Search

Les xarxes de bici urbanes són un punt central en moltes ciutats, grans o petites, per tal de reduir la contaminació ja que permeten fer recorreguts curts i mitjans de forma còmode sense necessitat del vehicle privat o el transport públic. És per aquest motiu que esdevé molt important tenir una bona xarxa que permeti que els usuaris puguin realitzar els seus desplaçament i en facin ús. En aquest treball s'analitza una xarxa ja existent i es desenvolupa i s'aplica un algoritme genètic que permet optimitzar les diferents estacions o punts de servei. Es creuen dades reals de les estacions d'una ciutat amb els viatges que s'hi van realitzar en un període de temps. La funció objectiu que es minimitza són els viatges fallits, que poden ser viatges que no s'haurien pogut realitzar perquè en aquell moment no hi havia bicis a l'estació o els viatges que no s'haurien pogut finalitzar perquè no hi havia espais lliures. Els cromosomes de la població tenen dos al·lels, un amb els espais de l'estació i l'altre amb el nombre de bicis a l'iniciar el dia. Es tracta d'un problema complex de resoldre els resultats del qual han estat molt bons, ja que, tot i executar-se amb poques generacions i una població inicial no gaire gran, s'ha millorat molt el resultat si es fa servir la xarxa real. Per aquest motiu es podria considerar que es tracta d'un bon algoritme per resoldre aquest tipus de problemes d'optimització. Urban bike-sharing is a central point in many cities in order to reduce pollution because it allows citizens to travel medium and short distances without using private car or public transport. For that reason, is important to have a good network to carry user's travels to make it useful. This work analyses an existent network and develops and applies a genetic algorithm to optimize all the bike stations. Real data of the trips and the stations are joined in a period to evaluate the algorithm. The Objective function is to minimize failed trips, which means that a user could not start because there is not a bike at that station or that a user could not finish because there are not empty docks. Chromosomes have two alleles, one with docks at station and the other with the number of bikes at the beginning of the day. The results of the algorithm applied to these complex problems are quite good because even executing with a few generations and small initial population it improves the real network results. Therefore, we could consider that a genetic algorithm is a good choice to resolve that kind of optimization problems. Las redes de bici urbanas son un punto central en muchas ciudades, grandes o pequeñas, para reducir la contaminación puesto que permiten hacer recorridos cortos y medianos de forma cómodo sin necesidad del vehículo privado o el transporte público. Es por este motivo que acontece muy importante tener una buena red que permita que los usuarios puedan realizar sus desplazamiento y hagan uso. En este trabajo se analiza una red ya existente y se desarrolla y se aplica un algoritmo genético que permite optimizar las diferentes estaciones o puntos de servicio. Se creen datos reales de las estaciones de una ciudad con los viajes que se realizaron en un periodo de tiempo. La función objetivo que se minimiza son los viajes fallidos, que pueden ser viajes que no se habrían podido realizar porque en aquel momento no había bicis a la estación o los viajes que no se habrían podido finalizar porque no había espacios libres. Los cromosomas de la población tienen dos alelos, uno con los espacios de la estación y otro con el número de bicis al iniciar el día. Se trata de un problema complejo de resolver los resultados del cual han estado muy buenos, puesto que, a pesar de ejecutarse con pocas generaciones y una población inicial no muy grande, se ha mejorado mucho el resultado si se usa la red real. Por este motivo se podría considerar que se trata de un buen algoritmo para resolver este tipo de problemas de optimización.

Orientador: Guilherme Palermo Coelho Dissertação (mestrado) - Universidade Estadual de Campinas, Faculdade de Tecnologia Resumo: O planejamento da estratégia de extração de petróleo em campos de águas profundas é uma das etapas essenciais da engenharia de reservatórios. É nessa etapa que são de?nidos os parâmetros de construção e operação do sistema de produção que será implantado no campo, de forma a otimizar algum critério, como o lucro gerado pelo campo ou o volume total de óleo que será extraído durante o período de concessão. Apesar de ser possível utilizar ferramentas clássicas para resolver esse problema de otimização, para que tais métodos funcionem diversas simpli?cações devem ser consideradas, principalmente por conta dos níveis de incerteza relacionados ao problema. Tais simpli?cações podem levar a soluções que não condizem com a realidade. Meta-heurísticas são ferramentas alternativas de otimização que se mostram adequadas para problemas onde os métodos clássicos não se mostram e?cientes, uma vez que, ao contrário das metodologias clássicas de otimização, as meta-heurísticas não exigem grandes adaptações para serem aplicadas ao problema que se deseja resolver. No entanto, é comum que tais ferramentas exijam um elevado número de avaliações de soluções candidatas até que uma solução adequada seja obtida. Esta característica é particularmente desfavorável para o problema de de?nição de estratégia de produção, uma vez que a avaliação de uma solução exige a simulação computacional da evolução do campo de petróleo durante todo o período de exploração. Essa simulação pode levar várias horas dependendo do nível de detalhamento do modelo do campo. Sendo assim, esse trabalho se propõe a estudar o uso de meta-heurísticas para a resolução do problema de de?nição de estratégia de produção, com o objetivo de reduzir o custo computacional para que uma solução de boa qualidade seja encontrada. De forma mais especí?ca, aqui foram desenvolvidas versões do algoritmo genético, bem como de seus operadores de busca, para o problema de posicionamento de poços de uma estratégia de produção. Tais modi?cações utilizam de conhecimento acerca do problema para explorar de forma mais e?ciente o espaço de busca. As versões clássica e propostas de algoritmos genéticos foram utilizadas para otimizar uma estratégia de produção em um modelo sintético de reservatório de petróleo, os resultados foram então apresentados e comparados com uma busca aleatória e uma ferramenta comercial de otimização, indicando que, apesar de não superar a ferramenta comercial em termos de qualidade de soluções, os algoritmos genéticos propostos foram capazes de obter soluções de boa qualidade em um tempo significativamente inferior Abstract: Planning the oil extraction strategy to be deployed in deep-water ?elds is one of the most important steps of reservoir engineering. In this step, several parameters related to the construction and operation of the production system to be implemented must be de?ned. These parameters are adjusted to optimize some criteria, such as pro?t or the total volume of oil that will be extracted during the concession period. Although it is possible to use classic tools to solve this optimization problem, these methods need several simpli?cations to work properly, mainly, due to all the uncertainties associated with this problem. Such simpli?cations can lead to solutions that do not satisfy reality. Meta-euristicsarealternativeoptimizationtoolssuitableforproblemswhereclassictools are not e?cient as they require little adaptation to be applied to a particular problem. However, such algorithms often require a high number of objective function evaluations to obtain high-quality solutions. This characteristic is particularly adverse the de?nition problem, since the evaluation of each candidate solution requires a computational simulation of the evolution of the oil ?eld throughout the period of operation. Each simulation can take several hours to complete, according to the level of details adopted for the ?eld¿s model. Therefore, this project aims to reduce the cost in the search to ?nd good solutions for production strategy de?nition problem by using meta-heuristics. In a more speci?c way versions of the genetic algorithm was developed for this problem. This modi?ed versions of the algorithms use problem knowledge to explore the search space more e?ciently. Both the classical versions of the genetic algorithm and the versions proposed here were used to optimize a production strategy in a synthetic model of oil reservoir. The results were then presented and compared with a random search and a commercial optimization tool. The results have shown that, although the genetic algorithms were not capable to overcome the commercial tool with respect to the quality of the solutions, they were still able of ?nding high quality solutions in a much shorter time Mestrado Sistemas de Informação e Comunicação Mestre em Tecnologia CAPES

