicijskom cijenu za brojne niskougljične tehnologije i fiksnim godišnjim izravnim plaćanjima za fleksibilno upravljanje kućanskim uređajima. Postoje dvije grupe odziva potrošnje na temelju kojih krajnji korisnici mogu pružiti fleksibilnost elektroenergetskom sustavu. Prva grupo obuhvaća cjenovno potaknute programe odziva potrošnje, dok druga grupa obuhvaća programe stimulirane različitim poticajima. U cjenovno potaknute programe spadaju različite opcije promijenjivih cijena električne energije: • Višetarifini sustavi - dvije ili više unaprijed određenih tarifa koje se izmijenjuju tijekom dana u unaprijed određenim fiksnim intervalima za vrijeme trajanja ugovora. Svaka tarifa pokriva dulji vremenski interval tijekom dana (npr. niže cijene tijekom noći i više cijene tijekom dana). Visoke cijene tijekom vršnog opterećenja - značajni porast cijena električne energije za vrijeme vršnog opterećenja tijekom nekoliko kritičnih dana u godini koji su unaprijed najavljeni. • Popusti tijekom vršnog opterećenja - popusti krajnjim korisnicima koji pristanu smanjiti svoju potrošnju za vrijeme vršnog opterećenja. • Postupno povećanje tarife ovisno o ukupnoj potrošnji - količina potrošene električne energije podijeljena je u više kategorija. Svaka kategorija odgovara određnoj cijeni koja raste s povećanjem potrošnje i prelaskom u iduću kategoriju. Ukupna mjesečna cijena jednaka je sumi umnoška cijene u pojedinoj kategoriji i količini potrošene električne energije u toj kategoriji. • Dinamičke cijene na satnoj razini - cijene određene dan-unaprijed koje variraju na satnoj razini te prate stanje na tržištu električnom energijom. Programi stimulirani poticajima mogu se podijeliti u četiri kategorije: • Direktno upravljanje potrošnjom - najčešće direktno upravljanje klimatizacijskim uređajima, grijanjem i bojlerima od strane operatora sustava. Operator sustava unaprijed odredi maksimalni broj aktivacija i vrijeme trajanja usluge. S obzirom na to da krajnji korisnici ugovorom pristanu na izravno upravljanje uređajima od strane operatora sustava, najčešće se ne šalje nikakva obavijest krajnjim potrošačima ili se pošalje neposredno prije aktivacije usluge. Krajnji potrošači dobivaju fiksni iznos naknade određene ugovorom te dodatnu nakandu ukoliko je operator sustava uslugu aktivirao. • Programi smanjenja potrošnje - programi u kojima se unaprijed najavi krajnjim korisnicima kada je potrebno smanjiti potrošnju (nekoliko minuta, sati ili čak dan unaprijed). Krajnji korisnik smanjuje svoju potrošnju automatiziranim načinom ili ručno, ovisno o uvjetima dogovorenim u ugovoru. • Programi prekida potrošnje - krajnji korisnici u ovom programu pristaju na djelomični ili potpuni prekid potrošnje za vrijeme vrlo visokih cijena električne energije ili u trenucima kada je ugrožena sigurnost elektroenergetskog sustava. • Tržišno orijentirani programi - obuhvaćaju agregiranje krajnjih korisnika od strane agregtora koji sudjeluje na tržištu pomoćnih usluga. Doktorska disertacija istražuje utjecaj različitih cijenovnih mehanizama na krajnje kupce u europskim zemljama, uzimajući u obzir ne samo cijene električne energije, već i naknade za prijenosnu i distribucijsku mrežu, različite vrste poreza i cjenovnu politiku vezanu uz OIE instalirane na strani krajnjeg kupca. Iako krajnji kupci imaju mogućnost odabira dinamičkih cijena koje reflektiraju tržišne cijene samo u osam zemalja u Europi, energetske politike ističu da države članice moraju omogućiti sklapanje ugovora s dinamičkim cijenama za krajnje kupaca s ugrađenim pametnim brojilima od strane barem jednog opskrbljivača. U disertaciji se uspoređuju troškovi krajnjih korisnika koji imaju mogućnost odabira jednotarifne cijene, dvotarifne cijene i dinamičke cijene pokazujući fleksibilnije ponašanje i smanjenje računa za električnu energiju kod dinamičkih cijena. Trenutna struktura cijena (niže prodajne cijene od kupovnih cijena) potiče krajnje kupce da preusmjere svoju fleksibilnu potrošnju u razdoblje s visokom proizvodnjom iz OIE, umjesto da višak prodaju opskrbljivaču. Nadalje, doktorska disertacija se bavi različitim oblicima agregacije koji rezultiraju ekonomskim povlasticama za sve uključene subjekte, ali i smanjenjem negativnog utjecaja varijabilne proizvodnje iz OIE. Prva vrsta agregiranja razmatrana u disertaciji je formiranje inovativnog oblika bilančne grupe koja se sastoji od agregatora krajnjih kupaca s vlastitiom proizvodnjom i vjetroelektrane (VE) koji sudjeluju na na tržištu dan unaprijed i na tržištu uravnoteženja. Problem je definiran kao stohastički dvorazinski model mješovitog cjelobrojnog programiranja baziran na Stackelbergovoj igri. U gornjoj razini promatra se maksimizacija profita agregatora i VE koji predstavljaju vođe u Stackelbergovoj igri određujući cijene aktivnim kupcima s vlastitiom proizvodnjom. Donja razina minimizira trošak električne energije aktivnih kupaca s vlastitom proizvodnjom koji na temelju cjenovnih signala iz gornje razine optimiziraju svoju potrošnju te sudjeluju u razmjeni energije s VE nastojeći smanjiti njena odstupanja od dan-unaprijed ugovorenog profila proizvodnje upravljajući vlastitim baterijskim spremnicima i fleksibilnim grijanjem. Promatrane su tri različite mogućnosti određivanja cijena krajnjim kupcima: unaprijed dugoročno definirane dvotarifne cijene, dinamičke promijenjive cijene koje određuje agregator optimizacijom te tržišne cijene uvećane za fiksnu naknadu od strane agregatora. Fokus je stavljen na uspredbu individualnog nastupa agregatora i VE na tržištu te njihovo zajedničko sudjelovanje kojim se smanjuju troškovi električne energije krajnjih kupaca i povećava profit VE zbog smanjenih troškova uravnoteženja koji proizlaze iz fleksibilnog ponašanja krajnjih kupaca koji služe kao tzv. amortizer između tržišta uravnoteženja i isprekidane proizvodnje iz VE. Rezultati su pokazali da dinamičke promjenjive cijene koje određuje agregator optimizacijom potiču krajnje korisnike na fleksibilno ponašanje kada je potrebno pomoći VE u smanjenju odstupanja. Tako određene cijene pogoduju i agregatoru i krajnjim korisnicima jer optimizacija uzima u obzir i cijene električne energije na tržištu te potiče krajnje korisnike na smanjenje potrošnje za vrijeme visokih cijena čime se direktno smanjuje trošak krajnjih korisnika i povećava profit agregatora. Model je dodatno istražio kako povećanje broja članova bilančne grupe koji imaju baterijski spremnik utječe na tijek novca unutar grupe. Povećanje broja krajnjih korisnika s baterijskim spremnikom u bilančnoj grupi ne umanjuje značajno trošak bilančne grupe. Međutim, kod tržišnih cijena uvećanih za fiksnu naknadu od strane agregatora povećanjem broja baterija u bilančnoj grupi značajno se smanjuje trošak krajnjih korisnika. Upravo zato što su te cijene više orijentirane na krajnjeg korisnika i smanjenje njegovog troška, ne potiču pružanje usluge fleksibilnosti VE. Druga vrsta agregacije usmjerena je na energetsku zajednicu. Energetska zajednica sudjelovat će u rekonstrukciji elektroenergetskog sustava sudjelovanjem građana u zelenoj tranziciji u različitim pravnim oblicima, poput udruga, zadruga, partnerstava, neprofitnih organizacija i malih / srednjih poduzeća. Metode izračuna troškova u energetskoj zajednici svrstane su u tri kategorije: modeli bazirani na teoriji igara, modeli bazirani na formiranju koalicije i modeli u kojima se cijene računaju ex-post izvan optimizacijskog algoritma. Modeli bazirani na teoriji igara su računski zahtjevni te njihova složenost raste eksponencijalno s brojem članova. Modeli bazirani na formiranju koalicije ponekad su ograničeni s ukupnim brojem članova u koalciji što spriječava formiranje velike koalicije koja omogućuje ostvarivanje najvećih ušteda što potencijalno dovodi do financijskog nezadovoljstva članova zajednice. Modeli u kojima se cijene računaju ex-post jednostavni su za implementaciju i garantiraju konvergenciju, međutim neki od njih ne rezultiraju nižim troškovima članovima zajednice. Model predstavljen u disertaciji riješio je taj problem uvođenjem dvostupanjskog računanja troška električne energije. Članovi zajednice u disertaciji opremljeni su solarnim panelima, fleksibilnim električnim grijanjem, baterijskim spremnikom, fleksibilnim kućanskim aparatima i imaju mogućnost pametnog punjenja EV. Zajednicom upravlja upravitelj zajednice (UZ) koji kontrolira fleksibilne uređaje u kućanstvu putem HEMS-a. UZ uzima u obzir stohastičku prirodu solarnih panela i potrošnje kućanstava pružajući poticaje za fleksibilnost članovima zajednice kako bi se smanjilo odstupanje od dan-unaprijed najavljenog rasporeda. Uštede ostavrene pametnim upravljanjem uređajima u kućanstvu dijele se među članovima energetske zajednice. UZ je entitet drugačiji od tradicionalnog opkrbljivača jer mu cilj nije ostvarivanje profita. To je zapravo platforma koja članovima zajednice pruža različite opcije pri monetiziranju fleksibilnosti, ali ih izlaže i riziku promjenjivosti cijena koji je inače tradicionalno snosio opskrbljivač. Kod tradicionalog ugovora s opskrbljivačem opisanog u disertaciji prema danskom modelu maloprodajnog tržišta (Danska je odabrana u disertaciji zbog javno dostupnih podataka o svim komponentama računa električne energije, međutim model se može primijeniti na bilo kojem maloprodajnom tržištu koje nudi mogućnost odabira dinamičkih cijena) krajnji korisnik snosi trošak električne energije, trošak uravnoteženja po svakom kupljenom ili prodanom kWh električne energije te mrežne trošakove za svaki kupljeni kWh električne energije. Energetska zajednica sudjeluje na tržištu kao jedna cjelina, a troškovi članova zajednice računaju se dan nakon isporuke energije, izvan optimizacijskog algoritma što ga čini jednostavnim za rješavanje. Energetska zajednica kao cjelina snosi trošak električne energije i mrežne trošakove za svaki kupljeni kWh električne energije. Za razliku od individualnog ugovora s opskrbljivačem gdje se plaća određeni trošak uravnoteženja za svaki kupljeni i prodani kWh električne energije, energetska zajednice snosi troškove uravnoteženja kada odstupa od dan-unaprijed predviđenog rasporeda. Smanjenje troškova uravnoteženja ostvareno je fleksibilnim poticajima. Cijene za unutarnju razmjenu energije između članova zajednice računaju se na temelju tri