Este trabalho fundamenta-se no conceito de máquina de comitê de redes neurais artificiais e tem por objetivo resolver o problema inverso de transferência radiativa em um meio unidimensional, homogêneo, absorvedor e espalhador isotrópico. A máquina de comitê de redes neurais artificiais agrega e combina o conhecimento adquirido por um certo número de especialistas aqui representados, individualmente, por cada uma das redes neurais artificiais (RNA) que compõem a máquina de comitê de redes neurais artificiais. O objetivo é atingir um resultado final melhor do que o obtido por qualquer rede neural artificial separadamente, selecionando-se apenas àquelas redes neurais artificiais que apresentam os melhores resultados na fase de generalização descartando-se as demais, o que foi feito neste trabalho. Aqui são utilizados dois modelos estáticos de máquinas de comitê, usando a média aritmética de conjunto, que se diferenciam entre si apenas na composição do combinador de saída de cada máquina de comitê. São obtidas, usando-se máquinas de comitê de redes neurais artificiais, estimativas para os parâmetros de transferência radiativa, isto é, a espessura óptica do meio, o albedo de espalhamento simples e as refletividades difusas. Finalmente, os resultados obtidos com ambos os modelos de máquina de comitê são comparados entre si e com aqueles encontrados usando-se apenas redes neurais artificiais do tipo perceptrons de múltiplas camadas (MLP), isoladamente. Aqui essas redes neurais artificiais são denominadas redes neurais especialistas, mostrando que a técnica empregada traz melhorias de desempenho e resultados a um custo computacional relativamente baixo. / This work is based on the concept of neural networks committee machine and has the objective to solve the inverse radiative transfer problem in one-dimensional, homogeneous, absorbing and isotropic scattering media. The artificial neural networks committee machine adds and combines the knowledge acquired by an exact number of specialists which are represented, individually, by each one of the artificial neural networks (ANN) that composes the artificial neural network committee machine. The aim is to reach a final result better than the one obtained by any of the artificial neural network separately, selecting only those artificial neural networks that presents the best results during the generalization phase and discarding the others, what was done in this present work. Here are used two static models of committee machines, using the ensemble arithmetic average, that differ between themselves only by the composition of the output combinator by each one of the committee machine. Are obtained, using artificial neural networks committee machines, estimates for the radiative transfer parameters, that is, medium optical thickness, single scattering albedo and diffuse reflectivities. Finally, the results obtained with both models of committee machine are compared between themselves and with those found using artificial neural networks type multi-layer perceptrons (MLP), isolated. Here that artificial neural networks are named as specialists neural networks, showing that the technique employed brings performance and results improvements with relatively low computational cost.

Esta tese apresenta um estudo sobre modelagem computacional onde são aplicadas meta-heurísticas de otimização na solução de problemas inversos de transferência radiativa em meios unidimensionais com albedo dependente da variável óptica, e meios unidimensionais de duas camadas onde o problema inverso é tratado como um problema de otimização. O trabalho aplica uma meta-heurística baseada em comportamentos da natureza conhecida como algoritmo dos vagalumes. Inicialmente, foram feitos estudos comparativos de desempenho com dois outros algoritmos estocásticos clássicos. Os resultados encontrados indicaram que a escolha do algoritmo dos vagalumes era apropriada. Em seguida, foram propostas outras estratégias que foram inseridas no algoritmo dos vagalumes canônico. Foi proposto um caso onde se testou e investigou todas as potenciais estratégias. As que apresentaram os melhores resultados foram, então, testadas em mais dois casos distintos. Todos os três casos testados foram em um ambiente de uma camada, com albedo de espalhamento dependente da posição espacial. As estratégias que apresentaram os resultados mais competitivos foram testadas em um meio de duas camadas. Para este novo cenário foram propostos cinco novos casos de testes. Os resultados obtidos, pelas novas variantes do algoritmo dos vagalumes, foram criticamente analisados. / This thesis presents a study on computational modeling where optimization metaheuristics are applied to the solution of inverse radiative transfer problems in heterogeneous media: in one-layer media with space-dependent single scattering albedo, and two-layer media, where the inverse problem is formulated as an optimization problem. It is applied a metaheuristic based on the natural behavior of fireflies, known as the firefly algorithm. Initially, comparative studies of performance were made with two other classic stochastic algorithms. The results indicated that the choice of the firefly algorithm was appropriate. Then, it was proposed other strategies that have been inserted into the original firefly algorithm. A first case was proposed where all the strategies were investigated and tested. The strategies with the best results were investigated in other two different cases. All the three proposed cases involved one-layer media with space-dependent scattering albedo. The strategies have been tested and evaluated, and those which presented the best competitive results were then implemented for radiative problems in two-layer media. For this new scenario five test cases were investigated, and the results obtained with the new strategies developed in this work were critically analyzed.

