Search

Prosjektet studerer kontroll av kraftelektronikkomformere for enkel batteriintegrasjon i AC-distribuerte fremdriftssystemer i skip ved hjelp av kontrollstrukturen kjent som "virtuell" synkronmaskin (VSM). Hovedformålet med en VSM er å etterligne oppførselen til en tradisjonell synkronmaskin (SM) i en kraftelektronikkomformer og dermed arve viktige elektromekaniske egenskaper som bidrar til nettstabilitet. Et VSM-basert energilagringssystem (ESS) er forventet å kunne forbedre nettstabilitet og øke energieffektiviteten i AC-distribuerte fremdriftssystemer i skip ved å gi en utslippsfri roterende reserve, muliggjøre sømløse overganger mellom nettfølgende og nettdannende drift av kraftelektronikkomformere. For å kunne verifisere effekten av å installere en VSM-basert ESS med dobbelinnmating i et to-buss fremdriftssystem har prosjektet utviklet en simuleringsmodell i tidsdomene. Det finnes mange forskjellige implementasjoner av VSMer, men dette prosjektet har valgt en løsning basert på en kvasi stasjonær elektrisk modell (QSEM), som utelater de dynamiske egenskaper av den elektriske SM modellen. En kontrollstruktur basert på en QSEM vil redusere kompleksiteten til VSMen, men samtidig opprettholde viktige elektromekaniske egenskaper til SMen. Simuleringsresultater har vist at den umiddelbare energitilførselen som batterier representerer kombinert med en VSM-kontrollerte kraftelektronikkomformere har økt nettstabiliteten ved å tilføre et "virtuelt" treghetsmoment samt en hurtig frekvenskontroll. Videre viser simuleringsmodellen at en VSM-basert ESS reduserer tidligere begrensninger knyttet til batteriintegrasjon i AC-distribuerte kraftsystemer og muliggjør sømløse overganger mellom lastdeling, nettstøtte og øydrift av ESSen. En ESS med dobbelinnmating gir muligheten til å drifte systemet med en "virtuelt" lukket buss som tillater energitransportering mellom AC-bussene i driftssituasjoner som krever separerte busser. Et videre arbeid med et energistyringssystem (EMS) som utnytter fleksibiliteten til en dobbelinnmatet VSM-basert ESS kan potensielt redusere driftskostnader og klimagassutslipp i skip. This project studies the control of power converters for simple battery integration in AC-distributed ship power systems utilizing the Virtual Synchronous Machine (VSM) control structure. The primary purpose of the VSM is to emulate the behaviors of a traditional Synchronous Machine (SM) in power converters and inherit important electromechanical properties contributing to grid stability. A VSM-based Energy Storage System (ESS) can potentially improve stability and efficiency in AC-distributed ship power systems by providing an emission-free spinning reserve and enabling seamless transitions between the grid-following and grid-forming operation of the power converter. However, to assess this, the project has developed a time-domain simulation model of a two-bus ship power system used to verify the effects of adding a dual-infeed VSM-based ESS in a two-bus ship power system. The VSM-controlled converter uses a reduced-order VSM architecture based on the Quasi-Stationary Electrical Model (QSEM), omitting the dynamical properties of the electrical model to reduce complexity while maintaining essential properties. Simulation results have shown that the immediate energy supply of batteries combined with the VSM-controlled converter design has increased grid stability by adding "virtual" inertia and providing a fast primary frequency control. Furthermore, the simulation model shows that the VSM-based ESS reduces previous limitations related to battery integration in AC-distributed power systems and enables seamless transitions between load sharing, grid support, and islanded operation of the ESS. The dual-in-feed configuration using a common DC bus also provides "virtual" closed bus ties and allows power cycling between each AC bus during operations with open bus ties. Further work on an Energy Management System (EMS) utilizing the increased flexibility of a dual-infeed configured VSM-based ESS can potentially reduce operational costs and GHG emissions in ships by optimizing energy utilization.

Forskningen på ‘fixed wing’ vertikale start- og landingsdroner (VTOL) vokser raskt, med økende fokus på smidige kontrollsystemer og kostnadsreduserende tiltak. Denne oppgaven har som mål å bidra til denne forskningen ved å presentere ulike sensorløse kontrollmetoder som gir muligheten til å låse propellene til en ‘fixed wing’ drone for å forhindre at de blir skadet under start og landinger og for å oppnå bedre aerodynamiske egenskaper under flyging for å redusere energiforbruket. Med sikte på å utvikle et sensorløst kontrollsystem for å låse drone propellene i en fast posisjon, ble det laget et simuleringsmiljø ved bruk av MATLABs Simulink hvor en permanent magnet synkronmotor designet av Alva Industries kunne simuleres. Ved å bruke et styringssystem som først ble utviklet i fordypningsprosjektet, ble ulike estimeringsmetoder implementert i simuleringsmiljøet for å få det sensorløse kontrollsystemet til å oppfylle kravene definert i oppgaven. De implementerte estimeringsmetodene var basert på ikke-lineære observatører, høyfrekvente strøminjeksjoner, stator monterte Hall-sensorer og Kalman-filtre som ble brukt som feedback til kontrollsystemet i et forsøk på å erstatte behovet for enkodere og på den måten bidra til reduksjon av systemkostnader. Etter simulering og optimalisering av kontrollsystemet ved bruk av ulike estimeringsmetoder, ble det funnet at kun to av dem ga tilfredsstillende resultater. Dette var de som var basert på de ikke-lineære observatørene og det utvidede Kalman-filteret kombinert med Hall-sensorer. Ved å bruke de ikke-lineære observatørene som feedback, som yter dårlig ved lave hastigheter, ble propellen magnetisk låst i posisjon ved å påføre en fast spenning til statorviklingene. Denne metoden har sine begrensninger da den er basert på åpen sløyfekontroll og kan derfor ikke garantere at propellen er låst i ønsket posisjon. På den andre siden, er metoden som involverer det utvidede Kalman-filteret kombinert med Hall-sensorer basert på lukket sløyfekontroll og kan i større grad sikre at propellen låses i ønsket posisjon. I tillegg ble en typisk metode for å estimere rotorposisjonen ved lave hastigheter ved bruk av høyfrekvente strøminjeksjoner funnet uegnet for motoren designet av Alva på grunn av små motorinduktanser. Avslutningsvis ble det funnet to ulike metoder som kan låse propellen i fast posisjon uten bruk av enkoder. Likevel kan bare én av dem garantere at propellen er låst og forblir