mehanizma određivanja cijena: neto metoda dijeljenja računa (eng. Bill Sharging Method Net), neto srednje-tržišna stopa (eng. Mid-Market Rate Net) i neto omjer opskrbe i potražnje (eng. Supply Demand Ratio Net), uzimajući u obzir neto potrošnju svakog člana zajednice (razliku između ukupne potrošnje i proizvodnje na satnoj razini), dan-unaprijed cijene energije i poticaje za fleksibilnost. Cijene se računaju u dvostupanjskom postupku koji jamči niže troškove električne energije svakom članu zajednice u odnosu na njihov pojedinačni ugovor s opskrbljivačem. U prvom stupnju računaju se cijene za kupnju i prodaju električne energije unutar energetske zajednice za svaki sat i za svaku metodu zasebno. Nakon toga slijedi izračun ukupnog dnevnog ili mjesečnog troška svakog člana zajednice uzimajući u obzir da li je član zajednice u pojedinom satu kupac ili prodavatelj električne energije. U drugom stupnju uspoređuje se trošak svakog člana zajednice s troškom pri individualnim ugovorom s opskrbljivačem. Ukoliko bilo koji član energetske zajednice ima viši trošak električne energije na kraju obračunskog razdoblja (dnevnog ili mjesečnog) radi se preraspodjela troškova. Preraspodjela troškova omogućuje da svaki član zajednice ima manji trošak unutar zajednice u usporedbi s individualnim ugovorom s opskrbljivačem. Izračuna se minimalna vrijednost potrebne preraspodjele troškova na temelju razlike u pojedinačnom trošku i trošku svakog korisnika energetske zajednice te se matematičkim modelom definiranom u disertaciji dolazi do novih troškova koji garantiraju niži trošak električne energije unutar zajednice za sve članove zajednice. Od tri navedene metode izračuna cijena električne energije unutar zajednice, neto srednje-tržišna stopa i neto omjer opskrbe i potražnje garantiraju niže cijene električne energije svakom članu zajednice. Zbog specifičnosti izračuna cijena u neto metodi dijeljenja računa, članovi zajednice s vlastitom proizvodnjom snose viši trošak električne energije jer se višak proizvodnje unutar zajednice dijeli besplatno, čime profitiraju isključivo krajnji kupci bez vlastite proizvodnje.

In comparison to other methods used for long-term monitoring of fetal health, fetal phonocardiography has the potential to be more convenient and affordable due to its non-invasive nature and the possibility of implementation on omnipresent devices such as smartphones. Fetal phonocardiography signals can oftentimes be misinterpreted due to various sources of sound in the womb. Therefore, the question remained whether a machine learning model trained for fetal heartbeat detection containing a conventional set of audio features could be improved by introducing features taken from signal representations processed by two methods: empirical mode decomposition (EMD) and pitch shifting. Furthermore, features based on psychoacoustics were proposed as an additional input to the model. In other words, the main goal of this research was to employ EMD and pitch shifting as preprocessing steps, as well as psychoacoustics descriptors such as perceptual linear prediction coefficients, in order to enable the utilization of additional characteristics from the phonocardiography signal. Features extracted in this fashion were assessed through the analysis of their relevance, usefulness and significance in relation to the performance metrics, such as accuracy or precision. Two raw datasets of audio data were employed as input, one custom recorded and collected through fetal heartbeat acquisition obtained from 8 pregnant women, while the other was taken as an available dataset of simulated fetal heartbeat sounds with different noise levels. Two scenarios with different inputs were introduced in this research. In Scenario A, the custom dataset was utilized to train a machine learning model from features originating from raw, filtered and EMD-processed versions of the audio signal. The results consistently indicated high ranking of features based on EMD and their ability to improve general detection accuracy once they were introduced to the set of audio features. Namely, the selected subset of combined audio and EMD-based features in comparison to all audio features, improved the detection accuracy by up to 10.28%. Scenario B contained 6 different cases, incorporating 3 extracted datasets generated from custom raw data and variable segmentation window lengths and 3 extracted datasets acquired through taking the simulated raw dataset with 3 different signal-to-noise ratio values for the fetal heartbeat sound signal. The results of feature selection and ranking methods indicated consistently high relevance of psychoacoustic features, especially in the case where frequency shifting was used as a preprocessing step. In addition to the random forest models trained with the selected feature subsets, two new classifiers (support vector machine with cubic kernel and bagged trees) were introduced to assess the impact of the entire sets of characteristics added through the proposed preprocessing and feature extraction methods. The analysis showed that the included dimensionality gained through the pitch shifting and EMD in the preprocessing steps with audio and psychoacoustic feature extraction raised the detection accuracy to a higher level, reducing misclassification rate up to nearly 3 times in some instances. As a final result of this research, the impact of the proposed preprocessing and feature extraction methods in the automatic FPCG classification was shown to be substantial. In addition, the showcased approaches have been demonstrated to be feasible for the implementation in an algorithm for real time usage, further highlighting the possible benefits of the system and its components in the biomedical industry. U usporedbi s drugim metodama za dugoročno praćenje zdravlja fetusa, fetalna fonokardiografija ima veliki potencijal postati prilagođenija i jeftinija alternativa s obzirom na neinvazivnost i mogućnost implementacije na svakodnevnim uređajima. Signali dobiveni navedenom metodom mogu često biti krivo interpretirani s obzirom na visoku razinu šuma koji proizlazi iz ostalih izvora zvuka u utrobi majke. Predložene su dvije metode predobrade zvuka otkucaja srca fetusa: empirijska dekompozicija modova i promjena frekvencije. Nadalje, uz konvencionalne "audio" značajke, izlučene su i značajke bazirane na psihoakustici. Finalno, primjenjene su metode strojnog učenja u svrhu procjene važnosti pojedinih značajki, kao i evaluaciji kvalitete klasifikacije. Korištena su 2 izvora podataka: podaci snimljeni na 8 trudnica u različitim tjednima trudnoće, kao i simulirani podaci inače primjenjiivi u sličnim istraživanjima. Rezultati pokazuju poboljšanje kvalitete klasifikacije, kao i visoku rangiranost pojedinih značajke iz podskupa predloženih metoda, pogotovo u slučaju kombinacije promjene frekvencije i psihoakustičkih značajki.

Transformator je vrlo važan dio elektroenergetskog sustava pa je samim time jedan od ključnih zahtjeva tijekom projektiranja dimenzioniranje na vanjski kratki spoj kako bi se osigurao nesmetani rad. Budući je kratki spoj jedan od češćih uzroka kvarova transformatora, nužno je imati kriterije čijim bi se zadovoljavanjem u fazi projektiranja transformatora omogućilo da transformator izdrži kratki spoj. Tema otpornosti i dimenzioniranja energetskih transformatora detaljno je obrađena u literaturi, dok za suvremene distribucijske transformatore s ovalnim folijskim namotima i dijamantnim papirom nema dovoljno literature. Upravo je to tema ove disertacije. Ova disertacija bavi se određivanjem numeričkih kriterija za dimenzioniranje ovalnih folijskih namota distribucijskih transformatora na kratki spoj. Objašnjeno je ispitivanje koje transformator mora zadovoljiti zajedno s izrazima za izračun udarne (dinamičke) komponente struje kratkog spoja, kao i uvjeti za prolaznost ispitivanja prema standardu IEC 60076-5. Prolazak ispitivanja transformatora na kratki spoj potvrđuje otpornost transformatora na kratke spojeve u mreži i osigurava nesmetan rad transformatora. Uz analitičke izraze za djelovanje sile na namote u transformatoru, obrađena su i posljedična radijalna naprezanja koja se javljaju tijekom kratkog spoja. Uz naprezanje na vlak i tlak te savijanje, posebna pozornost je dana naprezanju na izvijanje. Napravljeno je opsežno ispitivanje Youngovog modula elastičnosti kompozita čiji rezultati su korišteni za definiranje kriterija za dimenzioniranje transformatora na kratki spoj. Verifikacija samih kriterija napravljena je na 43 transformatora u rasponu snaga 160 1000 kVA koji su ispitani prema standardu IEC 60076-5. Neki od ispitanih transformatora bili su iz redovne proizvodnje dok su neki napravljeni isključivo u svrhu verifikacije novih kriterija. The transformer is an important part of the power system and must be designed to be able to withhold an external short circuit. External short-circuits in the power systems are still one of the most common causes of the transformer malfunction. Therefore, it is crucial to have criteria during the designing phase the fulfillment of which would allow the transformer to withstand a short-circuit. There is a lot of published literature about power transformers and how they can be made short-circuit proof while there is not enough literature for modern distribution transformers with oval foil windings and diamond dotted paper. That is the focus of this dissertation. This dissertation describes the numerical criteria for the oval foil windings designing to be able to withstand short-circuit in the power system. The short-circuit testing procedure together with the expressions for calculating short-circuit current, as well as the conditions for passing the test according to the IEC 60076-5 standard are given. Short-circuit forces and stress in the windings is explained. Along with the analytical expressions for the forces in the windings, the radial stresses that occur during a short circuit are described. In addition to tensile and compressive stress and bending, special attention is paid to buckling stress and Young's modulus of elasticity. The verification of the criteria was done on 43 transformers in the power range 160 1000 kVA that were tested according to the IEC 60076-5 standard.