Orientadores: Fernando Jose Von Zuben, Cynthia Cristina Martins Junqueira Dissertação (mestrado) - Universidade Estadual de Campinas, Faculdade de Engenharia Eletrica e de Computação Resumo: O estudo de antenas de micro?ta envolve a especi?cação de um amplo espectro de parâmetros físicos, e cada proposta de antena requer o emprego de algoritmos avançados de cálculo numérico para medidas de desempenho, como impedância de entrada, padrão de irradiação e ganho em faixas especí?cas de freqüência. O principal objetivo desse trabalho é desenvolver um ambiente computacional combinando um software de análise de estruturas eletromagnéticas, MStrip40, baseado no método dos momentos (MoM), e um método de busca evolutivo, denominado algoritmo genético. Buscam-se valores ótimos de parâmetros físicos de antenas de micro?ta, como comprimento e largura do elemento irradiador e da linha de alimentação, entre outros. São tratados vários tipos de antenas de micro?ta, e um elenco amplo de critérios de desempenho é considerado, mantendo sempre um valor baixo para o coe?ciente de onda estacionária na freqüência de projeto. A validação dos resultados gerados pelo ambiente computacional é alcançada através da fabricação da melhor antena obtida em cada caso de estudo, seguida por testes de desempenho do dispositivo físico fabricado, realizados em laboratórios especializados Abstract: The study of microstrip antennas involves the speci?cation of a broad spectrum of physical parameters, and each antenna proposal requires the use of advanced algorithms of numerical calculus to measure the performance, such as input impedance, radiation pattern and gain in speci?c frequency bands. The main objective here is to develop a computational environment joining a software for electromagnetic structure analysis, called MStrip40 and based on the Method of Moments (MoM), together with an evolutionary search method founded on a genetic algorithm. The outcome is composed of optimal values for physical parameters of the microstrip antenna, such as length and width of the radiator element and of the feeding line. Several kinds of microstrip antennas are treated, and a wide range of performance criteria are considered, always keeping a low value for the coef?cient of stationary wave at the design frequency. The validation of the results produced by the computational environment is achieved by means of the manufacture of the best obtained antenna in each case study, followed by performance evaluation of the manufactured physical device, which is accomplished in dedicated laboratories Mestrado Engenharia de Computação Mestre em Engenharia Elétrica