låst i riktig posisjon. For videre forskning foreslås det å verifisere de simulerte resultatene praktisk, og gjennomføre en energianalyse av de ulike kontrollmetodene. The research on fixed-wing vertical take-off and landing drones (VTOL) is growing rapidly, with an increasing focus on agile control systems and cost-reducing measures. This thesis aims to contribute to this research by presenting various sensorless control schemes that makes it possible to lock the propellers of a fixed-wing drone in order to prevent them from being damaged during takeoff and landings and to achieve better aerodynamic properties during flight to reduce energy expenditure. With the aim of developing a sensorless control system to lock the drone propellers into a fixed position, a simulation environment was created using MATLAB's Simulink where a permanent magnet synchronous motor designed by Alva Industries could be simulated. By using a control system that was first developed in the specialization project, various estimation methods were implemented in the simulation environment to make the sensorless control system meet the requirements defined in the thesis. The implemented estimation methods were based on nonlinear observers, high frequency current injections, stator mounted Hall sensors and Kalman filters which was used as feedback to the control system in an effort to replace the need of encoders and in that way contribute to reduction of system costs. After simulating and optimizing the control system using different estimation methods, it was found that only two of them gave satisfactory results. These were the ones based on the nonlinear observers and the Extended Kalman filter coupled with Hall sensors. By using the nonlinear observers as feedback, which performs poorly at low speeds, the propeller was magnetically locked into position by applying a fixed voltage to the stator windings. This method has its limitations as it is based on open loop control and therefore cannot guarantee that the propeller is locked into the desired position. On the other hand, the method involving the Extended Kalman filter coupled with Hall sensors is based on closed loop control and can to a greater extent ensure that the propeller is locked into the desired position. In addition, a commonly used method for estimating the rotor position at low speeds using high frequency current injections was found unsuitable for the motor designed by Alva due to small motor inductances. In conclusion, two different methods were found that can lock the propeller into a fixed position without the use of an encoder. Nevertheless, only one of them can guarantee that the propeller is locked and remains locked in the correct position. For further research, it is proposed to verify the simulated results practically and conduct an energy analysis of the different control methods.

De siste årene har etterspørselen etter posisjoneringsløsninger for propelldrift i droner økt i industrien, med to velkjente motorapplikasjoner, fremdriftsmotorer og vertikal take-off og landings motorer. Hensikten med posisjoneringen i fremdriftsmotorer for fastvingede ubemannede luftfartøyer er å sikre en trygg motorposisjon ved landingsøyeblikket. Siden luftfartøyene er designet med et lavt landingsutstyr for å redusere luftmotstanden og øke varigheten, krever det at propellen plasseres i en horisontal posisjon for å unngå mekanisk skade på propellbladene hvis den treffer bakken. For vertikale take-off og landingsmotorer er posisjonering nødvendig for å redusere luftmotstanden produsert av propellene når de ikke er parallell med kursvinkelen. Hvis mer av overflatearealet til propellbladene er eksponert i lengderetningen av luftfartøyet, bidrar det til mer luftmotstand som igjen bidrar til redusert varighet for dronen under flygningen. Denne oppgaven presenterer fire tilstandsmaskiner for posisjonskontroll og posisjonslåsing av en overflatemontert permanentmagnet synkron motor for propelldrift. Hvor posisjoneringsmetodene er enten med en proporsjonal-integral-derivert kontroller og en enkoder som tilbakekobling for posisjonskontroll, eller med en hall-effekt sensor som tilbakekobling og med en konstant strøm i motorfasene for posisjonslåsing. I tillegg anvender en av posisjonslåsemetodene et posisjonsestimat som tilbakekobling med en implementert hall-effekt sensorassistert ulineær observatør for å muliggjøre posisjonslåsing i mer enn én fast mekanisk posisjon. De eksperimentelle resultatene viser hvordan de foreslåtte løsningene kan gi tilstrekkelig ytelse både med og uten enkoder, og er utført med en vindforstyrrelse for å kunne vurdere ytelsen nært et reelt scenario, og sammenlignet opp mot de to nevnte motorapplikasjonene. In the recent years the demand for positioning solutions for propeller drives in drones has been increasing in the industry, with two prominent motor applications, propulsion motors and vertical take-off and landing motors. The purpose of the positioning in propulsion motor applications for fixed-wing unmanned aerial vehicles is to ensure a fixed motor position at the landing instant. As the aircraft is designed with low landing gear to reduce the drag and increase endurance, requires the propeller to be positioned in a horizontal position to avoid mechanical damage when hitting the ground. For the vertical take-off and landing motor applications, positioning is needed to reduce the additional drag from the propellers when not aligned with the heading angle. As more of the surface area of the propeller blades are exposed in the longitudinal direction the more drag is contributed, which reduces the endurance of the drone during cruise. This paper presents four state machines for position control and position locking of a surface-mounted permanent magnet synchronous motor propeller drive. Where the positioning methods are either with a proportional-integral-derivative controller and an encoder as feedback for position control, or with a hall-effect sensor as feedback and with a constant current to the motor phases for position locking. In addition, one of the position locking methods utilizes a position estimation as feedback with an implemented hall-effect sensor-assisted nonlinear observer solution to allow for position locking in more than one fixed mechanical position. The experimental results show how the proposed solutions can provide sufficient performance both with and without an encoder, and are conducted with wind disturbance to determine the performance close to a real scenario, and compared in relation to the two motor applications.