Cognitive load and emotional stress affect people’s daily lives, so the adequate assessment of cognitive load and emotional stress is of special importance. Inadequate amounts of cognitive load and emotional stress negatively affect human performance and, in the long run, the health of the individual. Assessing cognitive load and emotional stress is especially important in highly stressful professions such as the military, pilots, and air traffic controllers. Although cognitive load and emotional stress can be evaluated using different subjective questionnaires, their bias can lead to erroneous estimates. Objective methods based on neurophysiological signals collected using functional brain imaging techniques can be used to assess cognitive load and emotional stress. As part of this dissertation, two brain imaging techniques were used in cognitive load and emotional stress assessment: functional magnetic resonance imaging (fMRI) and functional near-infrared spectroscopy (fNIRS). These brain imaging techniques allow more precise and objective measurements of these mental states when compared to otherwise used approaches in cognitive load and stress estimation. Within this doctoral thesis, a special interest is given to the computation methods and classification algorithms in estimating cognitive load and emotional stress levels. Experiments based on stimulation paradigms aimed at classifying these constructs are described. Methods for feature extraction, classification models selection, classification results, as well as discussion of results for all conducted experiments are presented. Classification methods in the assessment of cognitive load and emotional stress allow the better-than chance estimation of cognitive load and emotional stress levels. The applications of such neurophysiological measurements enable more objective estimation of various mental states, such as emotional stress and cognitive load. The classification of the abovementioned mental states using various machine learning algorithms may make a significant improvement to the business-as-usual approach in many areas which may benefit from such objective emotional stress and cognitive load assessments. Developed methods and the obtained results are discussed in the context of prior work while considering limitations and proposing directions for future work. Kognitivno opterećenje i emocionalni stres utječu na svakodnevni život ljudi te je adekvatna procjena kognitivnog opterećenja i emocionalnog stresa od posebne važnosti. Neadekvatne količine kognitivnoga opterećenja i emocionalnoga stresa negativno utječu na ljudske performance i, dugoročno gledano, na zdravlje pojedinca. Procjena kognitivnoga opterećenja i emocionalnoga stresa posebno je važna u visoko stresnim profesijama kao što su vojska, piloti i kontrolori zračnog prometa. Iako se kognitivno opterećenje i emocionalni stres mogu procijeniti korištenjem različitih subjektivnih upitnika samoprocjene, njihova pristranost može dovesti do pogrešnih procjena. Za procjenu kognitivnoga opterećenja i emocionalnoga stresa mogu se koristiti objektivne metode temeljene na neurofiziološkim signalima prikupljenima korištenjem funkcionalnih tehnika oslikavanja mozga. U sklopu ove doktorske disertacije korištene su dvije tehnike funkcionalnoga oslikavanja mozga u procjeni kognitivnoga opterećenja i emocionalnoga stresa: funkcionalna magnetska rezonancija (engl. functional magnetic resonance imaging, fMRI) i funkcionalna blisko-infracrvena spektroskopija (engl. functional near-infrared spectroscopy, fNIRS). Ove tehnike oslikavanja mozga omogućuju preciznija i objektivnija mjerenja opisanih mentalnih stanja u usporedbi s inače korištenim pristupima u procjeni kognitivnoga opterećenja i emocionalnoga stresa. U okviru ove doktorske disertacije poseban je interes posvećen računarskim metodama i klasifikacijskim algoritmima u procjeni razina kognitivnoga opterećenja i emocionalnoga stresa. Opisani su eksperimenti utemeljeni na stimulacijskim paradigmama s ciljem klasifikacije ovih konstrukata. Prikazane su metode izlučivanja značajki, odabir modela klasifikacije, rezultati klasifikacije, kao i diskusija o rezultatima za sve provedene eksperimente. Metode klasifikacije u procjeni kognitivnoga opterećenja i emocionalnoga stresa omogućuju procjenu razina kognitivnoga opterećenja i emocionalnoga stresa. Primjena takvih neurofizioloških mjerenja omogućuje objektivniju procjenu različitih psihičkih stanja, poput kognitivnoga opterećenja i emocionalnoga stresa. Klasifikacija navedenih mentalnih stanja korištenjem različitih algoritama strojnog učenja može značajno poboljšati procjenu stanja u mnogim područjima. Razvijene metode i dobiveni rezultati u doktorskoj disertaciji opisuju se u kontekstu dosadašnjeg rada uz razmatranje ograničenja i predlaganje smjerova za budući rad.

Currently, electric automobiles are predominantly powered by interior permanent magnet electric machines which possess high efficiency and torque density in comparison to alternative machine types. Although the performance benefits are undisputed, the use of rare earth permanent magnet materials has historically been a commercial risk. To mitigate the potential issues, the automotive industry is considering alternative electric machine designs, which will either use none or a minimal amount of rare earth material. In parallel, synchronous reluctance machines are "magnet-free" electric machine type currently the focus of process industry. Electric power take-off was selected as the application niche for the introduction of synchronous reluctance machine to the automotive sector. The goals of the thesis are to improve the synchronous reluctance machine optimization process by reducing rotor radial cross-section parametric complexity, and to create a software framework for the rotor geometry feasibility validation. The research concentrates on a novel approach of defining rotor geometry on a shape object level, instead of the classical approach which uses geometrical primitives like points, lines, and arcs. This methodology allows simple and robust feasibility validation and robust calculation of at least one point inside every object, which is extremely important for assigning the material to different regions in the finite element analysis tool. The presented solution can be applied on any electric machine type. Furthermore, the thesis defines a set of absolutely feasible synchronous reluctance machine rotor geometries using a minimal set of parameters in order to reduce design complexity. Consequently causing a reduction of overall optimization time due to the smaller number of parameters that define the optimization problem. After selecting the geometry and commencing optimization, the optimization system (software code) must be able to detect if generated geometry is unfeasible (e.g., rotor barriers are overlapping). One of the proposed solutions is a novel concept of forced feasibility, where every unfeasible design is forced to change parameters until reaching feasibility. Otherwise, the infeasible candidate is rejected, which can affect the optimization convergence. Both approaches were compared concluding that forced feasibly causes faster optimization convergence. Scientific contribution of the thesis is reflected in: 1. Robust feasibility and region detection algorithm based on the shape object approach in the definition of electrical machine geometry. 2. Improvement of optimization convergence through the reduction of feasible geometry search-time for synchronous reluctance machines using novel forced feasibility approach 3. Method for synchronous reluctance machine rotor geometry parametrization with the reduced parameter set. U automobilskoj vuči trenutno prevladavaju strojevi s permanentnim magnetima postavljenima unutar rotora koje u usporedbi s alternativnim tipovima strojeva karakterizira visoka korisnost i gustoća momenta. Premda imaju najbolje tehničke karakteristike, korištenje permanentnih magneta s elementima rijetkih zemalja (eng. rare-earth) kao što su neodimij i disprozij predstavlja značajan problem. Automobilska industrija ne može dozvoliti značajno kolebanje cijene strateških sirovina, što je usmjerilo inženjere na razvoj električnih strojeva bez ili s minimalnom količinom permanentnih magneta s elementima rijetkih zemalja. Potencijalno rješenje koje isključuje korištenje permanentnih magneta je sinkroni reluktantni stroj. Kao niša za primjenu u automobilskoj industriji je određen priključak za pokretanje vanjskih tereta u električnim kamionima (eng. e-PTO - electric power take-off). Istraživanje se koncentrira na optimalnu parametrizaciju rotorskih barijera sinkronih reluktantnih strojeva, unaprjeđivanje postizanja izvedivosti rotorskih geometrija, i razvoj algoritama za baratanje geometrijom stroja temeljenom na objektno opisanim oblicima. Primjena rezultata istraživanja nije ograničena samo na automobilsku industriju jer se stečena znanja i zaključci mogu primijeniti i u energetici te procesnoj industriji. Sukladno navedenim koracima, doprinos provedenog istraživanja opisani kroz doktorsku disertaciju sastoji se od sljedećeg: 1. Robusni algoritmi za ocjenu geometrijske izvedivosti i pronalaženje geometrijskih područja temeljeni na objektnom pristupu definiranja oblika s primjenom u geometrijskom opisu električnog stroja. 2. Poboljšanje konvergencije optimizacije temeljeno na skraćenju vremena pronalaska izvedive geometrije električnog stroja korištenjem metode prisiljene geometrijske izvedivosti. 3. Metoda parametrizacije geometrije rotora sinkronog reluktantnog stroja sa smanjenim skupom parametara.

