Submitted by Fabio Sobreira Campos da Costa (fabio.sobreira@ufpe.br) on 2016-09-20T13:03:02Z No. of bitstreams: 2 license_rdf: 1232 bytes, checksum: 66e71c371cc565284e70f40736c94386 (MD5) Dissertação_Versão_Digital_REIGA.pdf: 3705889 bytes, checksum: 551e1d47969ce5d1aa92cdb311f41304 (MD5) / Made available in DSpace on 2016-09-20T13:03:02Z (GMT). No. of bitstreams: 2 license_rdf: 1232 bytes, checksum: 66e71c371cc565284e70f40736c94386 (MD5) Dissertação_Versão_Digital_REIGA.pdf: 3705889 bytes, checksum: 551e1d47969ce5d1aa92cdb311f41304 (MD5) Previous issue date: 2016-02-23 / CAPES / A Tomografia por Impedância Elétrica (TIE) é uma técnica que visa reconstruir imagens do interior de um corpo de forma não-invasiva e não-destrutiva. Com base na aplicação de corrente elétrica e na medição dos potenciais de borda do corpo, feita através de eletrodos, um algoritmo de reconstrução de imagens de TIE gera o mapa de condutividade elétrica do interior deste corpo. Diversos métodos são aplicados para gerar imagens de TIE, porém ainda são geradas imagens de contorno suave. Isto acontece devido à natureza matemática do problema de reconstrução da TIE como um problema mal-posto e mal-condicionado. Isto significa que não existe uma distribuição de condutividade interna exata para uma determinada distribuição de potenciais de borda. A TIE é governada matematicamente pela equação de Poisson e a geração da imagem envolve a resolução iterativa de um problema direto, que trata da obtenção dos potenciais de borda a partir de uma distribuição interna de condutividade. O problema direto, neste trabalho, foi aplicado através do Método dos Elementos Finitos. Desta forma, é possível aplicar técnicas de busca e otimização que objetivam minimizar o erro médio quadrático relativo (função objetivo) entre os potenciais de borda mensurados no corpo (imagem ouro) e os potencias gerados pela resolução do problema direto de um candidato à solução. Assim, o objetivo deste trabalho foi construir uma ferramenta computacional baseada em algoritmos de busca e otimização híbridos, com destaque para a Evolução Diferencial, a fim de reconstruir imagens de TIE. Para efeitos de comparação também foram utilizados para gerar imagens de TIE: Algoritmos Genéticos, Otimização por Enxame de Partículas e Recozimento Simulado. As simulações foram feitas no EIDORS, uma ferramenta usada em MatLab/ GNU Octave com código aberto voltada para a comunidade de TIE. Os experimentos foram feitos utilizando três diferentes configurações de imagens ouro (fantomas). As análises foram feitas de duas formas, sendo elas, qualitativa: na forma de o quão as imagens geradas pela técnica de otimização são parecidas com seu respectivo fantoma; quantitativa: tempo computacional, através da evolução do erro relativo calculado pela função objetivo do melhor candidato à solução ao longo do tempo de reconstrução das imagens de TIE; e custo computacional, através da avaliação da evolução do erro relativo ao longo da quantidade de cálculos da função objetivo pelo algoritmo. Foram gerados resultados para Algoritmos Genéticos, cinco versões clássicas de Evolução Diferencial, versão modificada de Evolução Diferencial, Otimização por Enxame de Partículas, Recozimento Simulado e três novas técnicas híbridas baseadas em Evolução Diferencial propostas neste trabalho. De acordo com os resultados obtidos, vemos que todas as técnicas híbridas foram eficientes para resolução do problema da TIE, obtendo bons resultados qualitativos e quantitativos desde 50 iterações destes algoritmos. Porém, merece destacar o rendimento do algoritmo obtido pela hibridização da Evolução Diferencial e Recozimento Simulado por ser a técnica aqui proposta mais promissora na reconstrução de imagens de TIE, onde mostrou ser mais rápida e menos custosa computacionalmente do que as outras técnicas propostas. Os resultados desta pesquisa geraram diversas contribuições na forma de artigos publicados em eventos nacionais e internacionais. / Electrical Impedance Tomography (EIT) is a technique that aim to reconstruct images of the interior of a body in a non-invasive and non-destructive form. Based on the application of the electrical current and on the measurement of the body’s edge electrical potential, made through of electrodes, an EIT image reconstruction algorithm generates the conductivity distribution map of this body’s interior. Several methods are applied to generate EIT images; however, they are still generated smooth contour images. This is due of the mathematical nature of EIT reconstruction problem as an ill-posed and ill-conditioned problem. Thus, there is not an exact internal conductivity distribution for one determinate edge potential distribution. The EIT is ruled mathematically by Poisson’s equations, and the image generation involves an iterative resolution of a direct problem, that