Litium-ion-batterier representerer en enorm industri og har vist seg å være en sentral teknologi for å redusere klimagasser. Ettersom elektrifiseringen av transportsektoren fortsetter å øke, vil de potensielle gevinstene ved å forlenge levetiden til litium-ion-batterier være betydelige. Denne optimaliseringen av levetiden vil kun være mulig gitt et tilstrekkelig estimat av den interne dynamikken og batteriets tilstand. En av disse interne tilstandene er batteriets restkapasitet. Denne oppgaven vil fungere som en fortsettelse av prosjektoppgaven. Prosjektoppgaven hentet inn eksperimentelle data, og ulike modeller ble designet og identifiserte parameterne. Denne oppgaven tar sikte på å estimere batteriets restkapasitet ved hjelp av en modellbasert estimeringsteknikk. Estimeringsmetodene som er implementert og sammenlignet er det utvidede Kalman-filteret og sentral-forskjells-Kalman-filteret. Med disse estimeringsmetodene klarer systemet å korrigere for en feil initial ladetilstand og estimere den innenfor en grense på $5\%$. Forholdet fra batteriets restkapasitet til åpen kretsspenning undersøkes også, og to metoder testes, et parametrisk polynom og lineær interpolasjon. Målestøy fra den virkelige verden kan ofte være et problem i eksperimentelle data, og et nullfase lavpassfilter brukes med gode resultater. Lithium-ion batteries represent an enormous industry and is shown to be a cornerstone in reducing greenhouse gasses. As the electrification of the transportation sector continues to increase, the potential gains of extending the life span of lithium-ion batteries are significant. This optimization of life span will only be possible given an adequate estimation of the internal dynamics and state of the battery. One such internal state is the State of Charge. This thesis will act as a continuation of the project thesis. The project thesis recorded experimental data, and different models were designed and identified. This thesis aims to estimate the state of charge using a model-based estimation technique. The estimation methods implemented and compared are the extended Kalman filter and the central difference Kalman filter. With these estimation methods, the system manages to correct for a wrong initial state of charge and estimate it within a $5\%$ bound. The relation from the state of charge to the open circuit voltage is also investigated, and two methods are tested, a parametric polynomial and linear interpolation. Real-world data noise can often be a problem, and a zero-phase low pass filter is used with good results.

Dieselelektriske marine kraftsystemer kan bli utsatt for store belastningsvariasjoner. Belastningen kan for eksempel variere imellom full fremdrifts-belastning og veldig lav hotel-belastning. Drivstoff-effektiviteten i et kraftsystem uten batteri-lagring kan være dårlig i situasjoner med lav belastning fordi hoved-generatorene vanligvis er dimensjonert for større belastninger. Implementasjon av en mindre variabel-hastighet diesel-generator foreslås som en løsning for å forbedre drivstoff-effektiviteten under situasjoner med lite belastning. En slik løsning kan være aktuell som et alternativ til et batterisystem, dersom dette ikke er tilgjengelig eller har for lav kapasitet til langvarig drift. Arbeidet presentert i denne oppgaven er sentrert rundt en Simulink-modell bestående av en variabel-hastighet diesel-generator (VSDG) og en DC/DC-omformer, som et grensesnitt mot en nett-tilkoblet virtuell synkronmaskin (VSM). VSMen kan både driftes i tilkobling hoved-generatorene og i øy-drift. Overgangen fra nett-tilkoblet til øy-drift er sømløs, som i en fysisk generator, uten noen endringer in driftsmodus. Styresystemet for generatoren og DC/DC-omformeren er hovedtemaet i denne oppgaven. En funksjonell verifisering av VSM-modellen ble gjennomført tidligere, i spesialiseringsprosjektet. Den virtuelle synkronmaskinen er derfor ikke i sentrum av denne oppgaven, men er implementert som et separat kontrollsystem for å danne en full AC-DC-AC-omformer. I situasjoner med svært lav belastning kan det være et potensiale for drivstoff-besparelser ved å redusere generatorhastigheten i tråd med belastningen. Drift ved redusert hastighet vil imidlertid medføre utfordringer i et tilfelle med en plutselig økning i belastning på grunn av tregheten i svingmassen til diesel-generatoren. En stor økning i belastning ved lav generatorhastighet kan medføre et betydelig spenningsfall, eller i verste fall, kvele dieselmotoren. For å holde systemet stabilt ved lastøkinger under redusert generatorhastighet, har en adaptiv lineær modellprediktiv regulator blitt implementert for å kontrollere generatorens hastighet og magnetisering samt DC-link strøm. En modell av systemet er laget som prediksjonsmodell til MPC og blir også brukt til små-signal tuning av et konvensjonelt kontrolldesign som er sammensatt av SISO PI-regulatorer. Modellen er laget med kjente teknikker for synkronmaskin- og kraftelektronikk -modellering. MPCen er sammenlignet med det konvensjonelle kontrollsystemet. Begge disse kontrollsystemene er funksjonstestet i en AC-DC-AC full-system-simulasjon med sprang i belastning og omkobling fra nett-tilkoblet til øy-drift. Den adaptive MPCen er bevist å kunne håndtere økninger i belastning, ved redusert generatorhastighet, bedre enn det konvensjonelle kontrollsystemet, gitt at den nødvendige spenningsreduksjon tillates. A diesel-electric ship power system may be subjected to large load variations, spanning from full propulsion load to hotel loads, or anywhere in between. The fuel efficiency in a power system without battery storage capabilities may be poor during low-load situations because the main generators are typically dimensioned for larger loads. A small variable-speed diesel generator implementation is proposed as a solution for improving fuel efficiency during low-load situations. Such a solution may take the role as an alternative in case a