Esta pesquisa consiste na solução do problema inverso de transferência radiativa para um meio participante (emissor, absorvedor e/ou espalhador) homogêneo unidimensional em uma camada, usando-se a combinação de rede neural artificial (RNA) com técnicas de otimização. A saída da RNA, devidamente treinada, apresenta os valores das propriedades radiativas [ω, τ0, ρ1 e ρ2] que são otimizadas através das seguintes técnicas: Particle Collision Algorithm (PCA), Algoritmos Genéticos (AG), Greedy Randomized Adaptive Search Procedure (GRASP) e Busca Tabu (BT). Os dados usados no treinamento da RNA são sintéticos, gerados através do problema direto sem a introdução de ruído. Os resultados obtidos unicamente pela RNA, apresentam um erro médio percentual menor que 1,64%, seria satisfatório, todavia para o tratamento usando-se as quatro técnicas de otimização citadas anteriormente, os resultados tornaram-se ainda melhores com erros percentuais menores que 0,04%, especialmente quando a otimização é feita por AG. / This research consists in the solution of the inverse problem of radiative transfer for a participating media (emmiting, absorbing and/or scattering) homogeneous one-dimensional in one layer, using the combination of artificial neural network (ANN), with optimization techniques. The output of the ANN, properly trained presents the values of the radiative properties [w, to, p1 e p2] that are optimized through the following techniques: Particle Collision Algorithm (PCA), Genetic Algorithm (GA), Greedy Randomized Adaptive Search Procedure (GRASP) and Tabu Search (TS). The data used in the training are synthetics, generated through the direct problem without the introduction of noise. The results obtained by the (ANN) alone, presents an average percentage error minor than 1,64%, what it would be satisfying, however, for the treatment using the four techniques of optimization aforementioned, the results have become even better with percentage errors minor than 0,03%, especially when the optimization is made by the GA.

Fundação Carlos Chagas Filho de Amparo a Pesquisa do Estado do Rio de Janeiro / A formulação explícita matricial desenvolvida nesta tese de doutorado foi proposta visando ser uma alternativa na solução de Problemas Inversos de estimativa de propriedades radiativas em meios participantes homogêneos unidimensionais usando a Equação de Transferência Radiativa para modelar a interação da radiação com o meio participante. A equação de transporte é formulada em forma matricial e o domínio angular é discretizado usando conceitos do método de ordenadas discretas e a expansão da função de fase do espalhamento anisotrópico em uma série de polinômios de Legendre. A formulação proposta consiste em uma formulação explícita para o problema inverso. Um arranjo apropriado das condições de contorno prescritas (fluxos incidentes) e dos fluxos emergentes nos contornos de uma placa permitem o cálculo direto do operador de transmissão, do operador albedo e do operador de colisão. A partir do operador de colisão calculado são obtidos os valores estimados dos coeficientes de extinção total e de espalhamento. São apresentadas as formulações para problemas em regime estacionário e em regime transiente, bem como os resultados para alguns casos-teste. / The explicit matrix formulation developed in the present thesis has been proposed as an alternative for the solution of Inverse Problems for radiative properties estimation in one-dimensional homogeneous participating media using Radiative transfer equation for the modeling of the radiation interaction with the participating medium. This transport equation is formulated in a matrix form and the angular domain is discretized using concepts of the discrete ordinates methods and the expansion of the function of phase function of anisotropic scattering in a series of Legendre polynomial. The formulation proposed consists on an explicit formulation for the inverse problem. An adequate assembly of the prescribed boundary conditions (incidents flux) and of the emerging flux at the boundaries of the slab allows the direct computation of the transmission, albedo and collision operators. From the computed collision operator estimated values for total extinction and scattering coefficients are obtained. The formulations for steady state and transient situations are presented, as well as test case results.

