Ces travaux de thèse, réalisés dans une équipe concevant des circuits analogiques neuromimétiques suivant le modèle d’Hodgkin-Huxley, concernent la modélisation de neurones biologiques, plus précisément, l’estimation des paramètres de modèles de neurones. Une première partie de ce manuscrit s’attache à faire le lien entre la modélisation neuronale et l’optimisation. L’accent est mis sur le modèle d’Hodgkin- Huxley pour lequel il existait déjà une méthode d’extraction des paramètres associée à une technique de mesures électrophysiologiques (le voltage-clamp) mais dont les approximations successives rendaient impossible la détermination précise de certains paramètres. Nous proposons dans une seconde partie une méthode alternative d’estimation des paramètres du modèle d’Hodgkin-Huxley s’appuyant sur l’algorithme d’évolution différentielle et qui pallie les limitations de la méthode classique. Cette alternative permet d’estimer conjointement tous les paramètres d’un même canal ionique. Le troisième chapitre est divisé en trois sections. Dans les deux premières, nous appliquons notre nouvelle technique à l’estimation des paramètres du même modèle à partir de données biologiques, puis développons un protocole automatisé de réglage de circuits neuromimétiques, canal ionique par canal ionique. La troisième section présente une méthode d’estimation des paramètres à partir d’enregistrements de la tension de membrane d’un neurone, données dont l’acquisition est plus aisée que celle des courants ioniques. Le quatrième et dernier chapitre, quant à lui, est une ouverture vers l’utilisation de petits réseaux d’une centaine de neurones électroniques : nous réalisons une étude logicielle de l’influence des propriétés intrinsèques de la cellule sur le comportement global du réseau dans le cadre des oscillations gamma. / These works, which were conducted in a research group designing neuromimetic integrated circuits based on the Hodgkin-Huxley model, deal with the parameter estimation of biological neuron models. The first part of the manuscript tries to bridge the gap between neuron modeling and optimization. We focus our interest on the Hodgkin-Huxley model because it is used in the group. There already existed an estimation method associated to the voltage-clamp technique. Nevertheless, this classical estimation method does not allow to extract precisely all parameters of the model, so in the second part, we propose an alternative method to jointly estimate all parameters of one ionic channel avoiding the usual approximations. This method is based on the differential evolution algorithm. The third chaper is divided into three sections : the first two sections present the application of our new estimation method to two different problems, model fitting from biological data and development of an automated tuning of neuromimetic chips. In the third section, we propose an estimation technique using only membrane voltage recordings – easier to mesure than ionic currents. Finally, the fourth and last chapter is a theoretical study preparing the implementation of small neural networks on neuromimetic chips. More specifically, we try to study the influence of cellular intrinsic properties on the global behavior of a neural network in the context of gamma oscillations.

Orientador: Paulo Morelato França / Tese (doutorado) - Universidade Estadual de Campinas, Faculdade de Engenharia Eletrica e de Computação / Made available in DSpace on 2018-08-05T20:11:19Z (GMT). No. of bitstreams: 1 Garcia_ViniciusJacques_D.pdf: 1756755 bytes, checksum: e845cc09a5de807da958e9792684e777 (MD5) Previous issue date: 2005 / Resumo: Depois da regulamentação do setor elétrico brasileiro, a qualidade no fornecimento de energia ganhou maior importância por parte das concessionárias. Neste contexto, o problema de restauração do serviço tem particular relevância pela relação com a freqüência e duração das interrupções no fornecimento: através de alterações na configuração original da rede, busca-se reduzir a carga não atendida sem deixar de observar as restrições de capacidade dos alimentadores, de queda de tensão nas barras de carga e de radialidade da rede. Considerando o caráter temporário destas manobras, torna-se desejável reduzir o grau de intervenção de modo a facilitar a restauração da configuração original. Nesta tese é considerado o problema multiobjetivo de restauração do serviço que compreende a minimização da carga sem fornecimento e do número de chaves