battery system is not available or has too low capacity for long term operation. The work presented in this thesis is done in connection with a Simulink simulation consisting of a variable-speed diesel generator (VSDG) and a DC/DC converter interfaced to an AC ship power system as a virtual synchronous machine (VSM). The VSM can operate both islanded and connected to the main generators. The transition from grid-connected to islanded operation is seamless, like in a physical generator, without any switching between operational modes. The VSDG and DC/DC converter control system is the main topic of this thesis. A functional verification of the VSM model was performed earlier, in the specialization project. The VSM is therefore not in the centre of this work but is implemented as a separate control system to form a full AC-DC-AC converter. During very low load situations there may be a potential for fuel savings by minimizing the generator speed in line with the load power. However, such a minimized speed operation will entail challenges during a sudden load increase due to the large diesel generator inertia. A large load increase during low speed operation would cause a significant voltage drop, or even worse, cause the diesel engine to stall. To keep the system stable during load surges at reduced generator speed, an adaptive linear model predictive controller is implemented to control the VSDG speed and excitation along with the DC/DC converter current. A model is made for the purpose of MPC prediction and is also used to perform small-signal tuning of a conventional control system made using multiple SISO PI controllers. The model is made using well known techniques for synchronous machine and power electronics modelling. The MPC performance is compared to that of the conventional SISO control design. Both control designs are verified in a full AC-DC-AC system simulation subjected to load changes and a transition from grid-connected to islanded operation. The MPC system is shown to handle load surges during reduced speed operation better then the SISO system, given that a momentarily output voltage reduction is allowed.

Den virtuelle synkronmaskinen er utviklet som en tilnærming for å gi fleksibel kontroll av omformere for energilagringssystemer koblet til kraftsystemer. Ved å etterligne driftstjenestene som en synkronmaskinen kan tilby, kan den virtuelle synkronmaskinene støtte driften av et kraftsystemet, samt bytte sømløst mellom nettilkobling og øydrift. I denne oppgaven er det gjennomført et analytiskstudie av et hybrid skipskraftsystem ved bruk av en virtuell synkronmaskin implantasjon for kraftkontroll. Små signal modellering og analyse med støtte fra en elektrisk simulering har blitt brukt til de analytiske studien av kraftsystemet. Hybridarkitektur i skip har vist mulighetene for å redusere utslipp og drivstofforbruk med opptil $ 30 \% $ til $ 50 \% $. Lastutjevning, toppbarbering, sentrifugeringsreserve, nullutslippsdrift og strategisk belastning er alle driftsmåter som kan bidra til redusert drivstofforbruk i skip. Bruk av en virtuell synkronmaskin koblet til et energilagringssystem kan gi de nevnte fordelene til hybridarkitekturen, samt gi en fleksibel distribuert kontroll av den kraftelektronisk omformer ved å imitere oppførselen til en synkronmaskin. Tilnærmingen av kraftsystemet har vært basert på linearisering, og har vist seg å være svært effektiv for å vurdere systemets stabilitet. For små forstyrrelser samsvarer den lineariserte tilstands-romrepresentasjonen tett med den elektriske modellen som inkluderer systemets ulineariteter. De presenterte numeriske resultatene bekreftet stabiliteten for små signal-modellen over hele det angitte driftsområdet til systemet. Gjennom parametrisk følsomhet er det observert at plasseringen av systemets kritiske egenverdier kan påvirkes ved å øke den virtuelle treghetskonstanten, $ T_a $, og dempningskoeffisienten $ k_ {d, v s m} $. En mer ønskelig frekvensrespons for systemet er oppnådd ved å øke initialverdiene til de nevnte parameterne. The virtual synchronous machine (VSM) has been developed as an approach to provide flexible control of power electronic converters with energy storage systems (ESSs) to power systems. By providing ancillary services of the synchronous machine (SM), it can support power system operation and transition seamlessly between grid-connected and islanded operation. In this thesis, an analytical study of a hybrid ship power system using a VSM for power control has been conducted. Small-signal modeling and analysis with the support of an electrical simulation have been used for the analytical study of the power system. Ship hybrid architecture has shown the capabilities of reducing emissions and fuel consumption by up to $30\%$ to $50\%$. Load leveling, peak shaving, spinning reserve, zero-emission operation, and strategic loading are all operating modes that can contribute to the reduction of fuel consumption. The VSM in conjunction with an ESS can provide the benefits of the hybrid architecture and at the same time provide flexible distributed control of power electronic converter, by imitating the behavior of the SM. In general, the applied approach based on linearization has proven to be very effective for assessing of the stability of the system. For small perturbations, the linearized state-space representation closely matches the electrical model that includes the non-linearities of the system. The presented numerical results verified the small-signal stability over the entire designated operating range of the system. Through parametric sweep analysis it has been observed that the placement of the system's critical modes can be affected by increasing the virtual inertia constant, $T_a$, and the damping coefficient $k_{d, v s m}$. A more desirable frequency response for the system has been achieved by increasing the initial values of the mentioned parameters.