treats the obtainment of the edge potentials through of an internal distribution of conductivity. The direct problem, in this dissertation, was applied through of Finite Elements Method. Thereby, is possible to apply search and optimization techniques that aim to minimize the mean square error relative (objective function) between the edge potentials measured in the body (gold image) and the potential generated by the resolution of the direct problem of a solution candidate. Thus, the goal of this work was to construct a computational tool based in hybrid search and optimization algorithms, highlighting the Differential Evolution, in order to reconstruct EIT images. For comparison, it was also used to generate EIT images: Genetic Algorithm, Particle Optimization Swarm and Simulated Annealing. The simulations were made in EIDORS, a tool used in MatLab/GNU Octave open source toward the TIE community. The experiments were performed using three different configurations of gold images (phantoms). The analyzes were done in two ways, as follows, qualitative: in the form of how the images generated by the optimization technique are similar to their respective phantom; quantitative: computational time, by the evolution of the relative error calculated for the objective function of the best candidate to the solution over time the EIT images reconstruction; and computational cost, by evaluating the evolution of the relative error over the amount of calculations of the objective functions by the algorithm. Results were generated for Genetic Algorithms, five classical versions of Differential Evolution, modified version of the Differential Evolution, Particle Optimization Swarm, Simulated Annealing and three new hybrid techniques based in Differential Evolution proposed in this work. According to the results obtained, we see that all hybrid techniques were efficient in solving the EIT problem, getting good qualitative and quantitative results from 50 iterations of these algorithms. Nevertheless, it deserves highlight the algorithm performance obtained by hybridization of Differential Evolution and Simulated Annealing to be the most promising technique here proposed to reconstruct EIT images, which proved to be faster and less expensive computationally than other proposed techniques. The results of this research generate several contributions in the form of published paper in national and international events.

Submitted by Fabio Sobreira Campos da Costa (fabio.sobreira@ufpe.br) on 2016-09-20T13:21:46Z No. of bitstreams: 2 license_rdf: 1232 bytes, checksum: 66e71c371cc565284e70f40736c94386 (MD5) Dissertação_Versão_Digital_REIGA.pdf: 3705889 bytes, checksum: 551e1d47969ce5d1aa92cdb311f41304 (MD5) / Made available in DSpace on 2016-09-20T13:21:46Z (GMT). No. of bitstreams: 2 license_rdf: 1232 bytes, checksum: 66e71c371cc565284e70f40736c94386 (MD5) Dissertação_Versão_Digital_REIGA.pdf: 3705889 bytes, checksum: 551e1d47969ce5d1aa92cdb311f41304 (MD5) Previous issue date: 2016-02-23 / CAPES / A Tomografia por Impedância Elétrica (TIE) é uma técnica que visa reconstruir imagens do interior de um corpo de forma não-invasiva e não-destrutiva. Com base na aplicação de corrente elétrica e na medição dos potenciais de borda do corpo, feita através de eletrodos, um algoritmo de reconstrução de imagens de TIE gera o mapa de condutividade elétrica do interior deste corpo. Diversos métodos são aplicados para gerar imagens de TIE, porém ainda são geradas imagens de contorno suave. Isto acontece devido à natureza matemática do problema de reconstrução da TIE como um problema mal-posto e mal-condicionado. Isto significa que não existe uma distribuição de condutividade interna exata para uma determinada distribuição de potenciais de borda. A TIE é governada matematicamente pela equação de Poisson e a geração da imagem envolve a resolução iterativa de um problema direto, que trata da obtenção dos potenciais de borda a partir de uma distribuição interna de condutividade. O problema direto, neste trabalho, foi aplicado através do Método dos Elementos Finitos. Desta forma, é possível aplicar técnicas de busca e otimização que objetivam minimizar o erro médio quadrático relativo (função objetivo) entre os potenciais de borda mensurados no corpo (imagem ouro) e os potencias gerados pela resolução do problema direto de um candidato à solução. Assim, o objetivo deste trabalho foi construir uma ferramenta computacional baseada em algoritmos de busca e otimização híbridos, com destaque para a Evolução Diferencial, a fim de reconstruir imagens de TIE. Para efeitos de comparação também foram utilizados para gerar imagens de TIE: Algoritmos Genéticos, Otimização por Enxame de Partículas e Recozimento Simulado. As simulações foram feitas no EIDORS, uma ferramenta usada em MatLab/ GNU Octave com código aberto voltada para a comunidade