Ce travail de thèse porte, en premier lieu et principalement, sur le diagnostic à base de modèle d’une classe de SLC (Systèmes Linéaires à Commutations). Une problématique récurrente dans ce type de problème concerne la prise en considération de façon explicite les deux aspects, continu et discret, constituant un SLC. Dans ce cadre, nous avons proposé une méthodologie de détection et de localisation de défauts qui combine les outils initialement dédiés au diagnostic des systèmes continus et d’autres spécifiques aux SED (Systèmes à Evénement Discrets). L’approche proposée est conçue autour de trois modules : deux types de générateurs de résidus (issus de l’Automatique continue) et un estimateur en-ligne de l’état discret, appelé diagnostiqueur (issu de l’Automatique événementielle). Notre diagnostiqueur utilise les deux types de résidus, provenant de la partie continue, afin d’identifier le mode de fonctionnement du SLC et d’isoler les défauts de capteurs. Les résidus utilisés pour la localisation des défauts de capteurs sont générés à travers un générateur développé autour d’un schéma DOS (Dedicated Observer Scheme) à base d’observateurs hybrides,à la fois robustes vis-à-vis des entrées inconnues et sensibles aux défauts de capteurs. En second lieu, sur la base des résultats obtenus à l’aide de l’approche de diagnostic développée, nous avons proposé une approche préliminaire de synthèse de lois de commande tolérantes aux défauts de capteurs stabilisante via un retour d’état. Cette approche permet de préserver les performances nominales du système (situation non défaillante)en présence d’un défaut de capteurs. L’idée consiste à reconfigurer le retour d’état en remplaçant le vecteur d’état estimé à partir d’une sortie en défaut par un autre estimé à partir d’une sortie saine. La redondance des estimations est assurée dans cette approche par un banc d’observateurs hybrides robustes qui fournit plusieurs estimations correctes des vecteurs d’état et de sorties. / This thesis focuses, in first and foremost, on the model-based diagnosis of a class of SLC (Switched Linear Systems). The basic idea is to consider the continuous and discrete aspects, forming an SLC, explicitly.In this context, we proposed a methodology for detecting and locating faults that combines the tools originally dedicated to the continuous systems and the DES (discrete event systems) diagnosis. The proposed approach is designed around three modules: two types of residual generators (from the continuous Automatic) and anon-line estimator of the discrete state, called diagnoser (from the event Automatic). Our diagnoser uses the residual generators issue from the continuous part to identify the SLC mode and isolate sensor faults.Residues used for fault location sensors are generated through a generator developed around a scheme DOS(Dedicated Observer Scheme) based on hybrid observers. These observers are robust vis-à-vis the unknown input and sensitive to sensor faults. Secondly, based on the obtained results using the previous diagnosis approach, we proposed a preliminary approach for fault-tolerant state-feedback control law synthesis. This approach preserves the nominal performance of the system (as non-defaulting) in the presence of defective sensors. The idea is to reconfigure the state feedback by replacing the state vector estimated from defected output by another estimated from non-defected one. Redundancy estimates is provided in this approach by a bank of robust hybrid observer that provides several accurate estimates of state vectors and outputs.

The main objective of this thesis is to provide a smooth video playout on the mobile device over wireless networks. The parameters that specify the wireless channel include: bandwidth variation, frame losses, and outage time. These parameters may affect the quality of the video negatively, and the mobile users may notice sudden stops during the playout video, i.e., the picture is momentarily frozen, followed by a jump from one scene to a different one. This thesis focuses on eliminating frozen pictures and reducing the amount of video data that need to be transmitted. In order to eliminate frozen scenes on the mobile screen, we propose three different techniques. In the first technique, the video frames are split into sub-frames; these sub-frames are streamed over different channels. In the second technique the sub-frames will be “crossed” and sent together with other sub-frames that are from different positions in the streaming video sequence. If some sub-frames are lost during the transmission a reconstruction mechanism will be applied on the mobile device to recreate the missing sub-frames. In the third technique, we propose a Time Interleaving Robust Streaming (TIRS) technique to stream the video frames in different order. The benefit of that is to avoid losing a sequence of neighbouring frames. A missing frame from the streaming video will be reconstructed based on the surrounding frames on the mobile device. In order to reduce the amount of video data that are streamed over limited bandwidth channels, we propose two different techniques. These two techniques are based on identifying and extracting a high motion region of the video frames. We call this the Region Of Interest (ROI); the other parts of the video frames are called the non-Region Of Interest (non-ROI). The ROI is transmitted with high quality, whereas the non-ROI is interpolated from a number of references frames. In the first technique the ROI is a fixed size region; we considered four different types of ROI and three different scenarios. The scenarios are based on the position of the reference frames in the streaming frame sequence. In the second technique the ROI is identified based on the motion in the video frames, therefore the size, position, and shape of the ROI will be different from one video to another according to the video characteristic. The videos are coded using ffmpeg to study the effect of the proposed techniques on the encoding size. Subjective and objective metrics are used to measure the quality level of the reconstructed videos that are obtained from the proposed techniques. Mean Opinion Score (MOS) measurements are used as a subjective metric based on human opinions, while for objective metric the Structural Similarity (SSIM) index is used to compare the similarity between the original frames and the reconstructed frames.