Orientador: Maria Teresa Moreira Rodrigues Tese (doutorado) - Universidade Estadual de Campinas, Faculdade de Engenharia Quimica Resumo: O presente estudo trata da otimização da logística de transporte e recepção de cana-de açúcar em indústria sucro-alcooleira, com base em modelos de filas de espera, seguido de modelo para otimização da produção, no âmbito da programação linear, cujo objetivo é a maximização do lucro total, advindo da comercialização de seus produtos (açúcar e álcool) e sub-produtos (bagaço de cana, torta de filtro, vinhaça e eletricidade). Para o estudo das filas de espera, a metodologia adotada consistiu na análise do sistema de transporte de cana da Usina Aliança, indústria localizada no interior do Estado da Bahia, Brasil. Os dados coletados alimentaram o software de planilha eletrônica excel, da Microsoft, empregado para o teste de aderência das distribuições de probabilidade. Com base no software Quntitative Systems for Business Plus - QSB+ foram feitas simulações de descarregamentos de cana. Para a programação linear, por sua vez, tomou-se por base estudo detalhado das operações unitárias de 95 indústrias desse segmento, realizado pelo Instituto de Pesquisas Tecnológicas do Estado de São Paulo - IPT. O recurso informatizado neste caso, consistiu também, do software excel, através do solver, tendo sido usada para a simulação de cenários relativos à maximização do lucro. Os resultados obtidos indicaram que um re planejamento da frota de transporte de cana, com diferentes capacidades é, de fato, mais indicado para a otimização do descarregamento nas usinas. Quanto à otimização da produção, conclui-se que o modelo formulado responde satisfatoriamente a variações de preços de seus produtos, indicando cenários de produção mais adequados. O presente estudo contribui para a melhoria da indústria de açúcar e álcool, fazendo ligação entre a função suprimentos e a função produção, de modo sistematizado, e apoiado em bases das Engenharias Química e de Produção. Abstract: This study is about optimizing sugar-cane transportation and delivery logistics in the sugar and alcohol industry, based on waiting-queue models following a model for optimizing production, using linear programming, with the objective of increasing the total profit resulting from the sale of the products and by-products: sugar-cane bagasse, filter cake, stillage, and electricity. The methodology employed in the waiting-queues study consisted of analyzing the transportation system of the Aliança Mill, a facility located in rural Bahia, Brazil. Data collected were manipulated with statistics tools found in Microsoft Excel, in order to perform the probability distributions adherence test. Quantitative Systems for Business Plus, QSB+ software was used to perform simulations of cargo unloading in which the number of cargo unloading facilities (dumpers) was varied. The mathematical model was based on a detailed study of the operating units of 95 industries of this area, developed by Instituto de Pesquisas Tecnológicas do Estado de São Paulo - IPT. The computer resources also included Microsoft Excel. Its "solver" function was used to simulate scenarios related to maximizing profit. The results show that re-planning the transportation fleet to include a mix of vehicles with different load capacities is recommended to achieve optimal cargo unloading at the mills. Concerning production optimization, this study shows that the model satisfactorily responds to possible price variations of products and by-products, indicating suitable production composition scenarios. This study thus contributes to processes improvement in the sugar and alcohol industry, linking the supply function to the production function in a systematized way, based on Chemical and Production Engineering. Doutorado Sistemas de Processos Químicos e Informática Doutor em Engenharia Química

Over the past years, deep neural networks have proved to be an essential element for developing intelligent solutions. They have achieved remarkable performances at a cost of a large size with deeper layers and millions of parameters. Therefore utilising these networks for developing augmented reality applications on limited resource platforms such as embedded devices or mobile phones is a challenging task. In this context, this thesis addresses the problem of neural networks compression and optimisation in order to enhance the performance of these models on limited resource systems.In the first part of this manuscript, we present an overview of different compression methods present in the literature as well as their strengths, their weaknesses and a brief comparison of these techniques. We are also interested in the methods that are allowing the optimisation of deep neural networks structure design, from simple modules to autonomous models building.In the second part, we show the feasibility of an augmented reality application in real-time using a deep learning model in order to achieve a face parsing task. By using adapted frameworks and an optimised architecture, we achieve the segmentation of different face components in real-time with a high level of consistency on an iPhone X.The last two parts are focused on the development and the evaluation of a new deep convolutional neural networks compression method. Based on a regularisation term which is defined on the filter coefficients of the model, our approach is inducing sparsity among the weights of the network. Thus, our method is redistributing the information between the model filters, enabling us to remove the filters with the smaller values. We show the performance of our method on classic classification tasks. We are also introducing the efficiency of our technique on more complex models and tasks such as classification, segmentation and detection problems while specialising these models on a subset of categories on several databases., Au cours de ces dernières années, les réseaux de neurones profonds se sont montrés être des éléments centraux dans le développement de solutions intelligentes. Ils ont atteint des performances remarquables au détriment de la grande taille de leurs modèles avec de nombreuses couches profondes et des millions de paramètres. Ainsi, utiliser ces modèles pour des applications en réalité augmentée devant fonctionner sur des plateformes possédant des ressources limitées comme des systèmes embarqués ou des smartphones est une tâche loin d'être évidente. Dans ce contexte, cette thèse s'intéresse au problème de la compression et de l'optimisation de réseaux de neurones dans le but de réduire la taille de leur modèle sur des systèmes ayant des ressources limitées.Nous présentons dans une première partie différentes méthodes de compression de la littérature ainsi que leurs forces, leurs faiblesses et une brève comparaison entre elles. Nous nous intéressons également aux méthodes permettant d'optimiser la construction des architectures de réseaux de neurones profonds, allant de simples modules à la construction autonome de modèles.Nous montrons dans une deuxième partie la faisabilité technique d'une application en réalité augmentée et en temps réel utilisant un modèle d'apprentissage profond pour de l'analyse de visages. En utilisant des frameworks adaptés et une architecture optimisée, nous arrivons à segmenter différents composants du visage de manière robuste en temps réel sur un iPhone X.Les deux dernières parties sont consacrées au développement et à l'évaluation d'une nouvelle méthode de compression de réseaux de neurones convolutifs profonds. Basée sur un terme de régularisation défini sur les coefficients des filtres du modèle, notre approche permet d'introduire de la dispersion au sein des poids du réseau. Cela a pour effet de redistribuer l'information entre les filtres du modèle, nous permettant de supprimer les filtres ayant une valeur faible suite à cette opération. Nous évaluons les performances de notre méthode sur des tâches de classification classiques. Nous introduisons aussi l'efficacité de notre technique sur des modèles et des tâches de classification, segmentation et détection plus complexes en spécialisant les modèles sur certaines sous-catégories de jeux de données.