manipuladas. Depois da definição matemática do problema, da revisão da literatura especializada e da descrição de um "framework" para problemas relacionados, são descritas duas heurísticas, uma construtiva e outra de melhoramento. A seguir, apresentam-se duas metaheurísticas para o problema, uma Busca Tabu e um Algoritmo Evolutivo, ambas baseadas em otimização de Pareto. Por fim, por meio de estudos práticos com sistemas de distribuição brasileiros, avalia-se experimentalmente a aplicabilidade das abordagens propostas / Abstract: After the Brazilian electric power market regulation, quality of service became a crucial concern of utilities. In fact, the service restoration has a particular importance since it is closely related to frequency and duration of service interruption: through network reconfigurations, one aims to reduce the non supplied load while respecting constraints like feeder and voltage limits as well as the maintenance of a radial structure. Considering that this emergency state is transitory existing only until the fault is eliminated, it is convenient to reduce the number of switching operations in order to make the return back to the original configuration easy. This work considers the multiobjective service restoration to minimize both the load not supplied and the number of switching operations. After defining the mathematical formulation proposed and presenting the bibliographical survey with the description of a new framework to related problems, two new heuristics are presented, one for constructive search and another one for neighborhood search. Next, two metaheuristics especially developed for the referred problem are described, both based on Pareto optimization. Finally, the effectiveness of these proposed methods are proved in a set of five systems, three of them referring to actual Brazilian systems / Doutorado / Automação / Doutor em Engenharia Elétrica

Orientadores: João Marcos Travassos Romano, Romis Ribeiro de Faissol Attux / Dissertação (mestrado) - Universidade Estadual de Campinas, Faculdade de Engenharia Eletrica e de Computação / Made available in DSpace on 2018-08-12T15:14:54Z (GMT). No. of bitstreams: 1 Dias_TiagoMacedo_M.pdf: 3219855 bytes, checksum: 5572e53d65cb457f420e78b3150dd6ee (MD5) Previous issue date: 2008 / Resumo: Esta dissertação se propõe a estudar um novo método para separação cega de fontes baseado no modelo Post-Nonlinear, que une uma ferramenta de busca global baseada em computação bioinspirada a uma etapa de busca local conduzida pelo algoritmo FastICA. A idéia subjacente à proposta é procurar obter soluções precisas e eficientes usando de maneira parcimoniosa os recursos computacionais disponíveis. A nova proposta foi testada em diferentes cenários, e, em todos os casos, estabeleceram-se comparações com uma abordagem alternativa, cujo passo de otimização não inclui o estágio de busca local (ou "memética"). Os resultados obtidos por meio de simulações indicam que um bom compromisso entre desempenho e custo computacional foi, de fato, atingido. / Resumo: Esta dissertação se propõe a estudar um novo método para separação cega de fontes baseado no modelo Post-Nonlinear, que une uma ferramenta de busca global baseada em computação bioinspirada a uma etapa de busca local conduzida pelo algoritmo FastICA. A idéia subjacente à proposta é procurar obter soluções precisas e eficientes usando de maneira parcimoniosa os recursos computacionais disponíveis. A nova proposta foi testada em diferentes cenários, e, em todos os casos, estabeleceram-se comparações com uma abordagem alternativa, cujo passo de otimização não inclui o estágio de busca local (ou "memética"). Os resultados obtidos por meio de simulações indicam que um bom compromisso entre desempenho e custo computacional foi, de fato, atingido. / Abstract: This work deals with a new method for source separation of Post-Nonlinear mixtures that brings together an evolutionary-based global search and a local search step based on the FastICA algorithm. The rationale of the proposal is to attempt to obtain efficient and precise solutions using with parsimony the available computational resources. The new proposal was tested in different scenarios and, in all cases, we attempted to establish grounds for comparison with an alternative approach whose optimization step does not include the local (memetic) search stage. Simulation results indicate that a good tradeoff between performance and computational cost was indeed reached. / Abstract: This work deals with a new method for source separation of Post-Nonlinear mixtures that brings together an evolutionary-based global search and a local search step based on the FastICA algorithm. The rationale of the proposal is to attempt to obtain efficient and precise solutions using with parsimony the available computational resources. The new proposal was tested in different scenarios and, in all cases, we attempted to establish grounds for comparison with an alternative approach whose optimization step does not include the local (memetic) search stage. Simulation results indicate that a good tradeoff between performance and computational cost was indeed reached. / Mestrado / Telecomunicações e Telemática / Mestre em Engenharia Elétrica

Submitted by Geandra Rodrigues (geandrar@gmail.com) on 2017-12-22T14:54:33Z No. of bitstreams: 1 patriciasilvaneves.pdf: 1941458 bytes, checksum: 16ab3b743d0b75134d320f08de292905 (MD5) / Approved for entry into archive by Adriana Oliveira (adriana.oliveira@ufjf.edu.br) on 2018-01-22T18:33:39Z (GMT) No. of bitstreams: 1 patriciasilvaneves.pdf: 1941458 bytes, checksum: 16ab3b743d0b75134d320f08de292905 (MD5) / Made available in DSpace on 2018-01-22T18:33:39Z (GMT). No. of bitstreams: 1 patriciasilvaneves.pdf: 1941458 bytes, checksum: 16ab3b743d0b75134d320f08de292905 (MD5) Previous issue date: 2017-08-31 / O presente trabalho apresenta a aplicação conjunta de uma técnica de otimização bioinspirada e de um Algoritmo Heurístico Construtivo (AHC) na resolução do problema de planejamento estático da expansão de sistemas de transmissão de energia elétrica. O algoritmo bioinspirado utilizado é uma versão modificada do Flower Pollination Algorithm (FPA), no qual foi introduzido o operador de seleção clonal, oriundo do Algoritmo de Seleção Clonal (CLONALG), com o objetivo de potencializar o processo de busca local do FPA. A versão modificada proposta neste trabalho foi nomeada de Clonal Flower Pollination Algorithm (CFPA). O CFPA realiza a otimização da expansão de sistemas de transmissão de energia elétrica, determinando, entre um conjunto de linhas (circuitos) de transmissão previamente definidas, quais devem ser construídas de modo a minimizar os custos de investimento e de operação do sistema elétrico, suprindo a demanda prevista para um dado horizonte de planejamento. De modo a aumentar a eficiência do processo de busca pelo CFPA, fez-se o uso de informações provenientes de um Algoritmo Heurístico Construtivo. Tais informações heurísticas são utilizadas na inicialização do CFPA e também na seleção de um conjunto reduzido das rotas mais relevantes à expansão, reduzindo o espaço de busca. Para aferir os resultados da metodologia proposta foram simulados os sistemas Garver, IEEE 24 Barras e o equivalente da região Sul do Brasil. Diante dos resultados, pode-se verificar que tanto a inclusão do operador de seleção clonal quanto as informações heurísticas foram capazes de aumentar a eficiência do FPA na resolução do problema aqui em estudo. / This work presents the application of a bio-inspired algorithm, together with a Heuristic Constructive Algorithm (HCA) in the solution of a power system static transmission expansion planning problem. The algorithm used is a modified version of the Flower Pollination Algorithm (FPA) that includes a clonal selection operator, from the clonal selection algorithm (CLONALG) that aims to improve the FPA local search process. The modified version proposed is entitled Clonal Flower Pollination Algorithm (CFPA). The CFPA realizes the power system transmission expansion planning, that is, it determines between a set of predefined transmission lines (circuits), which of them must be constructed in order to minimize the power systems investments and operation costs, while meeting the forecast demand in a given planning horizon. In order to increase the efficiency of the search process by the CFPA, information from an HCA has been utilized. That heuristic information has been used in the initialization process of the CFPA and also in the selection of a reduced set of most relevant lines candidates to the expansion plan, thus reducing the search space. To evaluate the results of the proposed methodology, the Garver, IEEE 24 Buses and South Brazilian Systems were simulated. Considering the results it can be verified that both the inclusion of the clonal selection algorithm and the heuristic information were able to increase the efficiency of the FPA in solving this problem.