de TIE. Os experimentos foram feitos utilizando três diferentes configurações de imagens ouro (fantomas). As análises foram feitas de duas formas, sendo elas, qualitativa: na forma de o quão as imagens geradas pela técnica de otimização são parecidas com seu respectivo fantoma; quantitativa: tempo computacional, através da evolução do erro relativo calculado pela função objetivo do melhor candidato à solução ao longo do tempo de reconstrução das imagens de TIE; e custo computacional, através da avaliação da evolução do erro relativo ao longo da quantidade de cálculos da função objetivo pelo algoritmo. Foram gerados resultados para Algoritmos Genéticos, cinco versões clássicas de Evolução Diferencial, versão modificada de Evolução Diferencial, Otimização por Enxame de Partículas, Recozimento Simulado e três novas técnicas híbridas baseadas em Evolução Diferencial propostas neste trabalho. De acordo com os resultados obtidos, vemos que todas as técnicas híbridas foram eficientes para resolução do problema da TIE, obtendo bons resultados qualitativos e quantitativos desde 50 iterações destes algoritmos. Porém, merece destacar o rendimento do algoritmo obtido pela hibridização da Evolução Diferencial e Recozimento Simulado por ser a técnica aqui proposta mais promissora na reconstrução de imagens de TIE, onde mostrou ser mais rápida e menos custosa computacionalmente do que as outras técnicas propostas. Os resultados desta pesquisa geraram diversas contribuições na forma de artigos publicados em eventos nacionais e internacionais. / Electrical Impedance Tomography (EIT) is a technique that aim to reconstruct images of the interior of a body in a non-invasive and non-destructive form. Based on the application of the electrical current and on the measurement of the body’s edge electrical potential, made through of electrodes, an EIT image reconstruction algorithm generates the conductivity distribution map of this body’s interior. Several methods are applied to generate EIT images; however, they are still generated smooth contour images. This is due of the mathematical nature of EIT reconstruction problem as an ill-posed and ill-conditioned problem. Thus, there is not an exact internal conductivity distribution for one determinate edge potential distribution. The EIT is ruled mathematically by Poisson’s equations, and the image generation involves an iterative resolution of a direct problem, that treats the obtainment of the edge potentials through of an internal distribution of conductivity. The direct problem, in this dissertation, was applied through of Finite Elements Method. Thereby, is possible to apply search and optimization techniques that aim to minimize the mean square error relative (objective function) between the edge potentials measured in the body (gold image) and the potential generated by the resolution of the direct problem of a solution candidate. Thus, the goal of this work was to construct a computational tool based in hybrid search and optimization algorithms, highlighting the Differential Evolution, in order to reconstruct EIT images. For comparison, it was also used to generate EIT images: Genetic Algorithm, Particle Optimization Swarm and Simulated Annealing. The simulations were made in EIDORS, a tool used in MatLab/GNU Octave open source toward the TIE community. The experiments were performed using three different configurations of gold images (phantoms). The analyzes were done in two ways, as follows, qualitative: in the form of how the images generated by the optimization technique are similar to their respective phantom; quantitative: computational time, by the evolution of the relative error calculated for the objective function of the best candidate to the solution over time the EIT images reconstruction; and computational cost, by evaluating the evolution of the relative error over the amount of calculations of the objective functions by the algorithm. Results were generated for Genetic Algorithms, five classical versions of Differential Evolution, modified version of the Differential Evolution, Particle Optimization Swarm, Simulated Annealing and three new hybrid techniques based in Differential Evolution proposed in this work. According to the results obtained, we see that all hybrid techniques were efficient in solving the EIT problem, getting good qualitative and quantitative results from 50 iterations of these algorithms. Nevertheless, it deserves highlight the algorithm performance obtained by hybridization of Differential Evolution and Simulated Annealing to be the most promising technique here proposed to reconstruct EIT images, which proved to be faster and less expensive computationally than other proposed techniques. The results of this research generate several contributions in the form of published paper in national and international events.