Fossildrevet tungtransport genererer betydelige utslipp av NOx og klimagasser som CO2. I Norge forventes tungtransport å øke med 65 prosent innen 2050, men Norge har som mål å redusere klimagassutslippene med 80 til 95 prosent innen 2050. Elektrifisering av tungtransport vil redusere NOx og klimagassutslipp betydelig. Elektriske lastebiler har vanligvis Li-ion batteripakker med en kapasitet på 100-300 kWh, og det er utfordrende å lade en så massiv batteripakke på rimelig kort tid. For å elektrifisere tungtransport er alternative lademuligheter for batteriene avgjørende for en gjennomførbar løsning. Trådløs effektoverføring fra en vei er et alternativ til de konvensjonelle plugg-baserte ladestasjonene som fungerer i dag for lading av batteri-elektriske biler. Nyere teknologiske fremskritt har gjort en slik løsning praktisk mulig med induktiv ladeteknologi. Det er mange fordeler for applikasjoner for trådløs lading, slik som at det ikke er behov for et mekanisk støpsel, noe som gjør den mer mekanisk robust og mer tilgjengelig for automatisk lading uten manuelle inngrep. En presis overlapping av spoler i trådløs induktiv lading er avgjørende for en effektiv ladesyklus. Det er her kombinasjonen av autonom kjøring og trådløs lading virker som en perfekt kombinasjon. Ved bruk av en rekke sensorer kan autonome kjøretøysystemer holde en presis posisjon for et elektrisk kjøretøy med centimeterpresisjon. SINTEF har brukt en 1:14 skalert elektrisk lastebilmodell med et tilhørende trådløst induktivt ladesystem for å demonstrere denne teknologien i mindre skala. Tidligere studenter ved NTNU har utviklet autonome kjøretøysfunksjoner for lastebilmodellen og Bluetooth-kommunikasjon med ladestasjonen. Det er fremdeles rom for forbedringer, og denne studien har bidratt til arbeidet med å nå en mer stabil og robust autonom datasynsbasert banesporing for å oppnå en mer effektiv ladesyklus. Et tilbakemeldingssignal ble mottatt ved å koble til det interne potensiometeret ved servoen som betjener styringen. Deretter ble en lukket-sløyfe PID kontroller implementert. I tillegg er det montert et nytt kamera øverst på førerhuset på lastebilmodellen som er kombinert sammen med et eksisterende kamera for å oppnå et større deteksjonsområde. Synsfeltet er utvidet fra 30 - 100 cm til 0 - 100 cm foran lastebilmodellen. Skarpe hjørner kunne tidligere overses, og disse forbedringene ga en mer presis styring. Det er laget en bane for testing, som inkluderer ladestasjonen for lastebilmodellen. Lastebilmodellens posisjon er logget med både en IMU og LiDAR for å gi en visuell visning av forbedret autonom kjøring. En ny funksjon er implementert for å aktivere og deaktivere ladestasjonsspolene i en nøyaktig posisjon. Det er utført forskjellige tester for å finne en optimal ladesyklus med aktivering og deaktivering av spolene ved ladestasjonen i forskjellige posisjoner. Nullpunktet brukes som startpunkt, som er den nøyaktige posisjonen der spolene begynner å overføre og motta effekt. Resultatet av testene viser at en nullpunktaktivering nådde den høyeste overførte energien ved 66 prosent effektivitet, men en 5 cm kortere aktivering av nullpunktet ga en samlet høyere effektivitet på 67 prosent. En mulighets-lading system er implementert for å teste statisk trådløs lading der en av de rektangulære vegspolene er byttet med en kvadratisk spole. Med en statisk løsning ble den høyeste effektiviteten nådd, ved 80 prosent effektivitet. Dette utgjør i gjengjeld ingen tilbakelagt avstand under lading. Fossil-fueled heavy freight transport generates significant emissions of NOx and greenhouse gases such as CO2. In Norway, heavy freight transport is expected to increase by 65 percent within 2050, but Norway aims to reduce greenhouse gas emissions by 80 to 95 percent within 2050. Electrifying heavy freight transport will significantly reduce NOx and greenhouse gas emissions. Electric trucks usually feature Li-ion battery packs with a 100-300 kWh capacity, and charging such a massive battery pack in a reasonably short time is challenging. For electrifying heavy freight transport, alternative charging opportunities for the batteries are essential for a feasible solution. Wireless power transfer from a road is an alternative to the conventional plug-in-based charging stations operating today for charging battery-electric vehicles. Recent technology advances have made such a solution practically feasible with inductive charging technology. There are many advantages for wireless charging applications, such as no need for a mechanical plug. This makes it more mechanically robust, and more accessible for automated charging with no manual intervention. The alignment of coils in wireless inductive charging is crucial for an efficient charging cycle. This is where the combination of autonomous driving and wireless charging seems like a perfect match. With the use of a variety of sensors, autonomous driving systems could keep a precise position of an electric vehicle down to the centimeter. SINTEF has used a 1:14 scale battery-electric truck model with an associated dynamic wireless inductive charging system for demonstrating this technology on a smaller scale. Previous students at NTNU have developed autonomous driving functions for the truck model and Bluetooth communication with the charging station. There is still room for improvements, and this study has furthered the work to reach a more stable and robust autonomous computer vision based path tracking to achieve a more efficient charging cycle. A feedback signal was received by connecting to the internal potentiometer at the servo operating the steering angle. Then a closed-loop PID controller for the steering was implemented. In addition, a new camera has been mounted at the top of the cab at the truck model and merged with an existing camera to reach a wider detection area. The field of view has been expanded from 30 - 100 cm to 0 - 100 cm in front of the truck model. Sharp corners could previously be overlooked, and these improvements gave more precise steering. A mockup track was made for testing, and the position of the truck model has been logged with both an IMU and LiDAR to provide a visual display of enhanced autonomous driving. A new function has been implemented to activate and deactivate the road-side coils at an exact position. Various tests have been performed to find an optimal charging cycle with activation and deactivation of the roadside coils in different positions. A null-point is used as the starting point, which is the position where the vehicle-side coil starts to receive power from the road-side coils. Results from the tests showed that the null-point activation reached the highest transferred energy at 66 percent efficiency, but a 5 cm shorter activation of the null-point gave an overall higher efficiency at 67 percent. There has also been implemented an opportunity charging alternative to test static wireless charging where one of the rectangular road-side coils was exchanged with a square road-side coil. With a static solution, the highest efficiency was reached at 80 percent efficiency. However, the increased efficiency comes at the cost of zero distance traveled while charging.