Utična hibridna električna vozila (engl. Plug-in Hybrid Electric Vehicle, PHEV) predstavljaju jednu od ključnih tranzicijskih tehnologija prema energetski učinkovitom, čistom, tihom i održivom transportu u budućnosti. Budući da su vozni ciklusi vezani uz uvjete vožnje i ponašanje vozača, oni su od ključnog značaja za procjenu potrošnje energije i emisija štetnih plinova, te se naširoko koriste kod projektiranja i parametriranja strukture PHEV pogona i njegove upravljačke strategije. Većina studija vezanih uz PHEV kao i druge tipove vozila temelje se na certifikacijskim voznim ciklusima, koji ne uzimaju u obzir realne uvjete vožnje uključujući nagib ceste, već se temelje na "umjetno" generiranom profilu brzine vozila u ovisnosti o vremenu. Međutim, za potrebe projektiranja i ispitivanja sustava upravljanja vozilom trebali bi se koristiti realistični vozni ciklusi, koji odražavaju stvarno ponašanje u vožnji i uvjete na cesti. U tom smislu, glavni cilj ovog rada usmjeren je na sintezu i validaciju realističnih voznih ciklusa predstavljenih vremenskim profilima brzine i ubrzanja vozila te nagiba ceste, koji su izvedeni iz snimljenih GPS podataka praćenja vozila. U radu se prvo uspostavlja stohastički model voznih ciklusa temeljen na Markovljevim lancima, koji uključuje kombinacije diskretnih vrijednosti brzine vozila, akceleracije vozila, nagiba ceste i vremenske promjene nagiba ceste kao Markovljeva stanja. U tu svrhu prikupljen je bogat skup voznih ciklusa gradskih autobusa u gradu Dubrovniku. Zatim se slučajnim uzorkovanjem višedimenzionalne matrice prijelaznih vjerojatnosti (engl. Transition Probability Matrix, TPM) generira veliki skup sintetičkih voznih ciklusa. Kako bi se smanjio zahtjev za memorijom, a istovremeno poboljšala računalna učinkovitost, predlaže se programska implementacija TPM u obliku rijetke matrice temeljene na rječniku ključeva. K tome, razmatra se postupak uključivanja mase putnika u sintezu voznih ciklusa temeljen na događajima zaustavljanja vozila. Zagušenje prometa uzima se u obzir kroz tvorbu zasebnih TPM za svaku kategoriju zagušenosti prometa, koja se provodi klasteriranjem snimljenih voznih ciklusa s obzirom na uspostavljene kriterije zagušenja prometa. Za potrebe validacije sintetičkih voznih ciklusa predlaže se neuronska mreža koja za dani vozni ciklus predviđa potrošnju goriva. Kao podrška validaciji sintetičkih voznih ciklusa, uzimaju se u obzir brojne statističke značajke iz vremenske i frekvencijske domene, uključujući jedinstvene pokazatelje unakrsne korelacije brzine i ubrzanja vozila te nagiba ceste. Značaj svake nominirane statističke značajke i njezin utjecaj na potrošnju goriva raščlanjuje se pomoću linearne regresijske analize i LASSO (engl. Least Absolute Shrinkage and Selection Operator) metode odabira značajki. Za jednoznačnu, jednokriterijsku validaciju sintetičkih voznih ciklusa u odnosu na skup snimljenih voznih ciklusa predlaže se regresijski model koji uključuje samo nekoliko najrelevantnijih statističkih značajki kao ulaze i potrošnju goriva kao izlaz. Također se kao potencijalno daljnje unaprjeđenje ispituje metoda validacije voznih ciklusa temeljem višestrukih kriterija. Predloženi pristupi validacije temelje se na minimiziranju odstupanja statističkih značajki sintetičkih voznih ciklusa u odnosu na očekivane vrijednosti snimljenih voznih ciklusa. Konačno, u radu se predlažu deterministički i stohastički modeli za predviđanje brzine vozila na horizontu u budućnosti, a s tim povezanim i značajki voznih ciklusa relevantnih za potrebe modelskog prediktivnog upravljanja PHEV pogonom. Prediktivni modeli temelje se na višeslojnoj unaprijednoj neuronskoj mreži, koja se parametrira korištenjem podatka o trenutnoj poziciji vozila, trenutnoj brzini ili povijesti brzina vozila, dobu dana i dana u tjednu, a kao izlaz daje sekvencu brzina vozila na pomičnom vremenskom horizontu. Točnost predviđanja brzine vozila demonstrira se na izdvojenom skupu snimljenih voznih ciklusa. Plug-in Hybrid Electric Vehicles (PHEV) represent a key enabling technology to make the transport system more efficient, cleaner, quieter, and less dependent on oil reserves. Since driving cycles are related to driving conditions and driver behaviour, they are crucial for estimating energy consumption and pollutant emissions, and they are widely used in designing and parameterising the structure of PHEV powertrain and its control strategy. Nowadays, various studies related to PHEVs, as well as other vehicle types, are usually based on certification driving cycles, which do not take into account realistic driving conditions including road slope, but are rather based on “artificially” generated vehicle velocity-time profiles. However, for the purposes of designing and testing a vehicle control system, realistic driving cycles should be used, which reflect the actual driving behavior and road conditions. In this sense, the main aim of this thesis is focused on the synthesis and validation of realistic driving cycles represented by vehicle velocity, acceleration and road slope time profiles, which are derived from recorded GPS vehicle tracking data. The thesis first establishes a stochastic model of driving cycles based on Markov chains, which includes combinations of discrete values of vehicle velocity, vehicle acceleration, road slope and road slope time derivative as Markov states. For this purpose, a rich set of driving cycles of city buses operating in the city of Dubrovnik has been collected. By random sampling of a multidimensional transition probability matrix (TPM), a large set of synthetic driving cycles is generated. To reduce the memory requirement while improving computational efficiency, the implementation of TPM in the form of a sparse matrix based on a dictionary of keys (DOK) is proposed. In addition, including of passengers' mass into driving cycle synthesis process is considered based on vehicle stopping events. Traffic congestion is taken into account through the formation of separate TPMs for each traffic congestion category, which is carried out through clustering of recorded driving cycles based on established traffic congestion criteria. For the purpose of synthetic driving cycles validation, a dedicated neural network (NN) is proposed, which predicts the vehicle fuel consumption for a given driving cycle. The NN approach is based on driving cycle time series input data arranged in a form of fixed‐dimension histogram of counted discrete values of vehicle velocity, acceleration, and road slope. In support of validation of synthetic driving cycles, a rich set of statistical features from the time and frequency domains are considered, including unique indicators of cross-correlation of vehicle velocity, vehicle acceleration, and road slope. The significance of each nominated statistical feature and its impact on fuel consumption is analyzed by using a linear regression analysis and Least Absolute Shrinkage and Selection Operator (LASSO) feature selection method. For unambiguous, single-criteria validation of synthetic driving cycles with respect to recorded driving cycles, a reduced-order LASSO regression model is proposed that takes only a few most significant statistical features as inputs and gives fuel consumption as output. The driving cycle validation method based on multiple criteria is also examined as a potential further improvement. The proposed validation approaches are based on minimizing the deviation of statistical features of synthetic driving cycles from the mean/expected values of recorded driving cycles. Finally, the thesis proposes deterministic and stochastic models for predicting vehicle velocity on receding time horizon, and accordingly driving cycle features relevant for model predictive control of PHEV. Predictive models are based on a feedforward multilayer neural networks, which are trained by using the data on current vehicle position, current vehicle velocity or velocity history, time of day, and weekday, and as an output give sequence of vehicle velocities on the receding time horizon. The accuracy of vehicle velocity prediction is demonstrated on a separate set of recorded driving cycles. The main aim of the thesis is to propose a systematic procedure of multi-dimensional driving cycle synthesis, which utilizes driving data recorded using a vehicle telemetry tracking system and at the end results in a minimum number of representative synthetic driving cycles for a wide range of driving conditions. The thesis is organized in eight chapters, whose content is summarized in what follows. Chapter 1, „Introduction“ outlines the motivation for the conducted research, presents the literature overview, and provides the main hypothesis and an overview of the thesis. Chapter 2, „Recording, pre-processing and analysis of driving cycles“ describes the procedure of driving data recording by utilizing GPS/GPRS vehicle tracking devices for the case of a city bus fleet. Pre-processing of recorded driving data is carried out in terms of extracting driving cycles defined by recorded data between end stations of a certain route, filtering of driving cycles with respect to various criteria, reconstruction of precise road slope profiles using regression method based on Gaussian processes, and categorization of driving cycles in accordance to a traffic congestion index. A comprehensive analysis of interdependence of vehicle velocity, acceleration, road slope, and passengers mass is performed to determine whether the synthesis of driving cycles should be carried out jointly with respect to velocity, acceleration, road slope and passengers mass' states or the road slope and passengers' mass can be modeled independently of velocity and acceleration. Chapter 3, „Synthesis of multidimensional driving cycles“ proposes a method for synthesis of multidimensional driving cycles based on Markov chains. Different ways of realizing the transition probability matrix (TPM) are considered along with related implementation aspects, including computational efficiency analysis. It is proposed to adopt a stochastic driving cycle model represented by an 8D TPM using vehicle velocity, acceleration, road slope, and road slope time derivative states, and based on this model a rich set of synthetic driving cycles is generated. An independent, vehicle stopping event-based synthesis method of passengers’ mass profiles is also proposed. Traffic congestion is considered through the formation of separate TPMs for three levels of congestion (light, medium, and heavy). The chapter concludes with a preliminary analysis of validity of the generated synthetic driving cycles, based on a comparison of distributions of characteristic statistical features between synthetic and recorded driving cycles. Chapter 4, „Regression model for predicting fuel consumption“ establishes a regression model for predicting fuel consumption based on appropriately formed input features of synthetic driving cycles and corresponding neural network (NN). The emphasis is on including a road slope feature when designing the model inputs to improve prediction accuracy. The proposed regression models are trained, validated, and tested on an augmented dataset of combined driving cycles, which are derived through random addition of recorded microcycles whose length corresponds to resolution of fuel consumption measurement. A comparative analysis of fuel consumption prediction accuracy is performed between NN and simpler, linear-in-parameter polynomial regression models, for the case with included and excluded road slope information. In addition, several methods for fine-tuning of NN models are conducted and described in detail, which are related to dimensionality reduction, finding the most appropriate architecture, determining the optimal learning rate, and regularizing the NN. Although the proposed regression model for predicting fuel consumption can be applied to various transport-related studies, in this thesis it is used for determining the fuel consumption related to a set of synthetic driving cycles generated in third chapter for the purpose of their validation. Chapter 5, „Multi-criteria validation of driving cycles“ deals with validation of multidimensional synthetic driving cycles generated in the third chapter. A rich set of driving cycle-related statistical features are nominated to describe different driving patterns, which in addition to general statistical indices includes frequency domain ones and cross-correlation of vehicle speed, vehicle acceleration and road slope features. A comparative analysis of the characteristic statistical features calculated for each recorded and synthetic driving cycle is performed to verify that the distributions of the statistical features for the synthetic driving cycles resembles well those obtained for the recorded driving cycles. For the needs of unambiguous validation of driving cycles, several lumped indicators of driving cycle representativeness are derived, which are based on consolidation of individual statistical features or similarity index between two vectors/matrices used to model driving cycles. The selection of the most appropriate representativeness indicator for the final validation of driving cycles is carried out based on correlation analysis in relation to fuel consumption deviation from the mean value of concatenated recorded driving cycles. Finally, a multi-criteria validation procedure is proposed, which highlights and analyzes several most representative driving cycles from the Pareto frontier of lumped representativeness indicators. Chapter 6, „Feature selection-supported validation of driving cycles“ deals with feature selection techniques based on linear regression analysis and least absolute shrinkage and selection operator (LASSO), which serves to determine the relevance of each nominated statistical feature. The competitiveness of the fuel consumption prediction accuracy of linear and LASSO regression models of reduced order is examined in relation to baseline neural network model established in the fourth chapter. The procedure of extracting representative synthetic driving cycles based on LASSO-predicted fuel consumption is presented and compared with the Euclidean distance lumped indicator-based approach. To further improve the process of unambiguous validation of synthetic driving cycles, a dual-criterion validation method based on both Euclidean distance and LASSO-predicted fuel consumption is examined, as well. Chapter 7, „Prediction of driving cycle features“ first describes the process of data preparation for learning, validation, and testing of a vehicle velocity prediction models. Next, static stochastic and deterministic deep feedforward neural network prediction models are proposed, where the former predicts the vehicle velocity distribution along the route, while the latter predicts the vehicle velocity profile on receding time horizon. A comprehensive analysis of the prediction accuracy of the developed predictive models is performed based on test dataset and considering different forecast quality indicators. Deterministic model is additionally examined in terms of influence of each input variable candidates (current vehicle position, current velocity or history of vehicle velocities, time of day, and weekday), velocity history length, and prediction horizon length on the accuracy of vehicle velocity profile prediction. Finally, a dynamic stochastic model is proposed, which combines the properties of both previously developed predictive models, in terms of predicting expectations and standard deviations of velocity patterns on the receding time horizon. Chapter 8, „Conclusion“ outlines the main results, guidelines for future research, and the following major contributions of the doctoral thesis: 1) Driving cycle model based on Markov chains, which in addition to vehicle velocity and acceleration as the Markov states accounts for the road slope, vehicle mass, and traffic congestion features; 2) Validation of driving cycle model based on recorded fuel consumption data and fuel consumption predictions obtained by using a neural network parameterized based on the driving cycle model features; 3) Stochastic model for predicting driving cycle statistical features meant to be applied in stochastic model predictive control of PHEV.