Submitted by Irene Nascimento (irene.kessia@ufpe.br) on 2017-01-30T18:17:26Z No. of bitstreams: 2 license_rdf: 1232 bytes, checksum: 66e71c371cc565284e70f40736c94386 (MD5) DISSERTACAO_LEO_DEFESA - FINAL - CATALOGADA PDF.pdf: 4585329 bytes, checksum: 4b70c80127866cd2da97a6217bb6a34f (MD5) / Made available in DSpace on 2017-01-30T18:17:27Z (GMT). No. of bitstreams: 2 license_rdf: 1232 bytes, checksum: 66e71c371cc565284e70f40736c94386 (MD5) DISSERTACAO_LEO_DEFESA - FINAL - CATALOGADA PDF.pdf: 4585329 bytes, checksum: 4b70c80127866cd2da97a6217bb6a34f (MD5) Previous issue date: 2016-08-16 / CNPQ / Um dos principais desafios no projeto de sensores baseados em Ressonância de Plasmons de Superfície — RPS — é maximizar sua sensibilidade. Neste trabalho é proposto o uso de dois algoritmos heurísticos, Monte Carlo e Enxame de Partículas, para otimização de sensores baseados em RPS em interfaces planares, i.e, nas configurações de Kretschmann e de Otto, sem o auxílio da aproximação lorentziana para a curva de ressonância. Devido à natureza probabilística dos algoritmos, consegue-se obter um método simples e robusto para atingir essa otimização. É feita uma comparação quanto à eficiência computacional dos algoritmos em relação ao método tradicional de otimização, ficando demonstrado que o método de Enxame de Partículas é o mais eficiente em relação às outras técnicas. Com o emprego desse método, a dependência espectral dos parâmetros ótimos é obtida para sensores utilizando vários metais nas configurações de Kretschmann e de Otto, tanto para aplicações em meios gasosos quanto em meios aquosos. Um aplicativo foi desenvolvido e sua funcionalidade demonstrada, que pode ser executado diretamente via web, com base na metodologia proposta, para otimização de sensores RPS em interfaces planares. / One of the main challenges in the design of surface plasmon resonance – SPR – sensor systems is to maximize their sensitivity. In this work one proposes the use of two heuristic algorithms, Monte Carlo and Particle Swarm, for optimization of SPR sensors in planar interfaces, i.e, in the Kretschmann and Otto configurations, without use of the Lorentzian approximation to the resonance curve. Because of the probabilistic nature of the algorithms, one manages to obtain a simple and robust method to achieve optimization. A comparison is made on the computational efficiency of the algorithm relative to the traditional method of optimization, showing that the particle swarm optimization method is more efficient compared to other techniques. By employing this method, the spectral dependence of optimum parameters is obtained for sensors using a wide range of metal films in the Kretschmann and Otto configurations, both for applications in gaseous an in aqueous media. An app was developed and its functionality can be demonstrated, by direct execution via web, based on the proposed methodology for optimization of SPR sensors on planar interfaces.

