Le développement de nouveaux convertisseurs d’énergie, plus efficients et plus respectueux de l’environnement, tels que les piles à combustibles, tend à s’accélérer. Leur diffusion à grande échelle suppose cependant des garanties en termes de sécurité et de disponibilité. Une solution possible pour ce faire est de développer des solutions de Prognostics and Health Management (PHM) de ces systèmes, afin de mieux les surveiller, anticiper les défaillances et recommander les actions nécessaires à l’allongement de leur durée de vie. Dans cet esprit, cette thèse porte sur la proposition d’une approche de pronostic dédiée aux piles à combustibles de types PEMFC à l’aide de filtrage particulaire.Le raisonnement s’attache tout d’abord à mettre en place une formalisation du cadre de travail ainsi que des exigences de mise en. Ceci se poursuit par le développement d’un modèle basé sur la physique permettant une estimation d’état de santé et de son évolution temporelle. L’estimation d’état est réalisée grâce à du filtrage particulaire. Différentes variantes de filtres sont considérées sur la base d’une de la littérature et de nouvelles propositions adaptées au PHM sont formulées et comparées à celles existantes. Les estimations d’état de santé fournies par le processus de filtrages ont utilisées pour réaliser des prédictions de l’état de santé futur du système, puis de sa durée devie résiduelle. L’ensemble des propositions est validé sur 4 jeux de données obtenus sur des PEMFC suivant des profils de mission variés. Les résultats montrent de bonnes performances de prédictions et d’estimations de durée de vie résiduelle avant défaillance. / The development of new energy converters, more efficient and environment friendly, such as fuelcells, tends to accelerate. Nevertheless, their large scale diffusion supposes some guaranties in termsof safety and availability. A possible solution to do so is to develop Prognostics and HealthManagement (PHM) on these systems, in order to monitor and anticipate the failures, and torecommend the necessary actions to extend their lifetime. In this spirit, this thesis deals with theproposal of a prognostics approach based on particle filtering dedicated to PEMFCs.The reasoning focuses first on setting a formalization of the working framework and theexpectations. This is pursued by the development of a physic-based modelling enabling a state ofhealth estimation and its evolution in time. The state estimation is made thanks to particle filtering.Different variants of filters are considered on the basis of the literature and new proposals adaptedto PHM are proposed and compared to existing ones. State of health estimates given by the filter areused to predict the future state of the system and its remaining useful life. All the proposals arevalidated on four datasets from PEMFC following different mission profiles. The results show goodperformances for predictions and remaining useful life estimates before failure.

A tarefa de operar um veículo submarino durante missões de inspeção de ambientes estruturados como, por exemplo, duto de usinas hidrelétricas, é feita principalmente por meio de referências visuais e uma bússola magnética. Porém alguns ambientes desse tipo podem apresentar uma combinação de baixa visibilidade e anomalias ferromagnéticas que inviabilizaria esse tipo de operação. Este trabalho, motivado pelo desenvolvimento de um veículo submarino operado remotamente (ROV) para ser usado em ambientes com essas restrições, propõe um sistema de navegação que utiliza o conhecimento prévio das dimensões do ambiente para corrigir o estado do veículo por meio da correlação dessas dimensões com os dados de um sonar de imageamento 2D. Para fazer essa correlação é utilizado o ltro de partículas, que é uma implementação não paramétrica do ltro Bayesiano. Esse ltro faz a estimação do estado com base nos métodos sequenciais de Monte Carlo e permite trabalhar de uma maneira simples com modelos não lineares. A desvantagem desse tipo de fusão sensorial é o seu alto custo computacional o que geralmente o impede de ser utilizado em aplicações de tempo real. Para que seja possível utilizar esse ltro em tempo real, será proposto neste trabalho uma implementação paralela utilizando uma unidade de processamento gráco (GPU) da NVIDIA e a arquitetura CUDA. Neste trabalho também será feito um estudo da utilização de duas congurações de sensores no sistema de navegação proposto neste trabalho. / The task of navigating a Remotely Operated underwater Vehicles (ROV) during inspection of man-made structures is performed mostly by visual references and occasionally a magnetic compass. Yet, some environments present a combination of low visibility and ferromagnetic anomalies that negates this approach. This paper, motivated by the development of a ROV designed to work on such environment, proposes a navigation method for this kind of vehicle. As the modeling of the system is nonlinear, the method proposed uses a particle lter to represent the vehicle state that is a nonparametric implementation of the Bayes lter. This method to work needs a priori knowledge of the environment map and to make the data association with this map, a 2D image sonar is used. The drawback of the sensor fusion used in this work is its high computational cost which generally prevents it from being used in real time applications. To be possible for this lter to be used in real time application, in this work is proposed a parallel implementation using a graphics processing unit (GPU) from NVIDIA and CUDA architecture. In this work is also made a study of two types of sensors conguration on the navigation system proposed in this work.