Neste trabalho é apresentado um estudo sobre o uso de quatro ferramentas computacionais não-tradicionais com a finalidade de detecção e classificação automática de defeitos de montagem e fabricação em câmbios automotivos a partir de sinais de vibração. Devido ao estado do engrenamento poder ser caracterizado por modulações em amplitude e fase, duas técnicas de demodulação são consideradas: a demodulação da amplitude e da fase dos sinais de engrenamento utilizando mínimos quadrados e a decomposição via Transformada Stockwell, na qual são empregadas janelas que se ajustam ás características de modulação do sinal. As outras duas abordagens levam em conta a amplitude das freqüências de engrenamento e suas bandas laterais. Em uma, são adotadas ferramentas estatísticas, como kurtosis, fator de crista e nível médio quadrático, aplicadas às parcelas temporais associadas às freqüências de interesse obtidas após aplicação da Transformada Choi-Williams aos sinais e na outra são calculadas a soma das magnitudes dos bispectros e a soma dos bispectros relacionados à freqüência de engrenamento dos sinais de vibração. As metodologias propostas foram aplicadas aos dados simulados de vibrações de uma caixa de transmissão, onde em um estudo preliminar foi avaliada a viabilidade do emprego de cada uma. As metodologias escolhidas - demodulação da amplitude e fase e a que utiliza a transformada de Choi-Williams, foram submetidas a uma análise de sensibilidade com sinais simulados de engrenamento para avaliar a capacidade de cada uma quanto ao grau de modulação presente nos sinais, sob diversas condições de contaminação por ruído. Finalmente, estas duas abordagens foram aplicadas aos sinais de vibração obtidos de caixas de transmissão automotivas preparadas para esta finalidade. In this work a study about the use of four mathematical-computational nontraditional tools with detection purpose and automatic classification of assembly and manufacturing defects in automotive gearboxes from vibration signals is presented. Due to the State of the gear meshing to be characterized by amplitude and phase modulations, two demodulation techiniques are considered. The demodulation of the amplitude and phase of the gear meshing signals using least squares and decomposition by Generalized S Transform, in the wich are used Windows that are adjusted to the characteristics of signafs modulation. The other two approaches, take into account the amplitude of the gear meshing frequencies and their lateral bands. In one, statistical tools, as kurtosis, crest factor and root mean square levei are adopted, applied to the time parcels associated to the frequencies of interest of Choi-Williams Transforms of the signals, and in another one is calculated the sum of the global Bispectrum magnitudes and also the sum of bispectrum related to the gear meshing frequency only. The proposed methodologies were applied to the simulated gear meshing vibration data, where a peliminary study is made for to evaluate the viability of each one. The chosen methodologies - Amplitude and phase demodulation and the one that uses Choi-Williams Transform were submitted to sensitivity analysis with simulated gear meshing signals in order to evaluate their capabilities how much to the modulation degree present in the signals, under several conditions of noise. Finaly, these two approaches were applied to the real vibration signals obtained of an automotive transmission. Tese (Doutorado)