Made available in DSpace on 2014-07-29T15:08:18Z (GMT). No. of bitstreams: 1 Dissertacao Sistemas Hidrotermicos.pdf: 1921130 bytes, checksum: 988097a7877583ede959085e07eade65 (MD5) Previous issue date: 2012-06-21 / Particle Swarm Optimization has been widely used to solve real-world problems, including the operation planning of hydrothermal generation systems, where the main goal is to achieve rational strategies of operation. This can be accomplished by minimizing the high-cost thermoelectric generation, while maximizing the low-cost hydroelectric generation. The optimization process must consider a set of complex constrains. This work presents the application of some recently proposed Particle Swarm Optimizers for a group of hydroelectric power plants of the Brazilian interconnected system, using real data from existing plants. There were performed some tests by using the standard PSO, PSO-TVAC, Clan PSO, Clan PSO with migration, Center PSO, and one approach proposed in this work, called Center Clan PSO, over three different mid-term periods. All PSO approaches were compared to the results achieved by a Non-linear Programming algorithm (NLP). Furthermore, another approach was proposed, based on Center PSO, named Extended Center PSO. It was observed that the PSO approaches presented as promising solutions to the problem, even better than NLP in some cases. / A Otimização por Enxame de Partículas tem sido amplamente utilizada na solução de problemas do mundo real, inclusive para o problema do planejamento da operação de sistemas de geração hidrotérmicos, em que o principal objetivo é encontrar estratégias racionais de operação. A solução é obtida através da minimização da geração térmica, alto custo, enquanto maximiza-se a geração hidrelétrica, que é de baixo custo. O processo de otimização deve considerar um conjunto complexo de restrições. Este trabalho apresenta a aplicação de uma abordagem recente chamada de Otimização por Enxame de Partículas para o problema com um grupo de usinas hidrelétricas do sistema interligado brasileiro, utilizando dados reais das usinas existentes. Foram realizados testes usando o PSO original, PSO-TVAC, Clan PSO, Clan PSO com a migração, Center PSO, e uma abordagem proposta neste trabalho, denominada Center Clan PSO, ao longo de três diferentes períodos de médio prazo. Todas as abordagens PSO foram comparadas com os resultados obtidos por um algoritmo de programação não linear (NLP). Além disso, uma outra abordagem foi proposta, com base no algoritmo Center PSO, chamada Extended Center PSO. Observou-se que as abordagens PSO apresentaram resultados promissores na solução do problema, com resultados até mesmo melhores, em alguns casos, que os obtidos pelo NLP.

Made available in DSpace on 2016-08-17T14:52:40Z (GMT). No. of bitstreams: 1 Dissertacao-OtilioPauloSilva.pdf: 2236988 bytes, checksum: e67439b623fd83b01f7bcce0020365fb (MD5) Previous issue date: 2016-03-04 / Breast cancer is now set on the world stage as the most common among women and the second biggest killer. It is known that diagnosed early, the chance of cure is quite significant, on the other hand, almost late discovery leads to death. Mammography is the most common test that allows early detection of cancer, this procedure can show injury in the early stages also contribute to the discovery and diagnosis of breast lesions. Systems computer aided, have been shown to be very important tools in aid to specialists in diagnosing injuries. This paper proposes a computational methodology to assist in the discovery of mass in dense and nondense breasts. This paper proposes a computational methodology to assist in the discovery of mass in dense and non-dense breasts. Divided into 6 stages, this methodology begins with the acquisition of the acquired breast image Digital Database for Screening Mammography (DDSM). Then the second phase is done preprocessing to eliminate and enhance the image structures. In the third phase is executed targeting with the Particle Swarm Optimization (PSO) to find regions of interest (ROIs) candidates for mass. The fourth stage is reduction of false positives, which is divided into two parts, reduction by distance and clustering graph, both with the aim of removing unwanted ROIs. In the fifth stage are extracted texture features using the functional diversity indicia (FD). Finally, in the sixth phase, the classifier uses support vector machine (SVM) to validate the proposed methodology. The best values found for non-dense breasts, resulted in sensitivity of 96.13%, specificity of 91.17%, accuracy of 93.52%, the taxe of false positives per image 0.64 and acurva free-response receiver operating characteristic (FROC) with 0.98. The best finds for dense breasts hurt with the sensitivity of 97.52%, specificity of 92.28%, accuracy of 94.82% a false positive rate of 0.38 per image and FROC curve 0.99. The best finds with all the dense and non dense breasts Showed 95.36% sensitivity, 89.00% specificity, 