Automated wireless power transfer can be seen as an enabling technology for autonomous vehicles. At SINTEF Energy Research a 1/14 scale electric truck model [13] had been used for demonstrating such technology. The truck was fitted with an induction system, but was otherwise stock. In this master thesis a system for autonomous path following has been developed and successfully implemented to demonstrate self-driving capabilities on a path that includes an inductive charger. The purpose of the vehicle is to visualize the concept of wireless charging for a self-driving truck, but on a smaller scale. Three different methods has been tested and implemented. 1. A LiDAR-based approach that utilizes Hector SLAM for mapping and localization together with a Pure Pursuit and Stanley Steering path tracking steering controller. 2. A machine-learning and camera-based approach that can navigate a track using deep learning and behavior cloning/supervised learning. This was inspired by the convolutional neural network used in Nvidia’s DAVE-2 self-driving car. 3. A classic computer vision approach for detecting lane curvature and using PID for control. ROS nodes that integrates the output from the SLAM- and deep learning-methods and combines them with a path recorder and a path tracking algorithm has been made, together with a graphical interface for monitoring vehicle states. Methods for estimating odometry without wheel encoders, dynamic speed control and obstacle detection with start/stop functionality has been implemented, along with a computer vision method for detecting ArUco markers. A complete, self-driving ROS-based system has been created. An Nvidia Jetson TX2 has been used as the main embedded computing unit. The system implements Hector SLAM as the primary SLAM method, since it does not require odometry. The Pure Pursuit path tracker was chosen as the preferred steering controller since it resulted in smooth and stable tracking of the path. The neural network of the machine-learning approach was able to steer the truck reliably and is implemented in the final version as a separate mode. This works independently of the LiDAR for when the LiDAR is out of range. The neural network has also been tested on the Udacity self-driving simulator [4]. The computer vision approach was found to be too demanding for the embedded hardware, and ended up not being used. The inductive charger system has successfully been connected and integrated to the Jetson TX2 through a CAN-bus interface in order to receive the battery state. The amount of charge received is dependent on the positioning of the truck, and is not optimized for a system with manual path creation. A better autonomous positioning system when driving across the charger is recommended, along with a better and more precise LiDAR and vehicle platform.

Et økende fokus på miljøpåvirkningen fra marine kraftsystemer som benytter seg av fossilt brensel har resultert i implementeringen av Utlippskontrollerte soner og et økende krav til nivåer av utslipp av forurensede gasser fra marine kraftinstallasjoner. En løsning for å imøtekomme disse kravene og redusere det totale forbruket av fossilt brensel for energiproduksjon er implementeringen av Energy Storage systems. Et eksempel på dette er et batterisystem. Hvis dette benyttes i et kraftsystem med AC-distribusjon er det mest vanlige grensesnittet for trefase transformasjon en trefase spenningskilde-konverter. Denne oppgaven undersøker en VSM implementasjon i et marint kraftsysem. Designet på VSMen er utført slik at frekvensen til den virtuelle rotoren blir produsert internt. Dette gjør at behovet for en PLL forsvinner. Dette gir rom for muligheten til at VSMen kan operere kraftsystemet alene, uten et behov for at andre komponenter skal styre frekvensen uten å bytte styresystem for VSMen. Kraftsystemet undersøkt er et tradisjonelt marint kraftsystem system bestående av en SM, en VSM og en lastmodell. En ulineær tilstandsrommodell har blitt laget for hver komponent. Disse har blitt linearisert til småsignalmodeller. Modellene består av VSM koblet til et nett, SM koblet til nett, VSM og SM koblet sammen og VSM, SM og lastmodell koblet sammen. Småsignal modellene utarbeidet har gjort det mulig å bruke lineære analyseverktøy for å analysere systemets stabilitet og system responsen. Småsignalmodellene ble validert ved å sammenligne responsen ved en forstyrrelse med tilsvarende modeller laget I Simulink. Dette ga muligheten for å validere modellene og vurdere responsen til systemet ved en forstyrrelse. En eigenverdianalyse ble utført på småsignalmodellene for å undersøke kritiske moder og vurdere stabilitetsmarginen. Analysen viser at systemet er stabilt og har en god dynamisk respons for transienter. Eigenverdianalysen viser at systemet har en god stabilitetsmargin. Systemet fungerer for alle komponenter tilkoblet og VSMen viser evnen til å tilpasse seg nettet den er koblet til. VSMen er i stand til å fungere for alle systemene introdusert uten å måtte endres. Dette viser muligheten for at en inverter kan operere et nett uten behov for en endring i kontrollsystemet. An increasing focus on the environmental impacts from marine power systems utilizing fossil fuel have resulted in implementation of Emission Controlled Areas, and an increasing demand for levels of pollution forces changes in the marine power systems. A solution for accommodating these demands and lowering the total use of fossil fuel for energy production is the implementation of energy storage systems. An implementation of such a system is a battery energy system. If this is used in a ship power system with AC-distribution the most common interface is for three phase transformation is three-phase voltage source converters. This thesis studies a VSM implementation in a marine power grid. The design of the VSM is done such that the frequency of the virtual rotor is produced internally, removing the need for a Phase Locked Loop. This also gives room for the possibility of islanded operation of the power grid with only the VSM connected, using the same control topology for the VSM. The power grid studied is designed as a traditional marine power grid with a SM, a VSM connected to a battery energy system and a load model. A nonlinear state space model of these components has been developed and small signal models have been made for each. Small signal models have been made for the VSM with grid connection and connected to the load model, SM with grid connection and connected to the load model and a full model containing all three components. This has given the possibility to study the dynamics of the system and gain insight in the stability. The small signal model developed have given the possibility to use linear analysis tools to evaluate stability and system response. The validation process gave room for validating the correctness of the small signal models and opened the possibility for assessing the system response of the disturbance. An eigenvalue analysis of the different small signal models has been done to determine critical modes and which states is influenced by these. This analysis has given further insight in the system response and the way that it is coupled. The eigenvalue analysis has also been tested for operational area of the power grid giving insight in how the system dynamics of the base case studied change as the total power flow is altered. The analysis shows that the system is stable and have a good dynamic response for transients. The eigenvalue analysis shows that the system has a good stability margin, meaning that the system is not in danger of becoming unstable. The system is able to function with all components (SM, VSM,load) connected. The VSM have shown the ability to adapt to the frequency in the grid by synchronizing to the other frequency setting component in the grid. This shows the possibility of a grid being operated by an inverter without the need for a change in control system.