Atomic Force Microscopy (AFM) appears as a very useful and versatile instrument for imaging surfaces and for characterizing materials properties at nanoscale such as electrical properties of thin layer materials. The electrical characterization notably involves a three-dimensional localization of the space charge within the submicron materials. Techniques derived from the AFM such as Electrostatic Force Microscopy (EFM) or Kelvin Force Microscopy (KFM); provide a considerable advantage, allowing the acquisition of new information down to nanoscale on the charge build up in dielectric materials and on their ability to store and dissipate charges. Concerning charge density determination, these techniques have been limited by their lack of in-depth sensitivity, i.e. it failed to determine precisely the spatial distribution of the charges in thin dielectric. Knowing charge distribution in the volume is mandatory for a better understanding of charge transport and trapping phenomena in thin dielectric films and at metal/dielectric interface. To overcome this issue, a new technique named Electrostatic Force Distance Curve (EFDC) has been proposed. The EFDC is based on the measurement of the electrostatic force between the AFM probe and the sample. Preliminary experimental results show that the EFDC seems to be sensitive to the location of the charge in 3D, or to electrostatic potential localization. However, the relative contribution of charge distribution (density, lateral/in-depth spreading) is unknown and need to be identified to point out EFDC ability to determine 3D charge distribution. The aim of this thesis is therefore to understand the relationship between the shape of EFDC and spatial distribution of the space charge in the material. In particular, answers have to be provided as the possibility to determine the relative contribution of charge distribution (density, lateral/in-depth spreading) using only EFDC. This is an inverse problem to be solved in order to recover, from experimental measurements, the electrostatic forces (EFDC) created by the charges, the spatial distribution of the charges (density, lateral/in-depth spreading). The thesis is divided into 3 parts. The first part is devoted to develop a realistic and complete model for electrostatic interaction between an AFM probe and a flat thin dielectric layer. In order to reproduce experimental conditions and to base the work on a well-controlled source of force, EFDCs were simulated over a polarized buried electrode within the dielectric. Results reveal a very good agreement between the model and experimental data. The second part is devoted to study EFDC sensitivity to charges localization in thin dielectric using this model. Numerical simulation is implemented to estimate the electrostatic force exerted on the AFM probe induced by a half-ellipsoid charge distribution. The sensitivity of the EFDC shape to charge localization is investigated for the charge cloud parameters as lateral radius, depth and charge density.Finally, we developed an inverse model to localize the injected charge in thin dielectric layers. This model will take an experimental force curve (EFDC) as the input and its output will be the charge cloud parameters: the radius, the density and the depth. Several optimization techniques are tested. The results show that different sets of parameters induce the same force curve (local minimums): charges with high depth and with low density seem to give the same results as surface charges with high density. In addition, we have proposed an original method to determine the lateral spreading of charges knowing only the distribution of the surface potential of the dielectric. By enriching the inverse model with this information, the results show that each EFDC corresponds to a single charge configuration. The optimization algorithm converges towards a global minimum; a unique triplet is then identified: radius, depth and charge density.; La microscopie à force atomique (AFM) est un instrument de mesure qui semble adapté à la caractérisation des matériaux diélectriques à l'échelle nanométrique. Cette caractérisation électrique passe notamment par une localisation tridimensionnelle de la charge d’espace au sein des matériaux submicroniques. Des techniques dérivées de l'AFM comme la microscopie à force électrostatique EFM, ou à force de Kelvin KFM, permettent d'obtenir de nouvelles informations, à l'échelle nanométrique, sur l'état de charge des isolants et sur leur capacité à stocker/dissiper les charges. Cependant, ces techniques ne permettent pas de connaître précisément la densité de charge ni sa répartition en profondeur, données indispensables pour une meilleure compréhension des phénomènes de transport et de piégeage de charge. Depuis quelques années, l’utilisation des courbes de force électrostatique, EFDC (Electrostatic Force-Distance Curve), basée sur la mesure des forces électrostatiques au cours de phases d'approche/retrait entre la sonde AFM et les charges stockées dans le matériau ont donné des résultats encourageant. En effet, les EFDC semblent montrer une sensibilité aux paramètres caractérisant la répartition spatiale de la charge dans les matériaux diélectriques (leur profondeur, leur répartition latéral, etc.). L’objectif principal de cette thèse est donc une meilleure compréhension des interactions en jeu lors des mesures EFDC pour la caractérisation électrique des matériaux de manière à améliorer la sensibilité de la mesure. En particulier, des réponses devront être apportées quant à la possibilité de connaître la répartition latérale et en profondeur des charges à l’aide des seules courbes EFDC. Il s’agit d’un problème inverse à résoudre pour recouvrer, à partir de mesures expérimentales, ici les forces électrostatiques créées par les charges, la répartition spatiale des charges. La thèse se décline en 3 parties. La première partie concerne le modèle 3D de l’AFM sous le logiciel commercial Comsol dans le but de simuler les EFDC induites par des charges en respectant 2 critères : un temps de calcul le plus faible possible, une précision la plus grande possible. La deuxième partie concerne l’étape de validation du modèle avec des données expérimentales. Pour avoir des données exploitables et au plus proche des conditions réelles, les charges sont dans un premier temps assimilées à des électrodes enfouies sous potentiel électrique. Les résultats issus de ces expériences sont ainsi comparés au modèle pour d’une part le valider et, d’autre part, initier notre étude sur la sensibilité des EFDCs sur les paramètres de localisation de la charge sur un cas où ses données de localisation sont accessibles expérimentalement (ici, les électrodes enfouies). Enfin, la troisième partie concerne l’utilisation de méthodes inverses dans le but de localiser la charge à partir des EFDCs. Plusieurs techniques d’optimisation sont testées. Les résultats montrent que des jeux de paramètres différents induisent une même courbe de force (plusieurs minimums locaux) : des charges en profondeur mais à faible densité semblent donner les mêmes résultats que des charges en surface mais à forte densité. De plus, nous avons proposé une méthode originale pour déterminer l’étalement latéral des charges en connaissant uniquement la distribution du potentiel en surface du diélectrique. En enrichissant le modèle inverse avec cette information, les résultats montrent que chaque EFDC correspond à une seule configuration de charges. L’algorithme d’optimisation convergeant vers un minimum global, un triplé unique est alors identifié : le rayon, la profondeur et la densité de charge.

No concreto de agregado leve, a resistência à compressão e o módulo de elasticidade são as propriedades mecânicas mais importantes e consequentemente as mais comumente analisadas. A relação entre os componentes do concreto de agregado leve e suas propriedades mecânicas é altamente não linear, e o estabelecimento de um modelo de previsão abrangente de tais características é usualmente problemático. Existem trabalhos que buscam encontrar essa relação de formas empíricas. Há também trabalhos que buscam aplicar técnicas de inteligência computacional para prever essas propriedades a partir dos componentes do concreto. Prever com precisão as propriedades mecânicas do concreto de agregado leve é um problema crítico em projetos de engenharia que utilizam esse material. O objetivo desta dissertação é avaliar o desempenho de diferentes métodos de inteligência computacional para prever a módulo de elasticidade e a resistência à compressão aos 28 dias de concretos de agregados leves em função do fator água/cimento, volume de agregado leve, quantidade de cimento e densidade do agregado leve. Para a escolha da melhor configuração de cada método, foi definida uma metodologia utilizando o algoritmo de otimização PSO (Particle Swarm Optmization). Por fim, é verificada a capacidade de generalização dos métodos através do processo de validação cruzada de modo a encontrar o método que apresenta o melhor desempenho na aproximação das duas propriedades mecânicas. In lightweight aggregate concrete, the compressive strength, the elastic modulus and specific weight are the most important properties and consequently the most commonly analyzed. The relationship between lightweight aggregate concrete components and their mechanical properties is highly nonlinear, and establishing a comprehensive predictive model of such characteristics is usually problematic. There are works that seek to find this relation of empirical forms. There are also works that seek to apply computational intelligence techniques to predict these properties from the concrete components. Accurately predicting the mechanical properties of lightweight aggregate concrete is a critical problem in engineering projects that use this material. The objective of this dissertation is to evaluate the performance of different computational intelligence methods to predict the elastic modulus and the compressive strength at 28 days of lightweight aggregates concrete as a function of water/cement factor, lightweight aggregate volume, cement quantity and density of the lightweight aggregate. In order to choose the best configuration of each method, a methodology was defined using the Particle Swarm Optmization (PSO) algorithm. Finally, the generalization of the methods through the cross validation process is verified in order to find the method that presents the best performance in the approximation of the two mechanical properties.