Disertacija sadrži istraživanje utjecaja procesa smanjenja stakleničkih plinova, posebice s obzirom na utjecaj obveze smanjenja ugljičnoga dioksida kao dominantnog stakleničkog plina, na konkurentnost nezavisnoga proizvođača električne energije. Sadržajno u disertaciji je razvijen model nezavisnoga proizvođača električne energije s ugrađenim sustavom za izdvajanje i skladištenje ugljičnoga dioksida. Unaprijeđena je matematička metoda za ocjenu utjecaja tehnoekonomskih značajki tehnologije smanjenja stakleničkih plinova za slučaj nezavisnoga proizvođača električne energije. Pritom je predočena povezanost između kvadratne krivulje potrošnje energije nezavisnoga proizvođača električne energije, specifičnosti pogonskog goriva te troškova emisijskih obveza u ovisnosti o izboru tehnologije za smanjenje stakleničkih plinova. Na osnovi prikupljenih i obrađenih tehničkih, ekonomskih i ekoloških čimbenika elektroenergetskih sustava jugoistočne Europe, primjenom nelinearnog programiranja, izrađene su simulacije koje su se koristile za analizu i ocjenu utjecaja procesa smanjenja stakleničkih plinova. Nadalje, temeljem rezultata dobivenih kroz proces simulacija te korištenjem unaprijeđene matematičke metode poboljšana je metoda za optimiranje izbora tehnologije smanjenja stakleničkih plinova za slučaj nezavisnoga proizvođača električne energije s obzirom na tržište električne energije i emisijskih obveza. Dokazano je da izbor optimalnog načina ostvarenja emisijskih obveza omogućava nezavisnom proizvođaču električne energije bolje tehničke, ekonomske i ekološke rezultate što na lokalnoj i globalnoj razini ima povoljan utjecaj na cijenu električne energije, konkurentnost industrije, te se umanjuje nepotrebno onečišćenje okoliša. Rezultati testiranja predložene metode opravdavaju predloženi pristup, a uočeni problemi ujedno su i smjernice za budući razvoj i moguća poboljšanja predložene metode. In the global process of reducing greenhouse gases (GHG), independent power producer (IPP) sustainability is directly correlated with its emissions treatment. Carbon dioxide (CO2) is considered to be the main cause of additional greenhouse effect and has become one of the key factors in IPP's sustainability. Because the cost of electricity is included in every product or service, unexplored impact of greenhouse gas reduction process on IPP's performance can cause significant consequences on local and global levels, which are manifested in rising electricity prices, poor investment climate, uncompetitive industry and unnecessary environmental pollution. Different approaches for solving the problem are proposed, but emission obligations still are and will be the subject of intensive research. Also, there are significant differences and inconsistencies, which are manifested in methods, units and assumptions used for estimating carbon dioxide emission reduction costs. To fulfill emission obligations an IPP (located in the EU) can choose between purchasing emission unit allowances (EA) within the European Union Emission Trading Scheme (EU ETS) or installing a carbon capture and storage system CCS [19], [20]. The CCS is expected to play a significant role in slowing down climate change, and it is estimated that it could reduce overall emissions by 20 % until 2050 [21]. Also, it is important to note that the CCS is not a solution for a low-carbon production generation mix, but its role is to gain the time needed to develop new low-carbon power generation technologies. The first part of dissertation presents the research problem, scientific contributions and an overview of all chapters. The second part called "Global Greenhouse Gas Reduction Process" describes the greenhouse effect. The main greenhouse gases have been defined, among which carbon dioxide is the dominant one. The reasons for the process of reducing greenhouse gases have been elaborated. The historical development of the legislation regarding the process of reducing greenhouse gases is presented, and the market of emission allowances in Southeast Europe is described, which is specific to uneven treatment of emission obligations. The third part describes the impact of the GHG reduction process on the IPP, which results in its obligation to purchase emission allowances or to install a CCS system. Carbon capture efficiency is estimated in range from 85 % to 95 % [25]. This means that an IPP with a built-in CCS system located in the European Union is obliged to purchase emission permits, but only for 5 % to 15 % of the carbon dioxide produced. Also, the mechanism of emissions trading in the European Union is described. The main three parts of an IPP equipped with CCS are described: 1. carbon capture (which includes CO2 preparation for transport), 2. transport of carbon dioxide (pipelines or shipping) and 3. storage of carbon dioxide (on land or under the sea). Furthermore, sources and locations of possible carbon storage sites in Southeast Europe are presented. In the fourth part, an IPP model with built-in CCS is defined. This model is based on the energy consumption square curve coefficients (zero, first and second degree) for IPP on coal. Coal was chosen as the fuel, since coal is the "core" fuel of the world's energy, and the process of reducing greenhouse gases is most pronounced at power plants that use coal as their primary energy source. The independent electricity producer model deals with its technical, economical and environmental parameters. The fifth part describes improved mathematical method for assessing techno-economic features of GHG reduction technology on an IPP. Technical, economical and environmental factors of an IPP are clearly and unambiguously connected and regulated, which enables a more accurate evaluation. Also, the advantages of proposed mathematical method for assessing the impact of techno-economic features of greenhouse gas reduction technology are: - a clear correlation between the cost of emission obligations and the efficiency of the power plant, - a clear correlation between the cost of emission obligations and the operating point (operating power) of the power plant and - a clear correlation between the cost of emission obligations and the specifics of the IPP's fuel. In the sixth part, using nonlinear programming, the impact of GHG reduction technology on the case of an independent electricity producer with respect to the electricity market and the emission obligations market was investigated in the Southeast European Power System model that does not have an identical emission management pattern, which may directly affect its sustainability. Based on the analysis of simulation results and using an improved mathematical method, a correlation was determined between the cost of electricity production IPP depending on the choice of greenhouse gas reduction technology with respect to the characteristics of the electricity and emissions market. The proposed method allows more accurate determination of the impact of greenhouse gas reduction technology on the independent electricity producer with regard to electricity market factors and emission obligations. The advantage of the proposed method is a clear idea of the interaction of several technical, economical and environmental factors such as: - IPP utilization factor (depending on the way emission liabilities are treated), - selection of IPP operating point (active power) (which may be between the technical minimum and rated power), - type of IPP propellant (eg coal, gas, lignite, etc.), - quality of delivered fuel (eg coal with different heating power or carbon composition) - total investment costs, - load factor, - installed power of the plant, - economic life of the plant, - interest rate, - fixed maintenance costs, - insurance costs, - administration costs, - staff costs, - variable maintenance costs of the plant (ammonia, SO2, NOx, ash) - fuel price, - method of transport and storage of separated carbon dioxide and - price of emission permits. The seventh part presents the results of the assessment of the techno-economic characteristics of greenhouse gas reduction technology in the case of IPP, in order to analyze its competitiveness on the electricity market. In the eighth chapter, a review of obtained results is given, and possible instructions for further research are presented, given the impact of the greenhouse gas reduction process on the IPP. Annex A presents the impact of neighboring power systems on the power system of Southeast Europe. Estimated data are presented separately for each border electricity system on an hourly basis, taking into account different times of the year. Annex B presents the expected inflow to hydropower plants in Southeast Europe on a weekly basis. Finally, an independent electricity power producer model equipped with a carbon capture and storage system has been developed. The mathematical method for techno-economic characteristics impact assessing of greenhouse gas reduction technology for an independent electricity power producer (IPP) has been improved. For an IPP, connection between energy consumption curve, fuel characteristics and emission obligation costs is presented. Based on the collected and processed technical, economic and environmental Southeast Europe electrical power system factors, using nonlinear programming, simulations were developed in order to analyze and assess the impact of greenhouse gas reduction processes. Furthermore, based on the results obtained through the simulations, method for optimizing the choice of carbon dioxide emission reduction technology has been improved. It has been proven that the optimal way to fulfill emission obligations for an IPP enables better technical, economic and environmental results, which could have a positive effect on electricity prices, industry competitiveness and reduces unnecessary environmental pollution. The results of previous research do not offer a complete solution but represent a current view based on the present development of engineering knowledge, which is constantly supplemented by new technological improvements.