Made available in DSpace on 2014-07-29T14:57:48Z (GMT). No. of bitstreams: 1 Dissertacao Carine Rodrigues da Costa.pdf: 991478 bytes, checksum: 516faf301aac129df1d69068892a5ea9 (MD5) Previous issue date: 2011-03-30 / The necessity of solving optimization problems in a reasonable computational time limit makes the development of heuristics be a large research area. Usually, developed heuristics for optimization problems are empirically evaluated by its application to a set of specific instances, comparing to quality solution and computational efforts. Besides, when presenting a new heuristic, the contributions should be scientifically evaluated and reported in an objective way. The quality of a computational experiment report may become evident the difficulty to reproduce the experiment or compare the results with those of other experiments. Part of the origin of these issues comes from the fact that there is no standard for reporting experiments in Computer Science. Therefore, the focus of this work is to investigate methods of conducting experimental research with heuristics, to examine what methods are more favorable and consistent in evaluating these. Thus, the investigation resulted in a compilation with contribution of several authors, which consisted in identifying a set of recommendations, including the formulation of a checklist representing the summary form of all the items that were seen in this study. The results of this review served as the basis for definitining the research and leading a sample study, which consisted in analysis of articles that deal with the Quadratic Assignment Problem (QAP), by checking the necessary items for understanding, reproduction and comparison of the performed experiments. / A necessidade de resolver problemas de otimização em um limite razoável de tempo computacional faz com que o desenvolvimento de heurísticas seja uma grande área de pesquisa. Usualmente, heurísticas desenvolvidas para problemas de otimização são avaliadas empiricamente, pela sua aplicação a um conjunto de instâncias específicas, comparando qualidade da solução e esforços computacionais. Além disso, ao se apresentar uma nova heurística, as contribuições devem ser avaliadas cientificamente e relatadas de uma maneira objetiva. Ao descrever um experimento computacional e relatar os resultados obtidos do mesmo, pode ficar evidente a dificuldade de reproduzir o experimento ou comparar os resultados obtidos com os de outros experimentos. Parte da origem dessas questões vem do fato de que não há padrão para o relato de experimentos na área de Computação. Portanto, o foco deste trabalho é investigar métodos de condução de pesquisa experimental com heurísticas, para analisar quais são os mais favoráveis e consistentes na avaliação destas. Desta forma, a investigação resultou em uma compilação com a contribuição de diversos autores, em que consistiu na identificação de um conjunto de recomendações, com a elaboração de um checklist, representando de forma sumarizada todos os itens vistos nesta pesquisa. Os resultados dessa revisão serviram como base para a definição da pesquisa e condução de um estudo exemplo, que consistiu na análise de artigos que tratam do Problema de Atribuição Quadrática (PAQ), com a verificação dos itens necessários para compreensão, reprodução e comparação dos experimentos realizados.

L’évolution continue des environnements de production et l’augmentation des besoins des clients, demandent un processus de production plus rapide et efficace qui contrôle plusieurs paramètres en même temps. Nous nous sommes intéressés au développement de méthodes d’aide à la décision qui permettent d’améliorer l’ordonnancement de la production. L’entreprise partenaire (Norelem) fabrique des pièces de précision mécanique, il faut donc prendre en compte les différentes contraintes de ressources (humaines et d’outillage) existantes dans l’atelier de production.Nous avons abordé l’étude d’un atelier d’ordonnancement de type open shop ou chemin ouvert, où une tâche peut avoir de multiples séquences de production puisque l’ordre de fabrication n’est pas fixé et l’objectif à minimiser est le temps total de séjour. Des contraintes d’affectation de ressources humaines (multi-compétences) et de disponibilité d’outillage ont été prises en compte.Des modèles mathématiques linéaires et non-linéaires ont été développés pour décrire la problématique. Etant donné que les méthodes exactes sont limitées aux instances de petites tailles à cause des temps de calcul, des méthodes de résolution approchées ont été proposées et comparées. De plus, nous avons abordé l’optimisation multi-objectif en considérant trois objectifs, la minimisation du temps total de séjour et l’équilibrage de charge des ressources (humaines et machines).L’efficacité des méthodes est prouvée grâce à des tests sur des instances théoriques et l’application au cas réel / The continuous evolution of manufacturing environments and the growing of customer needings, leads to a faster and more efficient production process that controls an increasing number of parameters. This thesis is focused on the development of decision making methods in order to improve the production scheduling. The industrial partner (Norelem) produces standardized mechanical elements, so many different resource constraints (humans and tools) are presented in its workshop.We study an open shop scheduling problem where one job can follow multiple production sequences because there is no fixed production sequence and the objective function is to minimize the total flow time. In addition, multi-skilled personnel assignment and tool’s availability constraints are involved.Mathematical models: linear and non-linear formulations have been developed to describe the problem. Knowing the exact method limitations in terms of instance sizes because of the duration, heuristics methods have been proposed and compared. Besides that, the multi-objective optimization was exposed to deal with three objectives as total flow time minimization and workload balancing concerning both, humans and machines.The efficiency of these methods was proved by several theoretical instance tests and the application on the real industrial case