92.00% accuracy, 0.75 the rate of false positives per image and 0, 98 FROC curve. / O câncer de mama hoje é configurado no senário mundial como o mais comum entre as mulheres e o segundo que mais mata. Sabe-se que diagnosticado precocemente, a chance de cura é bem significativa, por outro lado, a descoberta tardia praticamente leva a morte. A mamografia é o exame mais comum que permite a descoberta precoce do câncer, esse procedimento consegue mostrar lesões nas fases iniciais, além de contribuir para a descoberta e o diagnóstico de lesões na mama. Sistemas auxiliados por computador, têm-se mostrado ferramentas importantíssimas, no auxilio a especialistas em diagnosticar lesões. Este trabalho propõe uma metodologia computacional para auxiliar na descoberta de massas em mamas densas e não densas. Dividida em 6 fases, esta metodologia se inicia com a aquisição da imagem da mama adquirida da Digital Database for Screening Mammography (DDSM). Em seguida, na segunda fase é feito o pré-processamento para eliminar e realçar as estruturas da imagem. Na terceira fase executa-se a segmentação com o Particle Swarm Optimization (PSO) para encontrar as regiões de interesse (ROIs) candidatas a massa. A quarta fase é a redução de falsos positivos, que se subdivide em duas partes, sendo a redução pela distância e o graph clustering, ambos com o objetivo de remover ROIs indesejadas. Na quinta fase são extraídas as características de textura utilizando os índices de diversidade funcional (FD). Por fim, na sexta fase, utiliza-se o classificador máquina de vetores de suporte (SVM) para validar a metodologia proposta. Os melhores valores achados para as mamas não densas, resultaram na sensibilidade de 96,13%, especificidade de 91,17%, acurácia de 93,52%, a taxe de falsos positivos por imagem de 0,64 e a acurva Free-response Receiver Operating Characteristic (FROC) com 0,98. Os melhores achados para as mamas densas firam com a sensibilidade de 97,52%, especificidade de 92,28%, acurácia de 94,82%, uma taxa de falsos positivos por imagem de 0,38 e a curva FROC de 0,99. Os melhores achados com todas as mamas densas e não densas, apresentaram 95,36% de sensibilidade, 89,00% de especificidade, 92,00% de acurácia, 0,75 a taxa de falsos positivos por imagem e 0,98 a curva FROC.

Context. Automatic placement of surveillance cameras' 3D models in an arbitrary floor plan containing obstacles is a challenging task. The problem becomes more complex when different types of region of interest (RoI) and minimum resolution are considered. An automatic camera placement decision support system (ACP-DSS) integrated into a 3D CAD environment could assist the surveillance system designers with the process of finding good camera settings considering multiple constraints. Objectives. In this study we designed and implemented two subsystems: a camera toolset in SketchUp (CTSS) and a decision support system using an enhanced Particle Swarm Optimization (PSO) algorithm (HPSO-DSS). The objective for the proposed algorithm was to have a good computational performance in order to quickly generate a solution for the automatic camera placement (ACP) problem. The new algorithm benefited from different aspects of other heuristics such as hill-climbing and greedy algorithms as well as a number of new enhancements. Methods. Both CTSS and ACP-DSS were designed and constructed using the information technology (IT) research framework. A state-of-the-art evolutionary optimization method, Hybrid PSO (HPSO), implemented to solve the ACP problem, was the core of our decision support system. Results. The CTSS is evaluated by some of its potential users after employing it and later answering a conducted survey. The evaluation of CTSS confirmed an outstanding satisfactory level of the respondents. Various aspects of the HPSO algorithm were compared to two other algorithms (PSO and Genetic Algorithm), all implemented to solve our ACP problem. Conclusions. The HPSO algorithm provided an efficient mechanism to solve the ACP problem in a timely manner. The integration of ACP-DSS into CTSS might aid the surveillance designers to adequately and more easily plan and validate the design of their security systems. The quality of CTSS as well as the solutions offered by ACP-DSS were confirmed by a number of field experts. / Sarmad Rohani: 004670606805 Reza Shams: 0046704030897