There has been an increasing trend of integrating photovoltaic power plants (PVs). One of the important challenges for distribution system operators is to evaluate the total installed power of a PV that a particular network can host (or PV hosting capacity) while keeping voltage and element constraints within required limits. The major drawback of the existing methods for calculating PV hosting capacity is that they use the same installed power of the PV systems for all simulated PVs, as these methods do not use external data sources about building roofs. As a consequence, this has a significant impact on the final accuracy of the results. This paper presents a probabilistic methodology for calculating the PV hosting capacity in low voltage (LV) networks. The main contribution of this paper is the improved modeling of PV generation using actual building roof data when calculating the PV hosting capacity, as every building is treated according to its actual solar potential. Monte Carlo simulations with incorporated stochastic consumption and PV generation models are utilized for load flow calculations of the actual LV network. The simulation results presented in this paper prove that the proposed methodology increases the accuracy of the final PV hosting capacity calculations.

U ovom radu je opisan postupak estimacije i predviđanja zasjenjenja Sunca analizom slijeda slika neba, koji se temelji na analizi slika snimljenim širokokutnom kamerom. Kamera je instalirana na tlu kako bi se moglo pratiti kretanje Sunca od izlaska do zalaska. Motivacija za ovo istraživanje je dizajniranje mikromreže sa solarnom energijom te različitim vrstama pohrane energije. Opisana metodologija koristi slike neba snimljene svakih pet sekundi s kojima se estimira buduće kretanje oblaka, predviđa se trenutak zasjenjenja Sunca i trajanje sjene oblaka ispred Sunca. Slike neba snimljene širokokutnom kamerom najprije su obrađene na način da se otkloni efekt distorzije širokokutne leće kamere korištenjem ugrađenog modula iz programskog paketa MATLAB, koji se naziva Omni Directional Camera Calibration Toolbox. S obzirom na dinamičku prirodu promjene oblika, oblaci su estimirani oblikom pravokutnika kako bi sam algoritam bio što jednostavniji i brži. Obradom snimljenih slika predviđa se razina Sunčeve dozračenosti na osnovu predviđenog zasjenjenja, tj. predviđene razine vidljivog dijela Sunca te predikcije srednje vrijednosti direktne sunčeve dozračenosti od strane meteorološkog servisa. Dobiveni rezultati predviđanja sunčeve dozračenosti prikazani su u obliku krivulje količine direktne sunčeve dozračenosti u narednom periodu koji se potom koristi u postupku optimizacije tokova energije u mikromreži s fotonaponskim sustavom i sustavima pohrane zasnovanoj na kratkoročnom predviđanju Sunčeve dozračenosti i karakteristikama sustava pohrane energije. Analizirana mikromreža sastoji se i od više tipova pohrane energije s različitom učinkovitošću, kapacitetom i vremenom odziva. This dissertation describes the procedure of estimating and predicting the shading of the Sun by analyzing the sequence of images of the sky, which is based on the analysis of images recorded by a wide-angle camera. The camera is installed on the ground so that it can monitor the movement of the Sun from sunrise to sunset. The motivation for this research is the design of a microgrid with solar energy and different types of energy storage. The described methodology uses images of the sky taken every five seconds to estimate the future movement of clouds and predict the moment of shading of the Sun and the duration of the shadow of clouds in front of the Sun. Sky images taken with a wide-angle camera were first processed to eliminate the distortion effect of the wide-angle camera lens using the built-in module from the MATLAB software package, which is called the Omni Directional Camera Calibration Toolbox. Given the dynamic nature of the shape change, the clouds were estimated to be rectangular in shape to make the algorithm itself as simple and fast as possible. The processing of the recorded images predicts the level of solar radiation based on the predicted shading, i.e., the predicted level of the visible part of the Sun and the prediction of the mean value of direct solar radiation by the meteorological service. The obtained results of solar radiation prediction are presented in the form of a curve of the amount of direct solar radiation in the next period, which is then used in the process of optimizing energy flows in the microgrid with photovoltaic systems and storage systems based on short-term solar radiation prediction and energy storage system characteristics. The analyzed microgrid also consists of several types of energy storage with different efficiency, capacity, and response time.

Tvrtka Končar - Generatori i Motori d. d. ima dugogodišnje iskustvo u proizvodnji električnih rotacionih strojeva. Složeni uvjeti na svjetskom tržištu, prije svega pojačana konkurencija i stalni trend snižavanja cijene, uz povećanje zahtjeva, rezultiraju povećanom složenosti proizvoda i kraćim rokovima isporuke, čime uzrokuju dodatne zahtjeve na razvoj generatora. Električni generator je, i pored brojnih izvedbi, relativno poznat proizvod, s dugim životnim vijekom. Iako se proizvode dugi niz godina, nije poznato da je napravljen cjeloviti model razvoja i konstrukcije generatora. Iako doprinose boljem uvidu u ponašanje i svojstva proizvoda, nove metode proračuna konstrukcije i alati koji omogućuju simulacije fizikalnih stanja u proizvodu istovremeno čine razvojni proces složenijim i zahtjevnijim. Sustavnim pristupom analizi razvojnog procesa generatora, može se prepoznati prostor za poboljšanje razvojnog procesa i same konstrukcije i time doprinijeti naporima koji za cilj imaju razvoj jeftinijeg proizvoda, uz zadržavanje kvalitativnih značajki proizvoda, kao što su - uz temeljne funkcionalne značajke - zadani vijek trajanja, vrijeme isporuke itd. Cjeloviti pristup razvoju modela proizvoda i razvojnog procesa, posebno mehaničkih proračuna, omogućio bi uvid u relacije koje sada nisu vidljive u cijelosti, tj. omogućuje jednostavnije i bolje razumijevanje sustava. Boljim razumijevanjem relacija i interakcija unutar sustava, otvorit će se nove mogućnosti poboljšanja razvojnog procesa i generatora. Končar-Generators and Motors Inc. company has a long-term experience in production of electrical rotating machines. Tough conditions on the world’s market, most of all increased competition and continuous trend in price reduction and increased demands, result in increased complexity of the product thus causing additional demands on the generator development. Electric generator is, despite of its various designs, relatively known product with a long-term life cycle. Although generators are being produced for a long period of time, it is not known whether a thorough developing model and generator design has been made. New calculation methods in the design, as well as new tools, enable simulations of physical conditions of the product and although contribute to better insight into the behaviour and characteristics of the product, at the same time they make developing process more complicated and more demanding. With a systematic approach to the analysis of the generator developing process, additional space for improving the developing process and the design itself can be found, which contribute to the efforts aim, that is development of cheaper product, keeping qualitative characteristics of the product at the same time, such as (besides the basic functional characteristics) required life cycle, delivery time etc. Comprehensive approach to the development of product model and developing process, especially mechanical calculations, would enable insight into relations which are now not entirely visible, i.e., enables simpler and better understanding of the system. By better understanding of the relations and interactions within the system, new possibilities of improving the developing process and generator itself will appear.

Ovaj rad sastoji se od pet poglavlja u kojima ću kroz svoja iskustva i stečena znanja predstaviti 3 godine aktivnog istraživanja i rada na području FPV-a. Dronovi i bespilotne letjelice su danas nezamjenjivi alati pri izradi vizualnih materijala; slike, videa, u kartografiji. Zamijenili su gotovo u potpunosti snimanje iz zraka klasičnim metodama helikopterima i avionima, ali njihova upotreba je daleko šira nego što ću obrađivati u ovom radu. Fokusirati ću se na: povijest FPV-a, anatomiju mini multirotora/dronova, softver, regulativu i zakonski okvir, proces akvizicije snimaka, workflow obrade snimaka te zadnji dio ovog rada biti će posvećen procesu kreacije 360° VR FPV iskustva. S obzirom da je FPV, bilo kao profesionalan rad, bilo kao hobi veoma zahtjevan, mislim da je odličan pokazatelj kao prvo upornosti, a kao drugo širine potrebnog znanja ne samo na području multimedije nego i elektrotehnike, programiranja čak i fizike. Kako bi uopće došli do snimanja 360° VR videa pomoću FPV drona potrebno je veliko znanje osnova FPV kvadkoptera. Što je FPV dron specifičniji za neku upotrebu, to ga je teže izraditi i koristiti. Kako je FPV veoma mlado područje i još uvijek u značajnom razvoju, specifični pojmovi koji se koriste u FPV svijetu te tako i u ovom radu su riječi engleskog jezika, s obzirom da još uvijek nemamo te pojmove odgovarajuće prevedene na hrvatski jezik.