With the advent of smart grids and the higher level of requirement for the quality of supply, restoration of distribution networks is a timely topic that deserves to be revisited. In this paper, the concept of multi-agent automation in smart grids is applied to build a self-healing framework to be used for the restoration service. The optimal restoration solution is found solving a global optimization problem in a distributed way. The restoration problem including power flow equations is modeled as a second order cone programming problem and carried out on a 70-bus distribution network. The model is implemented in Matlab/Yalmip toolbox and solved using Gurobi solver.

As lipases (EC 3.1.1.3) hidrolisam ligações éster de triacilgriceróis numa interface água-óleo. Por possuírem diversas aplicações industriais, são as enzimas mais utilizadas em sínteses orgânicas e mais de 20% de biotransformações são realizadas com lipases. Os objetivos deste trabalho foram avaliar as condições nutricionais de meios de cultivo (fontes de óleos vegetais e animais, nitrogênio e carbono) para melhorar a produção de lipase por Candida viswanathii além de purificar a enzima em sistema aquoso bifásico e realizar a caracterização parcial da lipase. Entre os meios de cultivos analisados o meio de Vogel proporcionou a maior produção de lipase após 36 horas de cultivo (11,73 U/mL). Após analisar as fontes de carbono lipídicas e não-lipídicas e fontes de nitrogênio, um planejamento experimental de misturas foi aplicado para analisar o efeito de glicerol, lactose e sorbitol sobre a produção de lipase. Os resultados obtidos mostraram que a maior produção de lipase (20,41 U/mL) foi observada com 0,5% de lactose (p/v) e 0,5% (p/v) de sorbitol, sendo observado um aumento de 74% da produção de enzima. Nesse estudo observou-se que a lactose e sorbitol foram utilizados como adjuntos para a produção de lipase. A lipase produzida nas condições anteriores foi purificada em sistema aquoso bifásico (SAB) formado por polietilenoglicol (PEG) e fosfato de potássio. O maior coeficiente de partição (1,34) foi encontrado no ensaio com PEG 4000/fosfato a Temperatura de 40 °C e pH 7,0, bem como o maior balanço de atividade (50,73%). Na caracterização parcial, a lipase não sofreu influência da temperaturana faixa de 20 – 60 °C e obteve atividade máxima em pH 8,1. A enzima apresentou também elevada estabilidade em solventes orgânicos como metanol e etanol, sendo estas propriedades consideradas importantes para aplicações em processos biotecnológicos. Lipases (EC 3.1.1.3) hydrolyze ester bonds of triacylglycerols at a water-oil interface. By they have several industrial applications, they are the enzymes most used in organic synthesis and more than 20% of biotransformations are performed with lipases. The objectives of this work were to evaluate the nutritional conditions of culture media (plant and animal oils, nitrogen and carbon) to improve the production of lipase by Candida viswanathii in addition to purifying the enzyme in a biphasic aqueous system and to realize the partial characterization of lipase. Among the culture media analyzed, the Vogel medium provided the highest lipase production after 36 hours of culture (11,73 U/mL). After analyzing the lipid and non-lipid carbon sources and nitrogen sources, an experimental design of blends was applied to analyze the effect of glycerol, lactose and sorbitol on lipase production. The results showed that the highest production of lipase (20,41 U/mL) was observed with 0.5% lactose (w/v) and 0.5% (w/v) sorbitol, with an increase of 74% of enzyme production. In this study it was observed that lactose and sorbitol were used as adjuncts for the production of lipase. The lipase produced under the above conditions was purified in a aqueous two-phase systems (ATPS) consisting of polyethylene glycol (PEG) and potassium phosphate. The highest partition coefficient (1,34) was found in the PEG 4000/phosphate assay at temperature of 40 °C and pH 7,0, as well as the highest activity balance (50,73%). In the partial characterization, the lipase was not influenced by the temperature range of 20 - 60 °C and obtained maximum activity at pH 8,1. The enzyme also showed high stability in organic solvents such as methanol and ethanol, and these properties are considered important for applications in biotechnological processes.

This paper proposes a method of estimating area leakages in virtual areas of a water network to prioritize leak surveys for the areas. It estimates the leakage within the areas by minimizing the difference between the measured and the hydraulic-model-estimated pressure and flow rates. Leakage is distributed around junctions with risk models based on asset information around junctions such as pipe length. The accuracy was evaluated by a simulation study, where the impact of head measurement error was considered by a Monte-Carlo-type method. The mean absolute error of area leakage was less than 2 percent of the system flow.

New generation of high-resolution meteo-oceanographic predictions provides useful tools for routing of ships in short maritime distances. In this work, the optimal ship routing analysis is investigated in maritime routes in the Western Mediterranean Sea for relative short sea distances. A* algorithm is implemented in order to obtain the optimal path under an energetic wave events. The methodological aspects are presented including the grid description and weather prediction systems. The optimized cost function is the travel time and it is obtained considering the added resistance due to waves. A practical example considers the maritime route between Barcelona and Algiers. The results show the influence of this factor in the optimum path recovered by the algorithm. The relevance of the relative direction between wave and ship route is proven comparing the ship routing results.