In this thesis we propose a novel approach for tracking multiple objects using structural information. The objects are tracked by combining particle filter and frame description with Attributed Relational Graphs (ARGs). We start by learning a structural probabilistic model graph from annotated images. The graphs are then used to evaluate the current tracking state and to correct it, if necessary. By doing so, the proposed method is able to deal with challenging situations such as abrupt motion and tracking loss due to occlusion. The main contribution of this thesis is the exploration of the learned probabilistic structural model. By using it, the structural information of the scene itself is used to guide the object detection process in case of tracking loss. This approach differs from previous works, that use structural information only to evaluate the scene, but do not consider it to generate new tracking hypotheses. The proposed approach is very flexible and it can be applied to any situation in which it is possible to find structural relation patterns between the objects. Object tracking may be used in many practical applications, such as surveillance, activity analysis or autonomous navigation. In this thesis, we explore it to track multiple objects in sports videos, where the rules of the game create some structural patterns between the objects. Besides detecting the objects, the tracking results are also used as an input for recognizing the action each player is performing. This step is performed by classifying a segment of the tracking sequence using Hidden Markov Models (HMMs). The proposed tracking method is tested on several videos of table tennis matches and on the ACASVA dataset, showing that the method is able to continue tracking the objects even after occlusion or when there is a camera cut. / Nesta tese, uma nova abordagem para o rastreamento de múltiplos objetos com o uso de informação estrutural é proposta. Os objetos são rastreados usando uma combinação de filtro de partículas com descrição das imagens por meio de Grafos Relacionais com Atributos (ARGs). O processo é iniciado a partir do aprendizado de um modelo de grafo estrutural probabilístico utilizando imagens anotadas. Os grafos são usados para avaliar o estado atual do rastreamento e corrigi-lo, se necessário. Desta forma, o método proposto é capaz de lidar com situações desafiadoras como movimento abrupto e perda de rastreamento devido à oclusão. A principal contribuição desta tese é a exploração do modelo estrutural aprendido. Por meio dele, a própria informação estrutural da cena é usada para guiar o processo de detecção em caso de perda do objeto. Tal abordagem difere de trabalhos anteriores, que utilizam informação estrutural apenas para avaliar o estado da cena, mas não a consideram para gerar novas hipóteses de rastreamento. A abordagem proposta é bastante flexível e pode ser aplicada em qualquer situação em que seja possível encontrar padrões de relações estruturais entre os objetos. O rastreamento de objetos pode ser utilizado para diversas aplicações práticas, tais como vigilância, análise de atividades ou navegação autônoma. Nesta tese, ele é explorado para rastrear diversos objetos em vídeos de esporte, na qual as regras do jogo criam alguns padrões estruturais entre os objetos. Além de detectar os objetos, os resultados de rastreamento também são usados como entrada para reconhecer a ação que cada jogador está realizando. Esta etapa é executada classificando um segmento da sequência de rastreamento por meio de Modelos Ocultos de Markov (HMMs). A abordagem de rastreamento proposta é testada em diversos vídeos de jogos de tênis de mesa e na base de dados ACASVA, demonstrando a capacidade do método de lidar com situações de oclusão ou cortes de câmera.

Este artigo tem como objetivo verificar se há assimetria na rigidez de preços e de salários na economia brasileira; ou seja, se os agentes da economia são mais rígidos para baixo ou para cima para ajustarem seus preços e salários. Além disso, realiza-se a análise dos efeitos da política monetária e fiscal na dinâmica da economia. Para isso, utiliza-se um modelo Dinâmico Estocástico de Equilíbrio Geral (DSGE) não linear com custos de ajustamento assimétricos de preços e de salários com base no trabalho de Aruoba, Bocola e Schorfheide (2013). Esse modelo pode gerar rigidez de preços e de salários para baixo (ou para cima) que podem gerar não linearidades fortes. Diante da não linearidade gerada por esses aspectos, o modelo é solucionado através de um método de solução não linear e os seus parâmetros são estimados com a ajuda do Filtro de Partículas. O resultado encontrado é que tanto os preços quanto os salários nominais são mais rígidos para baixo e essas assimetrias na rigidez influenciam a dinâmica da economia quando esta sofre choques de política monetária e fiscal. / The objective of this article is to verify if there is asymmetry in the rigidity of prices and wages for the Brazilian economy; i.e. if the economic agents are more rigid downward or upward when adjusting their prices and wages. In addition, it performs the analysis of the effects of monetary and fiscal policy in the dynamics of the economy. For this, it uses a nonlinear model of Dynamic Stochastic General Equilibrium (DSGE) with asymmetric adjustment costs in prices and wages based on the work of Aruoba, Bocola and Schorfheide (2013). This model can generate prices and wages rigidity downward (or upward) that can produce strong nonlinearities. Considering the non-linearity generated by these aspects, the model is solved through a non-linear solution method and its parameters are estimated with the help of Particle Filter. The obtained result is that both prices and nominal wages are more rigid downwards and these asymmetries in rigidity influence the dynamics of the economy when it suffers shocks from monetary and fiscal policies.