Nesta tese é proposta uma implementação para a solução do problema inverso com as estimativas das propriedades radiativas (o albedo de espalhamento simples, a espessura ótica do meio e as reflectividades difusas) a partir dos valores das intensidades de radiação que deixam o meio participante utilizando uma abordagem híbrida de sistemas neuro-fuzzy (SNF), o qual combina a utilização de sistemas de inferência fuzzy com as redes neurais artificiais. Busca-se com a utilização desse sistema híbrido integrar a habilidade dos sistemas fuzzy no tratamento de informações inexatas, imprecisas, e vagas, e a capacidade das redes neurais artificiais de tratar o aprendizado por experiência e a generalização do conhecimento. É proposta também uma metodologia de máquinas de comitês neuro-fuzzy na solução deste problema inverso em transferência radiativa. Foi observado paralelamente que a solução dos sistemas neuro-fuzzy e dos sistemas híbridos de máquinas de comitê neuro-fuzzy, apresentam baixa qualidade nos resultados quando são utilizados os dados experimentais com os menores coeficientes de sensibilidade para os parâmetros que serão estimados. Por outro lado, quando são utilizados dados com maior sensibilidade, são obtidos melhores resultados. Esta abordagem procura evitar a possibilidade da não convergência desses métodos. / In this thesis is proposed an implementation for solving the inverse problem with the estimates of radiative properties (the single scattering albedo, the optical thickness of the media and the diffuse reflectivities) by the values of the intensities of radiation that leaves the participant medium using a hybrid approach of neuro-fuzzy systems, which combines the use of fuzzy inference systems with artificial neural networks. The use of this hybrid system try to include the ability of fuzzy systems in the treatment of inaccurate, imprecise, and vague data, and the ability of artificial neural networks to deal with learning from experience and widespread knowledge. Also is proposed a methodology for machines committees in neuro-fuzzy solution of this inverse problem in radiative transfer. It was observed in parallel that the solution of neuro-fuzzy systems and hybrid systems neuro-fuzzy committee machines, have a poor quality results when using the experimental data with the lowest sensitivity coefficients for the parameters that will be estimated. Moreover, when data are used with greater sensitivity, better results are obtained. This approach seeks to avoid the possibility of non-convergence in such methods.