Povećanje energetske učinkovitosti elektroenergetskog sektora je opredjeljenje svih nacionalnih prijenosnih operatera. Jedan od temeljnih ciljeva je smanjenje gubitka u elektroenergetskim mrežama. Električne mreže su jedan od najvažnijih dijelova elektroenergetskog sustava. Pouzdanost isporuke te kvaliteta električne energije neposredno ovisi o vrsti i izgrađenosti elektroenergetske mreže. Poznato je da se gubici u prijenosnoj mreži mogu definirati kao razlike preuzete i predane električne energije. Određivanje gubitaka na vodovima ima nekoliko temeljnih zadaća. Prva zadaća jeste odrediti razliku između izračunatih gubitaka i stvarnih gubitaka te odrediti razlog za tu razliku i da pokaže neophodne smjernice za poduzimanje mjera za smanjenje gubitaka. Drugi zadaća je procjena indeksa gubitaka za planiranje i formulaciju različitih planova. U ovom rada je prikazan novi pristup za smanjenje pogreške između izračunatih i stvarnih gubitaka korištenjem nove metode koja se bazira na tri modificirana moda, odnosno sasvim novim pristupom za izračun indeksa gubitaka. U ovom doktorskom radu predložena metoda znatno smanjuje pogrešku u odnosu na klasičnu metodu na tri moda. Također, razvijen je algoritam koji opisuje korake za implementaciju metode. Increasing the energy efficiency of the electricity sector is the commitment of all national transmission operators. One of the main goals is to reduce losses in electrical networks. Electrical networks are one of the most important parts of the electric grid. Delivery reliability and the quality of electricity directly depend on the electrical networks. It is well known that losses in the transmission network can be defined as the difference between the electricity taken and delivered. Determining the losses in overhead power lines has several basic tasks. One task is to determine the difference between calculated losses and actual losses, to determine the reason for the difference between calculated and actual losses, and to present the necessary data for taking measures to reduce losses. The second task is an estimation of the loss index for planning and formulation of various plans. The purpose of this paper is to present a new approach to reduce the difference between calculated and actual losses by modifying the method in three modes, or a new approach to calculating the loss index. The proposed method in this doctoral thesis significantly reduces the error in relation to the classic three-mode method. Also, an algorithm has been developed that describes the steps for implementing the method.

Cilj doktorske disertacije bio je predloziti metode za poboljsanje rada estimatora stanja i parametara asinkronog stroja. Disertacija je podjeljena u dva glavna dijela, prvi dio vezan za estimacije parametara prije pogona asinkronog stroja i drugi vezan za procjene stanja i parametara za vrijeme pogona asinkronog stroja. Prije ulaska u problematiku unaprijedenja rada estimatora, u disertaciji se strukturira potrebne matematicke modele asinkronog stroja i osnovne teorijske postavke potrebne za analizu dimanike, stabilnosti i konvergencije estimatora stanja i parametara, kao i dinamickih sustava. Prvi dio doktorata predstavlja metodu za odabir vremena uzorkovanja za tranzijente asinkronog stroja. Prethodno se pretpostavlja da se ne poznaje nikakve informacije o stroju. Metoda je zasnovana na aproksimaciji mjerenog tranzijenta direktnog uklopa asinkronog stroja pomocu reda kompleksnih eksponencijala i procjeni polova asinkronog stroja. Temelj kvalitetne aproksimacije je segmentacija prikupljenih podataka kako bi nelinearni sustav aproksimirali linearnim, za sto se uvodi optimizacijska procedura zasnovana na metaheuristi cnim gloalnom optimizatoru MIDACO. Na temelju procjenjenih polova modela, numerickim rjesavanjem nelinearne kriterijske jednadzbe dobiva se potrebno vrijeme uzorkovanja koje zadovoljava kriterij primjenjivosti Tustinove bilinearne transformacije na model asinkronog stroja. Odabrano vrijeme uzorkovanja usporedeno je sa Nyquistovim kriterijem i provjereno simulacijskim i eksperimentalnim primjerima. Zakljucuje se da su iskutveno odabrane frekvencije uzorkovanja od 5-15 kHz i do deset puta prebrza u odnosu na stvarno potrebno vrijeme uzorkovanja s kojim se moze pristupiti procjeni parametara asinkronog stroja. Drugi dio doktorata pociva na sintezi adaptivnog sustava za procjenu stanja i parametara asinkronog stroja za vrijeme pogona. Ovakav sustav je nuzan za ostvarivanje ispravne 164 5. Sazetak povratne veze za vektorsko upravljanje. Temeljna zadaca adaptivnog sustava je procjena varijabli stanja asinkronog stroja, stuje statora i magnetskog toka rotora asinkronog stroja kako bi se provela dobra orijentacija mt koordinatnog sustava prema polozaju prostornog vektora rotora. Uz ove zadace dodatno se procjenjuje brzina vrtnje asinkronog stroja koja nije mjerena pomocu mjernih clanova, a prati se i promjena otpora statora i vremenske konstante rotora kao vaznih parametara cije varijacije s temperaturom mogu narusiti rad reguliranog asinkronog stroja. Adaptivni sustav sintetiziran je na temelju pseudolinearne regresije Volterra transformiranog modela asinkronog stroja. Sustav je stabilan, a konvergencija parametara se postize do na preciznost primijenjene linearizacije, gdje je odstupanje brzine vrtnje minimalno. Sustav se dodatno prosiruje sa ltracijom signala struje koji je pojasnopropusne karakteristike. Filtar je zasnovan na ARMA modelu vremenskih nizova, adaptivan i samopodesavajuci. Rad ltra ispitan je na eksperimentalnim podacima, dok je rad cijelog adaptivnog procjenitelja stanja i parametara ispitan simulacijski. Pokazuje se da uvodenje ltra u povratnu vezu dodatno poboljsava konvergenciju parametara asinkronog stroja. The aim of this dissertation is to suggest new methods for improving induction machine state and parameter estimation. The dissertation is divided into two main sections. The rst part is dedicated to improving parameter estimation methods for oine estimation, and the second part is dedicated to state and parameter estimator synthesis for online estimation. Prior to explanation of the estimator improvements presented by the text, the dynamic modelling of an induction machine is structured and the fundamental concepts of dynamic system analysis, stability and convergence for state and parameter estimation is presented. The rst part of this thesis introduces a method for selecting the sampling time for induction machine transients. It is assumed that no information about the machine is known in advance. The method is based on complex exponential series approximation of the measured line start transient and lieanrized system pole estimation from the mentioned series. To achieve good approximation of nonlinear system with linear approximation, it is neccessary to segment the acquired data. The optimization method based on MIDACO metaheuristic global optimizer is applied for optimal data segmentation. From the determined system poles, the sampling time that satises the criterion of Tustin bilinear transformation accuracy is computed by numerical solution to the nonlinear criterion equation. The selected sampling time is compared to the Nyquist sampling time. It is concluded that the commonly selected 5-15 kHz sampling frequencies are up to ten times more then the actually needed samping time for oine parameter estimation. The second part of this thesis deals with synthesis of an adaptive system for simultaneous state and parameter estimation of induction machine during operation. This type of system is neccessary to achieve a good feedback needed for vector control of the induction machine. The main task of the adaptive system is state estimation. The estimated states are stator 166 5. Abstract currents and rotor ux linkages, wehere the estimation accuracy is demanded for porper rotor eld orientation of the mt coordinate system. Alongside the state estimation task, the speed of the rotation not measured by physical sensors is also estimated. The stator resistance and rotor time constant, important parameters that can vary with temperature and thus in uence the operating point of the machine, are also tracked in real time. The adaptive system is synthetised based on pseudolinear regresion of the Volterra based induction machine model. The system is stable and parametric convergence is agcieved up to a linearization error, introducing minimal speed estimation error. The system is additionaly expanded with bandpass current signal ltration. The lter is based on ARMA timeseries model, adaptive and selftuning. The lter operation is tested on experimental data while the entire state and estimation system is tested in simulations. It is shown that lter introduction in feedback further improves the parametric convergence of the system.