Esta tese foca no problema de estimação de estado e de identificação de parâametros para modelos afins por partes. Modelos afins por partes são obtidos quando o domínio do estado ou da entrada do sistema e particionado em regiões e, para cada região, um submodelo linear ou afim e utilizado para descrever a dinâmica do sistema. Propomos um algoritmo para estimação recursiva de estados e um algoritmo de identificação de parâmetros para uma classe de modelos afins por partes. Propomos um estimador de estados Bayesiano que utiliza o filtro de Kalman em cada um dos submodelos. Neste estimador, a função distribuição cumulativa e utilizada para calcular a distribuição a posteriori do estado assim como a probabilidade de cada submodelo. Já o método de identificação proposto utiliza o algoritmo EM (Expectation Maximization algorithm) para identificar os parâmetros do modelo. A função distribuição cumulativa e utilizada para calcular a probabilidade de cada submodelo a partir da medida do sistema. Em seguida, utilizamos o filtro de Kalman suavizado para estimar o estado e calcular uma função substituta da função likelihood. Tal função e então utilizada para identificar os parâmetros do modelo. O estimador proposto foi utilizado para estimar o estado do modelo não linear para vibrações causadas por folgas. Foram realizadas simulações, onde comparamos o método proposto ao filtro de Kalman estendido e o filtro de partículas. O algoritmo de identificação foi utilizado para identificar os parâmetros do modelo do jato JAS 39 Gripen, assim como, o modelos não linear de vibrações causadas por folgas. / This thesis focuses on the state estimation and parameter identi cation problems of piecewise a ne models. Piecewise a ne models are obtained when the state domain or the input domain are partitioned into regions and, for each region, a linear or a ne submodel is used to describe the system dynamics. We propose a recursive state estimation algorithm and a parameter identi cation algorithm to a class of piecewise a ne models. We propose a Bayesian state estimate which uses the Kalman lter in each submodel. In the this estimator, the cumulative distribution is used to compute the posterior distribution of the state as well as the probability of each submodel. On the other hand, the proposed identi cation method uses the Expectation Maximization (EM) algorithm to identify the model parameters. We use the cumulative distribution to compute the probability of each submodel based on the system measurements. Subsequently, we use the Kalman smoother to estimate the state and compute a surrogate function for the likelihood function. This function is used to estimate the model parameters. The proposed estimator was used to estimate the state of the nonlinear model for vibrations caused by clearances. Numerical simulations were performed, where we have compared the proposed method to the extended Kalman lter and the particle lter. The identi cation algorithm was used to identify the model parameters of the JAS 39 Gripen aircraft as well as the nonlinear model for vibrations caused by clearances.