Cette thèse s’inscrit dans la thématique d’évaluation des ouvrages par des méthodes nondestructives. Le double objectif est de développer un code permettant d’effectuer au choixl’identification de paramètres par analyse inverse en utilisant un algorithme méta heuristique, ou dedéfinir une métrologie optimale (nombre de capteurs, positions, qualité) sur une structure, en vued’une identification de paramètres. Nous avons développé un code permettant de répondre à cesdeux objectifs. Il intègre des mesures in situ, un modèle mécanique aux éléments finis de lastructure étudiée et un algorithme d’optimisation méta heuristique appelé algorithme d’optimisationpar essaim particulaire. Ce code a d’abord été utilisé afin de caractériser l’influence de la métrologiesur l’identification de paramètres par analyse inverse, puis, en phase expérimentale, nous avonstravaillé sur des problèmes d’interactions sol structure. Un travail a également été réalisé surl’identification et la caractérisation de défauts par sollicitations au marteau d’impact. Enfin unexemple d’optimisation de métrologie (nombre de capteurs, positions et qualité) a été réalisé enutilisant le code original adapté pour cette étude. / This thesis deals with non-destructive evaluation in civil engineering. The objective is of two-fold:developing a code that will identify mechanical parameters by inverse analysis using a metaheuristicalgorithm, and developing another code to optimize the sensors placement (with respect tothe number and quality of the sensors) in order to identify mechanical parameters with the bestaccuracy. Our code integrates field data, a finite element model of the studying structure and aparticle swarm optimization algorithm to answer those two objectives. This code was firstly used tofocus on how the sensors placement, the number of used sensors, and their quality impact theaccuracy of parameters’ identification. Then, an application on a soil structure interaction wasconducted. Several tests to identify and characterize defaults using an impact hammer were alsocarried on. The last application focused on the optimization of the metrology in order to identifymechanical parameters with the best accuracy. This last work highlights the possibilities of theseresearches for structural health monitoring applications in civil engineering project.

Le domaine de la traction électrique a suscité un très grand intérêt dans les dernières années. La conception optimale de l'ensemble moteur électrique de traction – onduleur doit prendre en compte une variété de critères et contraintes. Étant donnée la liaison entre la géométrie du moteur et la stratégie de commande de l'onduleur, l'optimisation de l'ensemble de traction doit prendre en considération, en même temps, les deux composants.L'objectif de la thèse est la conception d'un outil d'optimisation appliqué à un système de traction électrique légère qu'emploie un moteur à réluctance variable alimenté (MRVCE) par un onduleur triphasé en pont complet. Le MRVCE est modélisé en utilisant la technique par réseau de perméances. En même temps, la technique de commande électronique peut être facilement intégrée dans le modèle pour effectuer l'analyse dynamique du fonctionnement du moteur. L'outil d'optimisation réalisé utilise l'algorithme par essaim de particules, modifié pour résoudre des problèmes multi-objectif. Les objectifs sont liés à la qualité des caractéristiques de fonctionnement du moteur, en temps que les variables d'optimisation concernent la géométrie du moteur aussi que la technique de commande. Les performances de l'algorithme sont comparées avec ceux de l'algorithme génétique (NSGA-II) et d'une implémentation classique de l'algorithme par essaim de particules multi-objectif.Finalement, un prototype de moteur à réluctance variable est construit et le fonctionnement du MRVCE alimenté depuis l'onduleur triphasé en pont complet est implémenté et les outils de modélisation et d'optimisation sont validés / The interest for the electric traction applications has been growing in the last few years. The optimal design of the electric motor and of the inverter that powers it needs to consider a long list of restrictions and criteria. Because of the fact that the geometry of the motor and the switching strategy are closely linked, the optimization of the traction solution needs to consider both, at the same time.The objective of this thesis is the development of an optimization tool applied for the optimization of an electric traction solution that uses the switched reluctance motor (SRM) fed from a three phase full bridge inverter. The SRM is modeled using Permeance Network Analysis (PNA). The switching technique can be easily integrated in the model, which gives the possibility to run a dynamic analysis. The optimization tool created uses the Particle Swarm Optimization (PSO) algorithm, modified for multi-objective problems. The algorithms performances are compared with those of the Genetic Algorithm, using the NSGA-II multi-objective technique and with a classic version of multiple objective particle swarm optimizer (MOPSO).Finally, a SRM prototype is constructed and the drive solution using a full-bridge three phase inverter is implemented. The modeling and optimization tools are thus experimentally validated