Este trabalho fundamenta-se no conceito de máquina de comitê de redes neurais artificiais e tem por objetivo resolver o problema inverso de transferência radiativa em um meio unidimensional, homogêneo, absorvedor e espalhador isotrópico. A máquina de comitê de redes neurais artificiais agrega e combina o conhecimento adquirido por um certo número de especialistas aqui representados, individualmente, por cada uma das redes neurais artificiais (RNA) que compõem a máquina de comitê de redes neurais artificiais. O objetivo é atingir um resultado final melhor do que o obtido por qualquer rede neural artificial separadamente, selecionando-se apenas àquelas redes neurais artificiais que apresentam os melhores resultados na fase de generalização descartando-se as demais, o que foi feito neste trabalho. Aqui são utilizados dois modelos estáticos de máquinas de comitê, usando a média aritmética de conjunto, que se diferenciam entre si apenas na composição do combinador de saída de cada máquina de comitê. São obtidas, usando-se máquinas de comitê de redes neurais artificiais, estimativas para os parâmetros de transferência radiativa, isto é, a espessura óptica do meio, o albedo de espalhamento simples e as refletividades difusas. Finalmente, os resultados obtidos com ambos os modelos de máquina de comitê são comparados entre si e com aqueles encontrados usando-se apenas redes neurais artificiais do tipo perceptrons de múltiplas camadas (MLP), isoladamente. Aqui essas redes neurais artificiais são denominadas redes neurais especialistas, mostrando que a técnica empregada traz melhorias de desempenho e resultados a um custo computacional relativamente baixo. / This work is based on the concept of neural networks committee machine and has the objective to solve the inverse radiative transfer problem in one-dimensional, homogeneous, absorbing and isotropic scattering media. The artificial neural networks committee machine adds and combines the knowledge acquired by an exact number of specialists which are represented, individually, by each one of the artificial neural networks (ANN) that composes the artificial neural network committee machine. The aim is to reach a final result better than the one obtained by any of the artificial neural network separately, selecting only those artificial neural networks that presents the best results during the generalization phase and discarding the others, what was done in this present work. Here are used two static models of committee machines, using the ensemble arithmetic average, that differ between themselves only by the composition of the output combinator by each one of the committee machine. Are obtained, using artificial neural networks committee machines, estimates for the radiative transfer parameters, that is, medium optical thickness, single scattering albedo and diffuse reflectivities. Finally, the results obtained with both models of committee machine are compared between themselves and with those found using artificial neural networks type multi-layer perceptrons (MLP), isolated. Here that artificial neural networks are named as specialists neural networks, showing that the technique employed brings performance and results improvements with relatively low computational cost.

Esta tese apresenta um estudo sobre modelagem computacional onde são aplicadas meta-heurísticas de otimização na solução de problemas inversos de transferência radiativa em meios unidimensionais com albedo dependente da variável óptica, e meios unidimensionais de duas camadas onde o problema inverso é tratado como um problema de otimização. O trabalho aplica uma meta-heurística baseada em comportamentos da natureza conhecida como algoritmo dos vagalumes. Inicialmente, foram feitos estudos comparativos de desempenho com dois outros algoritmos estocásticos clássicos. Os resultados encontrados indicaram que a escolha do algoritmo dos vagalumes era apropriada. Em seguida, foram propostas outras estratégias que foram inseridas no algoritmo dos vagalumes canônico. Foi proposto um caso onde se testou e investigou todas as potenciais estratégias. As que apresentaram os melhores resultados foram, então, testadas em mais dois casos distintos. Todos os três casos testados foram em um ambiente de uma camada, com albedo de espalhamento dependente da posição espacial. As estratégias que apresentaram os resultados mais competitivos foram testadas em um meio de duas camadas. Para este novo cenário foram propostos cinco novos casos de testes. Os resultados obtidos, pelas novas variantes do algoritmo dos vagalumes, foram criticamente analisados. / This thesis presents a study on computational modeling where optimization metaheuristics are applied to the solution of inverse radiative transfer problems in heterogeneous media: in one-layer media with space-dependent single scattering albedo, and two-layer media, where the inverse problem is formulated as an optimization problem. It is applied a metaheuristic based on the natural behavior of fireflies, known as the firefly algorithm. Initially, comparative studies of performance were made with two other classic stochastic algorithms. The results indicated that the choice of the firefly algorithm was appropriate. Then, it was proposed other strategies that have been inserted into the original firefly algorithm. A first case was proposed where all the strategies were investigated and tested. The strategies with the best results were investigated in other two different cases. All the three proposed cases involved one-layer media with space-dependent scattering albedo. The strategies have been tested and evaluated, and those which presented the best competitive results were then implemented for radiative problems in two-layer media. For this new scenario five test cases were investigated, and the results obtained with the new strategies developed in this work were critically analyzed.