Dette er en masteroppgave om trådløs ladning av elbil, hvor den sentrale delen handler om hvordan kontrollere ladeprosessen. Denne oppgaven er et sluttprodukt av 5 år studie ved institutt for teknisk kybernetikk ved Norges Teknisk-Naturvitenskaplige Universitet (NTNU). Trådløs ladning har de siste årene tatt av for små elektronikk, for eksempel mobiler. Det er systemer med små strømmer og spenninger. Ved ladning av batteriet til en elbil, spenningen er signifikant høyere. Det er tatt utgangspunkt i en matematisk modell for trådløs overføring av effekt, og utvidet denne modellen med en konverter for å kunne kontrollere spenning eller strøm. Modellen for den trådløse overføringen består av en spenningskilde inn, kompenserende kondensator og overføringsspolen på sendersiden. På mottakersiden er det en mottakerspole, kompenserende kondensator og en diodebro for å rette vekselspenningen. To forskjellige konvertere er modellert, med til sammen tre forskjellige kontrollstrategier. Den utvidede modellen er skrevet på tilstandsromformat, og så linearisert for å kunne analyseres med bruk av lineære teorier. Systemmatrisene som er et resultat fra lineariseringen, har blitt analysert ved å se på egenverdiene. For hver egenverdi har det blitt sett på hvilke tilstander er med på å påvirke denne egenverdien, og hvilke av parameterne i systemet egenverdien er sensitiv ovenfor. Det blir konkludert med at en kontrollstrategi med to kontrollere i kaskade gir redusert oscillasjoner i systemet. Det små forskjell når det kommer til parameter sensitiviteten eller participation factoren for de to første casene. Systemet er sensitivt ovenfor vinkelfrekvensen og konverterkapasitansen. For den siste caset er konklusjonen at sensitiviteten til kontrollkoeffisienten er stor at bytte av referanse fra strøm til spenning inn over kondensator fører til tap av direkte kontroll av ladestrømmen. Wireless charging has in recent years taken off for small electronics, such as mobiles. There are systems with small currents and voltages. When charging the battery to an electric car, the voltage is significantly higher. This makes the system requirements different, and only a few systems are available today. The work is based on a mathematical model for wireless transmission of power, and the focus is on this model and extending it with a converter to control voltage or current on the receiving side. The model of the wireless transmission consists of a voltage source, compensating capacitor and the transmission coil on the transmitter side. On the receiving side, there is a receiver coil, compensating capacitor and a diode bridge to rectify to DC voltage. The DC voltage is to be controlled by a DC-DC converter. Two different converters are modelled, a boost converter and a buck converter. Two different control strategies are investigated for control of the boost converter, and one control strategy for the buck converter. For the boost converter, the first case is to use one controller to control the battery charging current. The second case is to use two controllers in a cascaded system to control both the battery charging current and the current in the converter inductance. For the buck converter, the control strategy is two controllers in cascade, with voltage over the converter capacitance as a reference and with control of the converter inductance. There is a total of three different control strategies. The model is written on state space format and then linearized to be analysed using linear theories. The system matrices, the result from linearization, have been analysed by looking at eigenvalues. For each eigenvalue, it has been seen which states help to influence this eigenvalue, and which of the parameters in the system the eigenvalue is sensitive to. There is no significant difference between the first two cases when it comes to the parameter sensitivity or the participation factor. The system is sensitive to the angular frequency and the converter capacitance. When it comes to the last case, there is a sensitivity to the entire system to the propositional control coefficient in the inductor current controller. The change of reference from the battery current to the voltage over the capacitance reduces the control of the charging current from directly controlling the current to indirect control. The conclusion is that a control strategy with two cascade controllers gives reduced oscillations in the system.

This thesis studies the effect of controlling gate switches in an APF using FCS-MPC. This gave a cascaded MPC structure. Where internal states and a calculated reference from the first MPC is used by a second MPC. The second MPC was designed to track the reference. An FCS-MPC with continuous variables was designed and compared to existing benchmark implementations. Including an ILS-FCS-MPC, an ICS-MPC and a hysteresis controller. The CCV-FCS-MPC was implemented in ACADO. It was based an existing ILS-FCS-MPC. Using an existing model formulation. The cost function was tuned based on theory on Sequantial Quadratic Programmin (SQP). Using a trial and error method to arrive at the final cost function expression. All gate controlling implementations were then verified in a simplified version of the ship power system. Then implemented in the Cascaded MPC structure with the existing harmonic mitigating MPC. First reference tracking performance and power loss was measured. Results show that an CCV-FCS-MPC with a tuned cost function outperforms hysteresis controllers, ICS-MPC with PWM and an existing ILS-FCS-MPC. With a reduction in SSE of at least 50\% compared to any other implementation, while also reducing power loss. Results show that THD in a ship power system is reduced for all MPC implementations, compared to a hysteresis controller. Both for an ICS-MPC, an ILS-FCS-MPC and a CCV-FCS-MPC. The CCV-FCS-MPC with a tuned cost function outperforms all other tested implementations. It reduces generator current THD by over 20\% compared to any other implementation. While also reducing power loss and maintaining bus voltage THD. Compared to the hysteresis controller.