Cette thèse, réalisée en collaboration avec Sagem Défense Sécurité, porte sur l'élaboration d'une stratégie de recherche efficace pour la résolution de problèmes de planification d'itinéraires de véhicules. Nous considérons ici en particulier les problèmes de planification avec contraintes de points de passage et de "capacité" (énergie, bande passante radio) appliquées au véhicule. Ce document propose une approche originale, hybridant un algorithme de colonies de fourmis avec un solveur de Programmation par Contraintes existant. Le premier est utilisé pour résoudre rapidement une version relaxée du problème. La solution partielle obtenue est alors employée pour guider la recherche du second, par le biais d'une méthode de sonde, vers les zones les plus prometteuses de l'espace d'état. Cette approche permet ainsi de combiner la rapidité des métaheuristiques et la complétude de la programmation par contraintes. Nous montrons expérimentalement que cette approche satisfait les exigences pour une utilisation du planificateur dans un cadre embarqué. / This thesis, led in collaboration with Sagem Defence & Security, focuses on defining an efficient search strategy to solve vehicle path planning problems. This work addresses more precisely planning problems in which waypoints and "capacity" constraints (energy, radio bandwidth) are applied to vehicles.This document proposes an original approach, mixing an Ant Colony algorithm with an existing Constraint Programming solver. The former is used to fastly solve a relaxed version of the problem. The partial solution returned is then employed to guide the search of the latter, through a Probe Backtrack mechanism, towards the most promising areas of the state space. This approach allows to combine the metaheuristics solving fastness and the Constraint Programming completeness. We experimentally show that this approach meets the requirements for an on-line use of the planner.

En 2014, dans le cadre du Plan Climat, les pays membres de l’Union Européenne, se sont engagés à réduire de près de 27% leur consommation d’énergie. L’un des axes d’études concerne l’augmentation de l’efficacité énergétique des transports urbains. Cette thèse a pour objectif de proposer une méthodologie afin de réduire la consommation énergétique de lignes de métro automatique tout en intégrant les perturbations de trafic qui se produisent dans des conditions normales d’exploitation. Le principe retenu dans ces travaux est de maximiser la réutilisation de l’énergie générée lors du freinage des trains, par les autres trains présents sur la ligne. Une première partie est dédiée à la modélisation électrique d’une ligne de métro automatique et à la présentation de méthodes permettant de calculer les flux de puissances entre les trains et les sous-stations d’alimentation. Ensuite, des algorithmes d’optimisation sont introduits pour effectuer l’optimisation des paramètres d’exploitation les plus influents dans une configuration idéale n’intégrant pas les aléas de trafic. Enfin, une méthodologie basée sur un apprentissage des données de simulation est développée dans le but de réaliser l’optimisation énergétique de la consommation en temps réel et en intégrant les perturbations de trafic. Cette dernière partie aura ainsi pour objectif de fournir une aide à la décision dans le choix des temps d’arrêts que doivent effectuer chaque train en station afin de maximiser la récupération de l’énergie issue du freinage. / In 2014, as part of the Climate Plan, EU member countries have committed to reduce by 27% their energy consumption. One of the main focal areas consists in increasing the energy efficiency of urban transports. This thesis aims to propose a methodology to reduce the energy consumption of automatic metro lines while integrating traffic disruptions that occur under normal operating conditions. The principle adopted in this work is to maximize the reuse of electrical energy generated during braking of the train, by other trains running on the line. First part is dedicated to the electrical modeling of an automatic metro line and development of methods to calculate power flows between trains and power substations. Then, optimization algorithms are introduced to perform optimization of the most influential operating parameters in an ideal configuration ignoring traffic fluctuations. Finally, a methodology based on learning simulation data is developed in order to achieve optimization of energy consumption integrating traffic disruptions in real time. This last part will thus purchase the objective to provide a decision support to determine optimal dwell times to be carried out by trains in each station, so as to maximize braking energy recovery.