Nesta dissertação é apresentada a comparação do desempenho de três estimadores - o Filtro de Kalman Estendido, o Filtro de Kalman Unscented e o Filtro de Partículas - aplicados para estimar a postura de um robô diferencial. Uma câmera foi fixa no teto para cobrir todo o campo operacional do robô durante os experimentos, a fim de extrair o mapa e gerar o ground truth. Isso permitiu realizar uma análise do erro de forma precisa a cada instante de tempo. O desempenho de cada um dos estimadores foi avaliado sistematicamente e numericamente para duas trajetórias. Os resultados desse primeiro experimento demonstram que os filtros proporcionam grandes melhorias em relação à odometria e que o modelo dos sensores é crítico para obter esse desempenho. O Filtro de Partículas mostrou um desempenho melhor em relação aos demais nos dois percursos. No entanto, seu elevado custo computacional dificulta sua implementação em uma aplicação de tempo real. O Filtro de Kalman Unscented, por sua vez, mostrou um desempenho semelhante ao Filtro de Kalman Estendido durante a primeira trajetória. Porém, na segunda trajetória, a qual possui uma quantidade maior de curvas, o Filtro de Kalman Unscented mostrou uma melhora significativa em relação ao Filtro de Kalman Estendido. Foi realizado um segundo experimento, em que o robô planeja e executa duas trajetórias. Os resultados obtidos mostraram que o robô consegue chegar a um determinado local com uma precisão da mesma ordem de grandeza do que a obtida durante a estimação de estados do robô. / In this dissertation, the performance of three nonlinear-model based estimators - the Extended Kalman Filter, the Unscented Kalman Filter and the Particle Filter - applied to pose estimation of a differential drive robot is compared. A camera was placed above the operating field of the robot to record the experiments in order to extract the map and generate the ground truth so the evaluation of the error can be done at each time step with high accuracy. The performance of each estimator is assessed systematically and numerically for two robot trajectories. The first experimental results showed that all estimators provide large improvements with respect to odometry and that the sensor modeling is critical for their performance. The particle filter showed a better performance than the others on both experiments, however, its high computational cost makes it difficult to implement in a real-time application. The Unscented Kalman Filter showed a similar performance to the Extended Kalman Filter during the first trajectory. However, during the second one (a curvier path) the Unscented Kalman Filter showed a significant improvement over the Extended Kalman Filter. A second experiment was carried out where the robot plans and executes a trajectory. The results showed the robot can reach a predefined location with an accuracy of the same order of magnitude as the obtained during the robot pose estimation.

A localização de objetos em uma imagem e acompanhamento de seu deslocamento numa sequência de imagens são tarefas de interesse teórico e prático. Aplicações de reconhecimento e rastreamento de padrões e objetos tem se difundido ultimamente, principalmente no ramo de controle, automação e vigilância. Esta dissertação apresenta um método eficaz para localizar e rastrear automaticamente objetos em vídeos. Para tanto, foi utilizado o caso do rastreamento da bola em vídeos esportivos, especificamente o jogo de futebol. O algoritmo primeiramente localiza a bola utilizando segmentação, eliminação e ponderação de candidatos, seguido do algoritmo de Viterbi, que decide qual desses candidatos representa efetivamente a bola. Depois de encontrada, a bola é rastreada utilizando o Filtro de Partículas auxiliado pelo método de semelhança de histogramas. Não é necessária inicialização da bola ou intervenção humana durante o algoritmo. Por fim, é feita uma comparação do Filtro de Kalman com o Filtro de Partículas no escopo do rastreamento da bola em vídeos de futebol. E, adicionalmente, é feita a comparação entre as funções de semelhança para serem utilizadas no Filtro de Partículas para o rastreamento da bola. Dificuldades, como a presença de ruído e de oclusão, tanto parcial como total, tiveram de ser contornadas. / The location of objects in an image and tracking its movement in a sequence of images is a task of theoretical and practical interest. Applications for recognition and tracking of patterns and objects have been spread lately, especially in the field of control, automation and vigilance. This dissertation presents an effective method to automatically locate and track objects in videos. Thereto, we used the case of tracking the ball in sports videos, specifically the game of football. The algorithm first locates the ball using segmentation, elimination and the weighting of candidates, followed by a Viterbi algorithm, which decides which of these candidates is actually the ball. Once found, the ball is tracked using the Particle Filter aided by the method of similarity of histograms. It is not necessary to initialize the ball or any human intervention during the algorithm. Next, a comparison of the Kalman Filter to Particle Filter in the scope of tracking the ball in soccer videos is made. And in addition, a comparison is made between the functions of similarity to be used in the Particle Filter for tracking the ball. Difficulties, such as the presence of noise and occlusion, in part or in total, had to be circumvented.