Cette thèse porte sur une méthodologie d’estimation des capacités d’un futur instrument spatioporté. Le cas d’étude est l’instrument Flexible Combined Imager (FCI) à bord du futur satellite Meteosat Troisième Génération Imageur (MTG-I), et plus particulièrement ses capacités à détecter des variations de quantité d’aérosols désertiques dans l’atmosphère. Une meilleure connaissance de ces aérosols fait partie des besoins régulièrement exprimés pour l’étude du climat, la prévision météorologique ou l’estimation de la ressource solaire dans des zones arides comme le Sahara. Ce type d’aérosols est abondant dans l’atmosphère. Leurs propriétés physico-chimique les rendent distinguables des autre types d’aérosols comme ceux résultant de la pollution d’origine anthropique, d’autant qu’ils sont émis dans des zones protégées des contaminations par ces autres types. Ils représentent donc un cas d’étude simple pour valider la méthodologie développée dans cette thèse.La méthodologie consiste à réaliser un simulateur de vue du sol par l’instrument, à effectuer de très nombreuses simulations des luminances mesurées par l’instrument sous diverses conditions atmosphériques et de l’albédo du sol, à analyser les résultats de manière à quantifier l’influence de chaque variable dans la variation de la luminance, puis à conclure quant aux capacités de détection grâce un critère de détectabilité prenant en compte les caractéristiques de l’instrument.Le simulateur développé a été validé par confrontation avec des mesures réelles de l’instrument SEVIRI à bord du satellite Meteosat Second Generation. L’innovation principale réside dans l’usage de l’approche d’analyse de sensibilité globale (GSA). Cette dernière quantifie l’influence de chaque variable séparément ainsi que les termes croisés. Elle exploite des fonctions de répartition statistique des variables extraites d’observations, et permet par conséquent d’obtenir une analyse de sensibilité réaliste. La GSA produit aussi des fonctionnelles modélisant l’influence d’une ou plusieurs variables sur la variabilité du signal observé et utilisables pour différentes applications dans la télédétection. / This thesis deals with a methodology to assess the capabilities of future spaceborne instruments. The case study is the Flexible Combined Imager (FCI) of the future Meteosat Third Generation Imaging mission (MTG - I), and in particular its ability to detect variations in load of desert aerosols in a realistically variable atmosphere. A better understanding of the behavior of these aerosols is part of regularly expressed needs for the study of the climate, weather forecast or assessment of the solar resource in arid areas such as the Sahara. This type of aerosols is abundant in the atmosphere. Their physical and chemical properties make them distinguishable from other types of aerosols such as those resulting from anthropogenic pollution, especially as they are emitted in areas protected from contamination by these other types. They therefore represent a simple case study to validate the methodology developed in this thesis.The methodology is to provide a simulator of the view of the instrument to perform a large number of simulations of the radiance measured under different atmospheric conditions and ground albedo, to analyze the results in order to quantify the influence of each variable in the variation of radiance, and then conclude on the capabilities of detection through a test of detectability taking into account the characteristics of the instrument.The developed simulator was validated by comparison against actual measurements of the SEVIRI instruments onboard Meteosat Second Generation satellites. The main innovation lies in the use of the global sensitivity analysis approach (GSA). The latter quantifies the influence of each variable separately as well as their crossed terms. Cumulative distribution functions were computed from actual observations and allow a realistic sensitivity analysis of the instrument. The GSA is also used to compute functional representation of the influence of one or more variables on the variability of the observed signal. The usefulness of such representations is discussed for various applications in remote sensing.