Cette thèse présente des algorithmes de résolution de quatre problèmes de clique : clique de poids maximum (MVWCP), s-plex maximum (MsPlex), clique maximum équilibrée dans un graphe biparti (MBBP) et clique partition (CPP). Les trois premiers problèmes sont des généralisations ou relaxations du problème de la clique maximum, tandis que le dernier est un problème de couverture. Ces problèmes, ayant de nombreuses applications pratiques, sont NP-difficiles, rendant leur résolution ardue dans le cas général. Nous présentons ici des algorithmes de recherche locale, principalement basés sur la recherche tabou, permettant de traiter efficacement ces problèmes ; chacun de ces algorithmes emploie des composants originaux et spécifiquement adaptés aux problèmes traités, comme de nouveaux opérateurs ou mécanismes perturbatifs. Nous y intégrons également des stratégies telles que la réduction de graphe ou la propagation afin de traiter des réseaux de plus grande taille. Des expérimentations basées sur des jeux d’instances nombreux et variés permettent de montrer la compétitivité de nos algorithmes en comparaison avec les autres stratégies existantes. / This thesis considers four clique problems: the maximum vertex weight clique problem (MVWCP), the maximum s-plex problem (MsPlex), the maximum balanced biclique problem (MBBP) and the clique partitioning problem (CPP). The first three are generalization and relaxation of the classic maximum clique problem (MCP), while the last problem belongs to a clique grouping problem. These combinatorial problems have numerous practical applications. Given that they all belong to the NP-Hard family, it is computationally difficult to solve them in the general case. For this reason, this thesis is devoted to develop effective algorithms to tackle these challenging problems. Specifically, we propose two restart tabu search algorithms based on a generalized PUSH operator for MVWCP, a frequency driven local search algorithms for MsPlex, a graph reduction based tabu search as well as effective exact branch and bound algorithms for MBBP and lastly, a three phase local search algorithm for CPP. In addition to the design of efficient move operators for local search algorithms, we also integrate components like graph reduction or upper bound propagation in order to deal deal with very large real-life networks. The experimental tests on a wide range of instances show that our algorithms compete favorably with the main state-of-the-art algorithms.

Nos travaux concernent la mise en œuvre de méthodologies pour la résolution de problèmes d’ordonnancement en industries agroalimentaires. Trois nouvelles approches basées sur les algorithmes génétiques, sont proposées pour la résolution de problèmes d’ordonnancement multi-objectifs : les algorithmes génétiques séquentiels (SGA), les algorithmes génétiques parallèles (PGA) et les algorithmes génétiques parallèles séquentiels (PSGA). Deux approches coopératives multi-objectifs en mode relais, SH_GA/TS et SH_GA/SA, hybridant toutes les deux des métaheuristiques de haut niveau, sont par la suite proposées. Un algorithme évolutionnaire et un algorithme de recherche locale sont, dans ce cas exécutés séquentiellement / The purpose of our works is the implementation of methodologies for the resolution of the agro-food industry scheduling problem. Three new approaches based on genetic algorithms are proposed to solve multi-objectives scheduling problems: sequential genetic algorithms (SGA), parallel genetic algorithms (PGA) and parallel sequential genetic algorithms (PSGA). Two high-level hybrid algorithms, SH_GA/TS et SH_GA/SA, are also proposed. The purpose in this hybridization is to benefit the exploration of the solution space by a population of individuals with the exploitation of solutions through a smart search of the local search algorithm