This thesis is concerned with the Pursuit-and-Evasion (PE) problem where the pursuer aims to minimize the time to capture the evader while the evader tries to prevent capture. In the problem, the evader has two advantages: a higher manoeuvrability and that the pursuer is uncertain about the evader??s state. Cooperation among multiple pursuer vehicles can thus be used to overcome the evader??s advantages. The focus here is on the formulation and development of frameworks and algorithms for cooperation amongst pursuers, aiming at feasible implementation on real and autonomous vehicles. The thesis is split into Parts I and II. Part I considers the problem of capturing an evader of higher manoeuvrability in a deterministic PE game. The approach is the employment of Forward Reachable Set (FRS) analysis in the pursuers?? control. The analysis considers the coverage of the evader??s FRS, which is the set of reachable states at a future time, with the pursuer??s FRS and assumes that the chance of capturing the evader is dependent on the degree of the coverage. Using the union of multiple pursuers?? FRSs intuitively leads to more evader FRS coverage and this forms the mechanism of cooperation. A framework for cooperative control based on the FRS coverage, or FRS-based control, is proposed. Two control algorithms were developed within this framework. Part II additionally introduces the problem of evader state uncertainty due to noise and limited field-of-view of the pursuers?? sensors. A search-and-capture (SAC) problem is the result and a hybrid architecture, which includes multi-sensor estimation using the Particle Filter as well as FRS-based control, is proposed to accomplish the SAC task. The two control algorithms in Part I were tested in simulations against an optimal guidance algorithm. The results show that both algorithms yield a better performance in terms of time and miss distance. The results in Part II demonstrate the effectiveness of the hybrid architecture for the SAC task. The proposed frameworks and algorithms provide insights for the development of effective and more efficient control of pursuer vehicles and can be useful in the practical applications such as defence systems and civil law enforcement.

Target tracking is one of the typical applications of wireless sensor networks: a large number of spatially deployed sensor nodes collaboratively sense, process and estimate the target state (e.g., position, velocity and heading). This thesis aimed to develop the collaborative information processing techniques that jointly address information processing and networking for the distributive estimation of target state in the highly dynamic and resources constrained wireless sensor networks. Taking into account the interplay between information processing and networking, this thesis proposed a collaborative information processing framework. The framework integrates the information processing which is responsible for the representation, fusion and processing of data and information with networking which caters for the formation of network, the delivery of information and the management of wireless channels. Within the proposed collaborative information processing framework, this thesis developed a suite of target tracking algorithms on the basis of the recursive Bayesian estimation method. For tracking a single target in wireless sensor networks, this thesis developed the sequential extended Kalman filter (S-EKF), the sequential unscented Kalman filter (S-UKF) and the Particle filter (PF). A novel extended Kalman filter and Particle filter hybrid algorithm, named as EKPF was also developed. The simulation results showed that the EKPF outperformed other three algorithms in terms of tracking accuracy and robustness. Moreover, to help evaluate the performance of the developed tracking algorithms, the posterior Cramer-Rao lower bound (PCRLB) which is the theoretical lower bound on the mean square error of the target state estimation was also computed. To tackle the measurement origin uncertainty in practical target tracking in wireless sensor networks, this thesis designed a Particle filter and probability density association filter (PDAF) hybrid algorithm, named as PF-PDAF for tracking a single target under the dual assumptions of clutter and missed detections. The PF-PDAF combines the advantages of PDAF algorithm in effectively solving the data association problem with the merits of PF that can accommodate the general non-Gaussian, nonlinear state space model. The PCRLB under measurement origin uncertainty was also derived and computed. For multiple target tracking in wireless sensor networks, this thesis designed a Particle filter and joint probabilistic data association filter (JPDAF) hybrid algorithm, named as PFJPDAF. The PF-JPDAF algorithm extends the traditional JPDAF to solve the general nonlinear non-Gaussian multiple targets tracking problems in wireless sensor networks. In the highly energy and communication bandwidth constrained wireless sensor networks, a critical consideration is that the information processing needs to be distributive. By adopting the hierarchical network architecture to achieve dynamic sensor nodes clustering and utilizing the Gaussian mixture model (GMM) to propagate estimation results amongst sensor clusters, this thesis developed the distributive PF, the distributive EKPF, the distributive PF-PDAF and the distributive PF-JPDAF tracking algorithms. Moreover, this thesis proposed a composite objective function incorporating both the information utility and the energy consumption measures to facilitate the sensing nodes selection in the distributive tracking algorithms. This composite objective function enables the distributive tracking algorithms to achieve the desirable tracking accuracy while still maintaining the lower energy consumption. / Thesis (Ph.D.) - University of Adelaide, School of Electrical and Electronic Engineering, 2008

<p>Navigation is about knowing your own position, orientation and velocity relative to some geographic entities. The sensor fusion considered in this thesis combines data from a dead reckoning system, inertial navigation system (INS), and measurements of the ground elevation. The very fast dynamics of aircraft navigation makes it difficult to estimate the true states. Instead the algorithm studied will estimate the errors of the INS and compensate for them. A height database is used along with the measurements. The height database is highly non-linear why a Rao-Blackwellized particle filter is used for the sensor fusion. This integrated navigation system only uses data from its own sensors and from the height database, which means that it is independent of information from outside the aircraft. </p><p>This report will describe the algorithm and illustrate the theory used. The main purpose is to evaluate the algorithm using real flight data, why the result chapter is the most important.</p>