Malgré l’existence de modèles précis, notre connaissance des structures à petite échelle des galaxies est toujours limitée par le manque de preuves observationnelles. Les progrès instrumentaux ont permis d’atteindre une haute résolution angulaire à l’aide des nouvelles générations de télescopes, mais celle-ci est restreinte à un faible nombre de cibles extragalactiques à causes des besoins de l’Optique Adaptative (OA). En effet, afin de permettre une mesure efficace du front d’onde, l’OA requiert une source brillante et ponctuelle proche de la cible scientifique, typiquement en dessous de 30 . La partie principale de cette thèse porte sur l’analyse de la dizaine de parsecs centrale des Galaxies à Noyaux Actifs (NAG) à l’aide de différentes techniques observationnelles et numériques. Nous avons dans ce contexte développé un code de transfert radiatif nous permettant d’analyser les données polarimétriques. La seconde partie de ce travail est dédiée à l’analyse d’images en proche infrarouges de galaxies à flambée d’étoiles afin de contraindre les paramètres décrivant les super amas stellaires, jeunes cocons de poussière très massifs abritant une formation d’étoiles très soutenue, à l’aide de données obtenues avec l’instrument CANARY, démonstrateur de nouvelles technologies d’OA. / Despite having strong theoretical models, the current limitation in our understanding of the small-scale structures of galaxies is linked to the lack of observational evidences. Many powerful telescopes and instruments have been developed in the last decades, however one of these strongest tools, namely Adaptive Optics (AO), can only be used on a very limited number of targets. Indeed, for AO to be efficient, a bright star is required close to the scientific target, typically under 30 . This is mandatory for the AO systems to be able to measure the atmospheric turbulence and this condition is rarely satisfied for extended extragalactic targets such as galaxies. The main part of this thesis work consisted in going deeper in the analysis of the inner tens of parsecs of Active Nuclei (AGN) by combining different techniques to obtain and to interpret new data. In this context, we developed a new radiative transfer code to analyse the polarimetric data. A second part of my work was dedicated to a high angular resolution study of Super Star Clusters (SSC) in a new system, thanks to data obtained with the AO demonstrator CANARY instrument.

Les machines électriques synchrones à aimants permanents sont susceptibles de rencontrer un disfonctionnement suite à un échauffement non maîtrisé. L’objectif de cette étude est de développer un modèle thermique générique et prédictif pouvant simuler diverses situations d’intérêts: régime permanent, régime transitoire, mode dégradé, entrefer immergé, haute vitesse. Pour cela, la méthode nodale est utilisée pour développer le modèle thermique générique. En parallèle, un banc d’essai et un prototype sont conçus pour valider le modèle. L’étude de sensibilité des résultats du modèle à certains paramètres montrent que certains coefficients de convection, certaines conductances de contact et la conductivité thermique radiale du bobinage ont une influence considérable sur les résultats du modèle. Cependant ces paramètres sont mal connus, car ils sont issus des formules empiriques ou des abaques. Grâce au prototype et au modèle développé, ces paramètres sont identifiés. Trois méthodes d’identification sont testées pour aboutir à une stratégie d’identification: les algorithmes génétiques, la méthode de Gauss-Newton et la méthode de Levenberg-Marquardt. Plusieurs essais sont effectués sur le prototype instrumenté. La mesure des températures à des lieux précis du prototype permet d’identifier les paramètres mal connus et de valider le modèle. / Permanent magnet synchronous machines are likely to break down due to poorly controlled heating. The goal of this study was to develop a generic and predictive thermal model to calculate the temperature of machines during the design phase simulating temperatures at various states. These states include: steady state, transient state, fault mode, axial circulating of a cooling fluid in the air-gap and high speed. The lumped parameter method was used to develop this generic thermal model. Meanwhile, a test bench and a prototype instrumented with thermocouples were manufactured to validate the model at the same time. Sensitivity studies of the results of the model to some parameters demonstrated that some convective coefficients, contact conductances and the thermal conductivity of the winding in the radial direction influenced the model. However, these parameters are poorly known, because empirical formulas or abacus are used to calculate them. Using, the prototype and the developed model, these parameters were identified. Three methods of identification were tested in order to find a strategy for the identification: the genetic algorithms method, the Gauss-Newton method and the Levenberg-Marquardt method. Many tests were done on the prototype. The measure of the temperatures on the specific place allows to identify these parameters and to validate the model.