Nepal has huge hydropower potential of 83,000 MW, of which 42,000 MW is technically and economically feasible. According to the annual report of Nepal Electricity Authority, the total installed capacity of hydropower in Nepal is only 733.57 MW. Although bestowed with tremendous hydropower resources, only about 40% of Nepal s population has access to electricity and remaining 60% faces severe load shedding. In this scenario, the cross border transmission links between Nepal and India serves two purposes. First, since Nepal still is short of electric power, these transmission links can be used to import power from India. Second, in future when there is surplus electric power in Nepal, it can be used to transfer power to India. In the thesis, a model of integrated power system of Nepal is made in the Matlab Simulink. All the major power plants with capacity above 30 MW has been considered. The plants are connected to major loads with the transmission lines. In the start of the thesis, it was planned to find the real parameters of generators, governors, excitation systems, power transformers, transmission lines and loads. But later because of difficulty in finding the real parameters, the standard values were used. The grid on the Indian side is represented by three phase voltage source. The best place for transmission link with India is in the central region between Butwal and Dhalkebar. Among different options for putting cross border transmission line, Dhalkebar-Muzaffarpur link is considered for the thesis. Different possible configuration were made using HVAC and HVDC in the cross border link. The HVAC link is considered first and then HVDC is kept in parallel. The power through the HVDC transmission line is regulated by setting the parameter values in the controller of the HVDC model. The simulation result shows that the system is unstable when there is HVAC in the cross border link. With HVDC only, the system is stable and also the control of the power flow through cross border is possible. Different condition of power transfer has been obtained and presented. Even though the values of generators parameters were taken as standard parameters, the excitation and governor parameters were same for all power plants. Either real value of the parameters should be used or standard values should be used according to the size and location of the power plant. The location of hydropower plants and load center are exact. The rated capacity of Upper Tamakoshi hydro plant is taken as 100 MVA, although its actual capacity is 540 MVA. It was because for its rated capacity above 100 MVA, the system becomes unstable. During the simulation, the control of power flow through HVAC was not possible. The model involving HVAC was not stable also. Stable condition were found using HVDC only.

In this work, previous small-signal modelling and analysis carried out for a VSM implementation has been adapted and expanded to include a synchronous machine (SM) with a control scheme similar to those of traditional generation units and to the one used in the VSM implementation. A non-linear state-space analytical model has been developed for a traditional power generation unit in grid-connected mode (SM system). The model allows the study of the SM system in similar conditions/context as the grid-connected VSM implementation (VSM system), and facilitates their comparison. Moreover, it can represent the most relevant dynamic characteristics of a round or salient-pole rotor SM, including the amortisseur circuits or the representation of the corresponding damping torque in the swing equation of a reduced-order version of the model. The developed model has been linearised analytically, to obtain the respective small-signal model. The eigenvalue analysis of the modelled systems has indicated that most of the SM system eigenvalues (modes) are slower or not as well damped as the VSM system ones. Moreover, most SM system modes have shown to be relatively fixed and determined mainly by non-tunable parameters. The modelled exciter has presented little capability of increasing the damping of oscillations, and its integral gain has demonstrated to be the tunable parameter that can most easily cause instability in the SM system. The location of the 4 slowest and least damped SM system eigenvalues has shown to be limited by non-tunable parameters, indicating that the performance or robustness of such system cannot be significantly improved. A trade-off has been observed in the placement of the VSM system critical modes: increasing the virtual impedance improves the damping of oscillations, while making the system response slower. However, parametric sweep analysis has indicated that a good compromise between performance and robustness could be achieved by decreasing the virtual impedance and increasing the phase lock loop proportional gain.

The development of far-offshore wind farms and other large-scale renewable energy sources, together with the increasing needs for long distance power transmission is resulting in more HVDC systems being integrated into the traditional AC power network. Due to the capability for operation in isolated AC grids, VSCs are becoming the preferred technology for HVDC systems. Additionally, VSC HVDC allows for more flexible power control within a network than the conventional LCC systems. Moreover, the need for flexible transmission capacity to balance fluctuating power generation from renewable sources over wide geographical areas combined with the corresponding potential benefits in a deregulated power market is expected to favour the concept of VSC based Multi-terminal HVDC system (MTDC). MTDC provides enhanced reliability and functionality and reduces the cost and conversion losses. However, most HVDC transmission schemes are currently constructed as point-to point connections, and there is not yet any clear standardization of voltage levels. Thus, DC/DC converters will become necessary if existing or emerging HVDC links operating with different voltages and different configurations. i.e. monopolar and bipolar should later be interconnected into MTDC configurations. DC/DC converters might also be needed for power flow control in meshed MTDC grids. The goal of this thesis is to summarize the requirements for DC/DC converters in HVDC applications and focus on the modelling and control of DC/DC converters for various applications. The modelling includes both switching and average models. Four types of DC/DC converters intended for different applications are modelled using MATLAB/ Simulink® platform. The functionality of the developed models are demonstrated by simulations in MTDC grid based on the CIGRÉ B4 DC grid test system.