Nous nous intéressons dans cette thèse à la résolution du problème de tournées sélectives (Team Orienteering Problem - TOP) et ses variantes. Ce problème est une extension du problème de tournées de véhicules en imposan tcertaines limitations de ressources. Nous proposons un algorithme de résolution exacte basé sur la programmation linéaire en nombres entiers (PLNE) en ajoutant plusieurs inégalités valides capables d’accélérer la résolution. D’autre part, en considérant des périodes de travail strictes pour chaque véhicule durant sa tournée, nous traitons une des variantes du TOP qui est le problème de tournées sélectives multipériodique (multiperiod TOP - mTOP) pour lequel nous développons une métaheuristique basée sur l’optimisation par essaim pour le résoudre. Un découpage optimal est proposé pour extraire la solution optimale de chaque particule en considérant les tournées saturées et pseudo saturées .Finalement, afin de prendre en considération la disponibilité des clients, une fenêtre de temps est associée à chacun d’entre eux, durant laquelle ils doivent être servis. La variante qui en résulte est le problème de tournées sélectives avec fenêtres de temps (TOP with Time Windows - TOPTW). Deux algorithmes exacts sont proposés pour résoudre ce problème. Le premier est basé sur la génération de colonnes et le deuxième sur la PLNE à laquelle nous ajoutons plusieurs coupes spécifiques à ce problème. / We focus in this thesis on developing new algorithms to solve the Team Orienteering Problem (TOP) and two of its variants. This problem derives from the well-known vehicle routing problem by imposing some resource limitations .We propose an exact method based on Mixed Integer Linear Programming (MILP) to solve this problem by adding valid inequalities to speed up its solution process. Then, by considering strict working periods for each vehicle during its route, we treat one of the variants of TOP, which is the multi-period TOP (mTOP) for which we develop a metaheuristic based on the particle swarm optimization approach to solve it. An optimal split procedure is proposed to extract the optimal solution from each particle by considering saturated and pseudo-saturated routes. Finally, in order to take into consideration the availability of customers, a time window is associated with each of them, during which they must be served. The resulting variant is the TOP with Time Windows (TOPTW). Two exact algorithms are proposed to solve this problem. The first algorithm is based on column generation approach and the second one on the MILP to which we add additional cuts specific for this problem. The comparison between our exact and heuristic methods with the existing one in the literature shows the effectiveness of our approaches.

Recently, continuous-valued optimization problems have received a great amount of focus, resulting in optimization algorithms which are very efficient within the continuous-valued space. Many optimization problems are, however, defined within the binary-valued problem space. These continuous-valued optimization algorithms can not operate directly on a binary-valued problem representation, without algorithm adaptations because the mathematics used within these algorithms generally fails within a binary problem space. Unfortunately, such adaptations may alter the behavior of the algorithm, potentially degrading the performance of the original continuous-valued optimization algorithm. Additionally, binary representations present complications with respect to increasing problem dimensionality, interdependencies between dimensions, and a loss of precision. This research investigates the possibility of applying continuous-valued optimization algorithms to solve binary-valued problems, without requiring algorithm adaptation. This is achieved through the application of a mapping technique, known as angle modulation. Angle modulation effectively addresses most of the problems associated with the use of a binary representation by abstracting a binary problem into a four-dimensional continuous-valued space, from which a binary solution is then obtained. The abstraction is obtained as a bit-generating function produced by a continuous-valued algorithm. A binary solution is then obtained by sampling the bit-generating function. This thesis proposes a number of population-based angle-modulated continuous-valued algorithms to solve binary-valued problems. These algorithms are then compared to binary algorithm counterparts, using a suite of benchmark functions. Empirical analysis will show that the angle-modulated continuous-valued algorithms are viable alternatives to binary optimization algorithms. Copyright 2012, University of Pretoria. All rights reserved. The copyright in this work vests in the University of Pretoria. No part of this work may be reproduced or transmitted in any form or by any means, without the prior written permission of the University of Pretoria. Please cite as follows: Pamparà, G 2012, Angle modulated population based algorithms to solve binary problems, MSc dissertation, University of Pretoria, Pretoria, viewed yymmdd < http://upetd.up.ac.za/thesis/available/etd-02242012-090312 / > C12/4/188/gm / Dissertation (MSc)--University of Pretoria, 2012. / Computer Science / unrestricted