This work deals with the development of a framework based on computational and optimization methods for algorithmic composition, more precisely, for the generation of guitar solos. The proposed approach was considered semiautomatic because it makes use of small melodic fragments (licks), previously created from human models. The solos generated are from the musical style Blues and they are applied over a well-known harmonic model called 12-Bar Blues. A licks database was created in which small instances containing a subset of them were randomly derived so as to diversify the possible candidates to be in the solo that will be generated. Once the instance is created, one solves an optimization problem that consists of determining the optimal sequence of a subset of licks by using a integer linear programming model. A set of rules was implemented for creating a matrix that de nes the transition cost between the licks. The outputs generated were stored in the MusicXML format and they can be read by most applications that provide support for this type of le and are capable of displaying it using the tablatures format. The solos created were evaluated by a sample of 173 subjects, classi ed as beginners, intermediates and professional musicians. A web application was developed to streamline the evaluation process. The results obtained show that the solos whose licks were optimally sequenced were statistically much better evaluated than those randomly sequenced, which indicates that the proposed methodology was capable of producing, on average, solos with a favorable percentage of acceptance. Este trabalho trata do desenvolvimento de um framework baseado em m etodos computacionais e de otimização para a composição algoritmica, mais especifi camente, para a geração de solos de guitarra. A abordagem proposta foi considerada semiautomatica pois faz uso de pequenos fragmentos mel odicos (licks) previamente criados a partir de modelos humanos. Os solos gerados possuem caracter sticas do estilo musical Blues e s~ao aplicados sobre um modelo de harmonia bastante conhecido denominado 12-Bar Blues. Um banco de dados de licks foi criado, do qual são realizados sorteios de instâncias menores do conjunto, diversifi cando os possíveis candidatos a estarem no solo a ser gerado. De posse da inst^ancia, um problema de otimiza c~ao, que consiste em sequenciar de forma otimizada um subconjunto de licks, e resolvido utilizando um modelo de programa c~ao linear inteira. Implementou-se um conjunto de regras para a criação de uma matriz que de ne o custo de transição entre os licks. As sa das geradas s~ao armazenadas no formato MusicXML e podem ser lidas pela maioria dos aplicativos que possuam suporte a esse tipo de arquivo e disponibilizem visualiza c~ao no formato de tablaturas. Os solos criados foram avaliados por uma amostra de 173 indiv duos, classi cados como m usicos iniciantes, intermedi arios e pro fissionais. Uma aplicação web foi desenvolvida para agilizar o processo de avaliação. Os resultados obtidos demonstram que os solos cujos licks foram sequenciados de forma otimizada foram estatisticamente mais bem avaliados que aqueles sequenciados aleatoriamente, indicando que a metodologia proposta foi capaz de produzir, em m edia, solos com percentual de aceitação favorável.

[ES] El presente trabajo trata sobre la búsqueda de los parámetros idóneos para la optimización de la fabricación de prototipos a través del proceso de impresión 3D mediante el uso del material ABS. Para ello, se han seleccionado las variables que se consideran más significativas y se han establecido una serie de parámetros para valorar la pieza obtenida. Se ha realizado un diseño de experimentos a dos niveles que consta de 64 ensayos aleatorios para analizar cómo afectan las variables a los parámetros., [EN] The present work deals with the search for suitable parameters for optimizing prototyping via 3D printing using ABS material. To do this, we selected the variables that are considered significant and stablished a number o parameters to evaluate the product obtained. Has made a design of experiments on two levels consisting of 64 randomize trials using the Stat Graphics program to analize how variables affect the stablished parameters., [CA] Aquest treball tracta sobre la busca dels paràmetres idonis per a l’optimització de la fabricació de prototips a través de la impressio 3D per mitjà de l’ús del material ABS. Per tal d’aconseguir-ho, s’han seleccionat les variables que es consideren més significatives i s’han establit una sèrie de paràmetres per a valorar la peça obtinguda. S’ha realitzat un disseny d’experiments a dos nivells que consta de 64 assajos aleatoris per mitjà del programa Stat Graphics per a analitzar com afecten les variables als paràmetres establits.

This paper was presented at the 3rd Micro and Nano Flows Conference (MNF2011), which was held at the Makedonia Palace Hotel, Thessaloniki in Greece. The conference was organised by Brunel University and supported by the Italian Union of Thermofluiddynamics, Aristotle University of Thessaloniki, University of Thessaly, IPEM, the Process Intensification Network, the Institution of Mechanical Engineers, the Heat Transfer Society, HEXAG - the Heat Exchange Action Group, and the Energy Institute. Thrust of the divergent part of axially symmetric micronozzle is under the study. It’s input to total thrust is considered by means of analysis of relative thrust determined as divergent part’s thrust related to nozzle’s thrust without divergent part when gas issuing into vacuum. An inviscid one-dimensional flow is used as start condition of analysis. For this case, it is shown from conservation laws that divergent part of infinite length has finite relative thrust depending only on sort of issuing gas. Analysis of the influence of shear stress on thrust of divergent part with the use of theory of laminar boundary layer shows that optimal nozzle wall angle at the exit increases and optimal length decreases with decreasing of nozzle’s dimension in comparison with initial inviscid case. This conclusion is approved by results of numerical simulation of flow inside nozzles with throat diameter 10 micrometers and various form of divergent part based on Navier – Stokes equations with both no-slip and slip wall conditions. Detailed analysis of flow shows advantage of micronozzles with wall form far from traditionally used in “large” thrusters. This study is partially supported by the Russian Foundation for Basic Research (Project 11-08-00422-а).

In the present article we study the reconstruction problem for clusters of linear cracks inside a rectangular domain. The parameters to be reconstructed are the number of cracks and the size and slope of each defect. The scanning is performed by a single ultrasonic transducer placed at a certain boundary point. The input data, used for the reconstruction algorithm, is taken as measured oscillation amplitudes over an array of chosen boundary points. The proposed numerical algorithm is tested on some examples with multiple clusters of cracks whose position and geometry are known a priori.