Cilj je ovog rada razvoj višekorisnickog energetski ucinkovitog predajnog sustava kako bi se postigao bežicni prijenos analogne informacije na kratke udaljenosti koristeci pritom ultraširokopojasne radio impulse kao nositelje informacije. Predajni sustav cine modulator temeljen na sklopovima za vremensko kodiranje informacije i generator ultra-širokopojasnih radio impulsa. Sklop za vremensko kodiranje informacije izveden je integralnim impulsno frekvencijskim modulatorom (IPFM) kojim se moduliraju analogne velicine (napon, struja) u niz unipolarnih impulsa na nacin da je vremenski interval izmeu susjednih impulsa funkcija analognog, modulacijskog signala. Svojstvo višekorisnickog pristupa postiže se kodiranjem svakog korisnika koji šalje informacije u istom komunikacijskom kanalu jedinstvenom celijom za kašnjenje, cija vrijednost kašnjenja IPFM signala kodira korisnika koji šalje informaciju. Na navedeni nacin, modulator generira niz impulsa ciji meusobni razmak kodira informaciju dok je u širini impulsa kodiran korisnik koji šalje navedenu informaciju. Izostankom generatora takta modulatora i mikroupravljaca postiže se njegova energetska ucinkovitost. Impulsni niz s izlaza modulatora dovodi se na generator ultra-širokopojasnih radio impulsa koji generira ultra-širokopojasni impuls za svaki ulazni i silazni brid navedenog impulsnog niza. S obzirom na svojstvo kratkog intervala trajanja ultra-širokopojasnog impulsa, istima se postiže svojstvo malog iznosa energije po impulsu. S ciljem miniturizacije sklopa kao i postizanja vece energetske ucinkovitosti predajnog sustava, isti je implementiran u CMOS integrirani krug. U sklopu rada ukupno su razvijena dva generatora ultra-širokopojasnih impulsa. Prvi se temelji na oblikovanju bridova pravokutnog signala, dok se drugi temelji na križno prospojenom LC oscilatoru s ugraenim sklopom za upravljanje kojim se automatski kontrolira njegovo uklapanje i isklapanje s obzirom na struju oscilatora, temperaturu okoline integriranog kruga kao i utjecaj kvantnog šuma i šuma sacme na pocetak oscilacija. Kako bi se izmjerio utjecaj višekorisnickog pristupa na pogrešku u prijemu simbola, IPFM modulator i demodulator implementirani su u FPGA, gdje se emuliranjem primopredajnog sustava promatra utjecaj parametara modulacije na omjer pogrešno zaprimljenih simbola. Temeljem dobivenih rezultata, provedena su mjerenja nad ukupno cetirima IPFM predajnicima koji asinkrono i kontinuirano šalju informaciju unutar istog komunikacijskog kanala. Za potrebe istog, u CMOS tehnologiji izraen je detektor ultra-širokopojasnih impulsa kojim se iste pretvara u unipolarne s ciljem njihove jednobitovne digitalizacije, dekodiranja i, naposljetku demodulacije signala od navedenog korisnika. Rad je rezultirao s ukupno cetirima znanstvenim doprinosima a to su: 108 1. Izrada višekorisnickog ultra-širokopojasnog predajnika temeljenog na integralnom impulsno frekvencijskom modulatoru i kodiranju korisnika jedinstvenim celijama za kašnjenje u tehnologiji CMOS. 2. Vrednovanje znacajki impulsno integralnog frekvencijskog modulatora u tehnologiji CMOS. 3. Izrada sustava za procjenu ucestalosti pogreške simbola dekodiranih impulsnih nizova odaslanih s višestrukih višekorisnickih integralnih impulsno frekvencijskih modulatora temeljenih na kodiranju korisnika jedinstvenim celijama za kašnjenje. 4. Izrada ultra-širokopojasnog generatora impulsa s promjenjivom spektralnom gustocom snage temeljenog na LC oscilatoru u tehnologiji CMOS. The aim of this thesis is to develop a multi-user energy efficient transmission system in order to achieve wireless transmission of analog data over short distances using ultra-wideband radio pulses as data carriers. The transmission system consists of a modulator based on circuits with time data coding and an ultra-wideband radio pulse generator. The circuits with time data coding are made of an Integral Impulse Frequency Modulator (IPFM) which modulates an analog quantity (voltage, current) into a series of unipolar pulses in such a way that the time interval between adjacent pulses is a function of the analog modulation signal. The multiuser approach feature is achieved by encoding each user who sends data in the same communication channel with a unique delay cell whose IPFM signal delay value encodes the user sending the data. In this way, the modulator generates a series of pulses whose spacing encodes the data while the pulse width encodes the user sending the aforementioned data. The absence of a clock modulator generator and a microcontroller achieves its energy efficiency. A pulse array from the output of the modulator is fed to an ultra-wideband radio pulse generator that generates an ultra-wideband pulse for each input and output edge of the pulse array. Given the short duration of the interval of the ultra-wideband pulse, they achieve the property of a small amount of energy per pulse. With the aim of miniaturizing the circuit as well as achieving greater energy efficiency of the transmission system, it has been implemented in the CMOS integrated circuit. As part of the paper, a total of two ultra-wideband pulse generators were developed. The first was based on the design of the edges of a rectangular signal, while the second was based on a cross-connected LC oscillator with a built-in control that automatically controls the on and off switch with respect to the oscillator current, integrated circuit ambient temperature, and the influence of quantum noise and shot noise to the beginning of the oscillation. To measure the impact of multi-user access on symbol reception error, the IPFM modulator and demodulator were implemented in the FPGA, where the effect of modulation parameters on the ratio of incorrectly received symbols was observed by emulating the transceiver system. Based on the obtained results, measurements were performed on a total of four IPFM transmitters that sent data asynchronously and continuously within the same communication channel. For the same requirement in the CMOS technology, a detector of ultra-wideband pulses was made, which converted these impulses into unipolar ones with the aim of their one-bit digitization, decoding, and, finally, demodulation of the signal from the specified user. 110 The paper resulted in a total of four scientific contributions: 1. Development of a multi-user ultra-wideband transmitter based on an integral pulse frequency modulator and user coding with unique delay cells in CMOS technology. 2. Evaluation of pulse frequency modulator features in CMOS technology. 3. Development of a system for estimating the error rate of decoded pulse array symbols transmitted from multiple multi-user integral pulse frequency modulators based on coding users with unique delay cells. 4. Development of an ultra-wideband pulse generator with variable power spectral density based on an LC oscillator in CMOS technology.

Predviđanje sportskih procesa, vršeno na temelju iskustva ili znanja o određenom procesu i korištenjem informacija o događaju, zanimljivo je široj javnosti u vidu sportskog klađenja, a dostupnošću velikih količina podataka sve češće postaje i tema znanstvenih istraživanja. Osim uobičajenih statističkih metoda za analizu podataka, koristi se i strojno učenje kako bi se ostvarili što bolji rezultati predviđanja sportskih ishoda. Rezultati znanstvenih istraživanja o sportskim događajima posebno su zanimljivi ekspertima, trenerima, sportskim menadžerima i upravama sportskih klubova koji ih koriste u svrhu vrednovanja učinka igrača i momčadi, kod odabira igrača, identifikacije sportskih talenata, definiranja novih strategija, itd. U disertaciji je tema istraživanja primjena metoda strojnog učenja u predviđanju sportskih ishoda. Osim pruženog detaljnog uvida u dostupnu literaturu i trenutna postignuća u području, napravljena je i analiza znanstvenih radova i ostvarenih rezultata istraživanja te analiza dostupnih podataka o sportskim događajima kako bi se identificiralo značajke od interesa za izradu modela predviđanja sportskih ishoda. U radu su opisani postojeći indeksi korisnosti te je predložen sveobuhvatni indeks korisnosti prilagodljiv različitim sportovima koji predstavlja temelj predložene metode. Ispitana je hipoteza da za učinkovit model nije potrebno poznavanje cijele povijesti, već je dovoljno pronaći reprezentativni dio povijesti koji se u ovom slučaju naziva optimalnim vremenskim prozorom. Također, zaključeno je kako je događaje moguće klasificirati u različite kategorije kako bi se primjenom prilagodljive metode predviđanja dobili još bolji rezultati. U radu je predstavljen sveobuhvatni indeks korisnosti kojim se može, ovisno o konkretnom problemu, vrednovati učinak igrača ili momčadi, a koji se u kasnijim fazama koristi kao početna točka predviđanja ishoda. Predložen je i postupak optimiranja parametara sveobuhvatnog indeksa korisnosti korištenjem kombinacije linearnih i nelinearnih doprinosa, a predstavljen je i algoritam izračuna i prilagodbe optimalnog vremenskog prozora sa svrhom ograničavanja doprinosa događaja iz daleke prošlosti. Sam postupak optimiranja je iterativan, a osim pronalaska optimalnog doprinosa, uključuje i postupak definiranja redoslijeda optimizacije skupa korištenih značajki. Optimalni vremenski prozor predstavlja kontinuirani vremenski period koji se koristi u svrhu određivanja relevantnosti statističkih podataka o prethodnim događajima s ciljem pronalaska podskupa skupa za učenje koji najbolje opisuje trenutno stanje analiziranog procesa, a da pritom ne doprinosi posljedičnom smanjenju rezultata predviđanja. U radu je predložen način izračuna optimalnog vremenskog prozora na temelju prosječnog učinka ili prosječnog indeksa korisnosti. Proces izračuna i prilagodbe optimalnog vremenskog prozora uključuje dva koraka. Jedan je 131 izračun početnog vremenskog prozora, a drugi je prilagodba vremenskog prozora na eventualne promjene. U svrhu poboljšanja rezultata predviđanja predložen je i način identifikacije događaja povećane neizvjesnosti. Događaji povećane neizvjesnosti omogućuju primjenu prilagodljivog postupka u vidu određivanja razine složenosti i načina predviđanja s konačnim ciljem poboljšanja rezultata predviđanja. Predloženi model ispitan je korištenjem skupa podataka o utakmicama NBA lige. Although the sporting process prediction, which is based on the experience or knowledge of a particular process and the use of information about the event, is interesting to the general public mainly in the form of sports betting, the availability of large data amounts is increasingly becoming a topic of scientific research. Currently, in addition to the application of common statistical methods to analyze the available data, machine learning is used to achieve the best possible results in predicting sports outcomes. Scientific research results on sporting events are of particular interest to experts, coaches, sports managers and the management of sports clubs who use them to evaluate the players’ and team’s performance, to select players, to identify sporting talents, to define new strategies, etc. The research topic of the dissertation pertains to the application of machine learning methods in predicting sports outcomes. In addition to providing a detailed insight into the available literature and the current achievements in the field, an analysis of scientific papers and research results was carried out, together with an analysis of the available data on sporting events, all in order to identify the features of interest for the development of models for predicting sports outcomes. The dissertation describes the existing efficiency indexes, specific to individual sports, which are used to assess the players’ and team’s performance; and it proposes a comprehensive efficiency index which can be used to assess performance in different sports. The proposed comprehensive efficiency index is the basis of the proposed predicting method. The hypothesis that was tested states that an effective model does not require the knowledge of the entire history, but that it is enough to find a representative part of history and use it to make predictions with satisfactory accuracy, which is achieved by applying the optimal time window. In the proposed method, the time window, in addition to being determined before the beginning of the application, is adjusted on the basis of the data present during the application itself, i.e. during the testing of the model. The dissertation presents a comprehensive efficiency index which can, depending on the specific problem, evaluate the players’ or team’s performance, and which is to be used in the later stages as a starting point for predicting the outcome. A procedure of optimizing the parameters of the comprehensive efficiency index by using a combination of linear and nonlinear contributions is also proposed, and an algorithm for calculating and adjusting the optimal time window is presented in order to limit the event contribution from the distant past. The goal of the optimization process 133 is to reduce the dimensionality problem by identifying irrelevant and redundant features for the purpose of faster and more efficient execution of the prediction algorithm. The optimization process is iterative, and in addition to finding the optimal contribution, it also includes the process of defining the order of the optimization of the used set of features. The optimal time window is a continuous period used to determine the relevance of the statistical data of previous events. The goal of the optimal time window is to find a training subset that best describes the current state of the analyzed process, without consequently reducing the prediction results. The dissertation proposes a method of calculating the optimal time window based on the average process performance or average process efficiency index. The process of calculating and adjusting the optimal time window involves two steps – one is to define the initial time window, and the other is to adjust the time window in case of any changes. The proposed model was evaluated by using the NBA league game dataset.