Several methodological issues in setting up a calibration process for traffic microsimulation models are still unresolved. The influence of individual parameters of microscopic models on simulated traffic dynamics is also far from clear. To address those issues, the paper sets up a methodology based on the sensitivity analysis of traffic flow models (the one used here is AIMSUN). Sensitivity analysis was performed by means of a series of 30 analyses of variance. These were designed to evaluate the effect of parameters on the variance of the simulated outputs and to draw a general inference about ( a) the proper interval for the aggregation of measurements, ( b) the proper measure of performance (e.g., traffic counts versus speeds), ( c) the proper traffic measurement locations, and ( d) the subset of parameters to calibrate. The analysis allowed quantification of the effect of the single parameters on different traffic phases. For example, it was possible to quantify the extent of the influence of the parameter reaction time on simulated outputs in locations in which free flow, rather than congested conditions, occurs. The great differences between parameters in affecting the different traffic phases suggested that parameters are likely to be calibrated independently, that is, using data from different locations. The first evidence of the possibility of breaking the calibration problem into two subproblems is given. This entails great benefits in regard to computational time, given the exponential computational complexity of the calibration problem.

Este trabalho aborda a modelagem matemática do processo de extração de óleos essenciais por arraste a vapor e a otimização deste processo em escala industrial. Utilizouse dados experimentais de óleos essenciais de alecrim e citronela obtidos em uma unidade industrial de extração por arraste a vapor. O modelo é constituído de um sistema de equações diferenciais parciais, que representam o balanço de massa unidimensional nas fases vapor e sólido. Para resolver o modelo desenvolvido, utilizouse o método das linhas com o uso da técnica de diferenças finitas para discretizar a coordenada espacial e suas derivadas, resultando em um sistema de equações diferenciais ordinárias implementado no simulador EMSO. Os parâmetros do modelo foram estimados pelo método dos poliedros flexíveis, utilizando dois conjuntos de dados experimentais e um terceiro para sua validação. O modelo ajustou-se bem aos dados experimentais industriais, resultando, para duas condições ambientais e geográficas das plantas aromáticas de alecrim, em coeficiente global de transferência de massa de 8,64 x 10 -4 s -1 e 9,10 x 10 -4 s -1, e a constante de equilíbrio de 2,72 x 10 -4 m3=kg e 1,07 x 10 -2m3/kg. Enquanto que para a citronela obteve-se coeficiente global de transferência de massa de 8,29 x 10 -4s -1 e 1,23 x 10 -3 s -1, e constante de equilíbrio de 1,25 x 10 -2 m3/kg e 1,52 x 10 -3 m3/kg. Foi proposto também um problema de otimização para determinar o tempo ótimo de extração e o número de bateladas por dia. Os resultados da otimização do processo indicam que se pode reduzir o tempo de extração, permitindo que outras extrações sejam realizadas por dia, maximizando o volume de óleo extraído ao fim do dia. This work has as objective the development of a mathematical model of essential oils extraction by steam distillation and the optimization of this process on an industrial scale. We used experimental data obtained from rosemary and citronella in an industrial steam distillation unit. The model consists a system of partial differential equations, which represent the one-dimensional mass balance in steam and solid phases. To solve the developed model, we used the method of lines with the finite differences technique to discretize the spatial coordinate and its derivatives, resulting in a system of ordinary differential equations implemented in the EMSO simulator. The model parameters were estimated by the method of flexible polyhedra, using two experimental runs and a third experimental run to test the model. The model fitted to the experimental data industry, resulting, for two environmental and geographical conditions of rosemary, in mass transfer coefficients of 8,64 x 10 -4 s -1 and 9,10 x 10 -4 s -1, and the equilibrium constant of 2,72 x 10 -4 m3/kg and 1,07 x 10 -2m3/kg. For citronella, the obtained mass transfer coefficient of 8,29 x 10 -4 s -1 and 1,23 x 10 -3 s -1 and equilibrium constant of 1,25 x 10 -2 m3/kg and 1,52 x 10 -3 m3/kg. It was also proposed an optimization problem to determine the optimum extraction time and number of batches per day. The optimization results indicate that the process can reduce the extraction time, allowing other extractions are performed each day, maximizing the essential oil volume extracted in the day.

The paper documents MINTOOLKIT for GNU Octave. MINTOOLKIT provides functions for minimization and numeric differentiation. The main algorithms are BFGS, LBFGS, and simulated annealing. Examples are given.

Special Issue Engineering of Configurable Systems; International audience; This paper presents an extension of our AAA rapid prototyping methodology for the optimized implementation ofreal-time applications onto reconfigurable circuits. This extension is based on an unified model of factorized datadependence graphs as well to specify the application algorihtm, as to deduce the possible implementations ontoreconfigurable hardware, in terms of graphs transformations. This transformation flow has been implemented inSynDEx, a system level CAD software tool.

The paper deals with the design of a composite arm, to be employed in the architecture of industrial, measuring robots. Based on the real working conditions, the arm is reduced to a cantilever beam, shaped as a thin walled tube, supporting a mass at the free end. An optimization study is carried out, aiming to determine the laminate properties capable of providing the maximum fundamental frequency (FF) in bending. The optimization relies on an approximate solution, obtained by modifying the Rayleigh energy method to account for the shear effect. It results in a closed-form formula, explicitly correlating FF with composite elastic properties. Two classes of laminates, namely (± θ)sand ( 0 ± 45n)s, are examined in order to assess their ability to fulfil the requirement of a maximum FF value. Moreover, the advantages and drawbacks of composites compared to metals are highlighted for the application under concern. Finally, the validity of the proposed solution is demonstrated by comparison with the results of a finite element analysis.