Approved for public release; distribution is unlimited / Results of a highly vectorized and multitasked model of the world ocean circulation were analyzed. This model which uses realistic physics, geometry, and forcing on a high-resolution grid, was run on the NCAR Cray X-MP/48 using a robust-diagnostic strategy. Twenty years of model integration using one-half degree horizontal resolution and 20 levels of vertical resolution were accomplished after 200 wall-clock hours at a maximum FORTRAN performance speed of 450 megaflops. Seven key regions of the world ocean were analyzed using an ocean model processor. A representation of the global ocean circulation emerged that compared well with observations and that included strong advective features, fronts, and subtropical meanders. A diagnostic analysis program was developed to analyze meridional heat and volume transports. The results in all basins appear to be reasonable when compared to the results of other studies. For example, an anomalous northward heat transport of 3.8 x 10 to the 14th power W at 30 deg in the South Atlantic compares favorably with the estimate of 4.2 x 10 to the 14th power W at 32 deg S by Bennett (1978) using hydrographic data. The results of simulations conducted in this study can be compared and contrasted against the results of future eddy-resolving simulations. Keywords: Digital simulation, Advection, Heat transport, Meridional volume transport, Oceanographic fronts, Meanders, Thermoclines, Finite difference analysis. Theses. (EDC) / http://archive.org/details/analysisofresult00bart / Lieutenant Commander, United States Naval Reserve

[pt] O objetivo deste trabalho é projetar e implementar um algoritmo eficiente, confiável e preciso para a interseção de superfícies triangulares que representam modelos geológicos complexos. A grandeza das coordenadas espaciais desses modelos, em contraste com o relativamente pequeno tamanho médio de seus elementos, levam a problemas numéricos que podem gerar modelos ruins ou a erros graves do modelador geométrico. Além disso, um alto nível de precisão é desejável para se evitar erros de modelagem que possam gerar acidentes no campo de exploração. Neste trabalho, é proposta uma solução para reduzir os problemas numéricos com o uso de algumas estratégias geométricas e da Aritmética Exata. Exemplos demonstram estes problemas de robustez e validam o algoritmo proposto. / [en] The goal of this work is to design and to develop an efficient, reliable, and accurate algorithm for the intersection of triangular surfaces that represent complex geological models. The wide range of these models coordinates in contrast with the relatively small average size of its elements lead up to numerical instability problems, which may generate bad models or crash the geometric modeler. Additionally, a high degree of precision is desired in the model to avoid accidents in the field of oil exploration. In this work, it is proposed a solution to reduce the numerical issues by the use of some geometrical strategies and the Exact Arithmetic. Examples are used to demonstrate these robustness problems and to validate the proposed algorithm.

Cette thèse porte sur l'analyse de l'endommagement par fatigue et optimisation fiabiliste des structures soumises à des vibrations aléatoires. Le but de l'optimisation fiabiliste est de trouver le compromis entre le coût et la fiabilité. Plusieurs méthodes, telles que la méthode hybride et la méthode OSF ont été développées. Ces méthodes ont été appliquées dans des cas statiques et certains cas dynamiques spécifiques. Dans la réalité les structures sont soumises à des vibrations aléatoires qui peuvent provoquer un endommagement par fatigue. Dans cette thèse on présente la stratégie numérique de calcul de l'endommagement par fatigue dans le domaine fréquentiel et on propose une extension des méthodes RBDO dans le cas des structures soumises à des vibrations aléatoires. Aussi, une méthode RHM est développée. Enfin,une application industrielle qui porte sur la modélisation de la partie mécanique du banc HALT est présenté. / This thesis deals with the fatigue damage analysis and reliability-based design optimization (RBDO) of structures under random vibrations. The purpose of an RBDO method is to find the best compromise between cost and safety. Several methods, such as Hybrid method and OSF method have been developed. These methods have been applied in static cases and some specific dynamic cases. In fact, structures are subject to random vibrations which can cause a fatigue damage. In this thesis we present the strategy of calculation of the fatigue damage based on the Sines criterion in the frequency domain developed in our laboratory. Then, an extension of the RBDO methods in the case of structures subjected to random vibrations is proposed. Also, an RHM method is developed. Finally, we present an industrial application where we propose a model of the mechanical part of the HALT chamber.

La pile à combustible (PàC) est un dispositif qui transforme l'énergie chimique en énergie électrique. Ce dispositif nécessite un certain nombre d’auxiliaires pour son fonctionnement. Afin d'assurer des performances en termes de sécurité, de fiabilité et de durée de vie de la PàC, des systèmes de diagnostic et de commande adéquats sont indispensables. Ainsi, cette thèse est une contribution au problème du diagnostic de défauts et à la commande de la PàC de type PEM. Le premier volet de ce travail est consacré au développement de méthodes de diagnostic appliquées à la PàC. Pour ce faire, deux approches ont été proposées. La première concerne la synthèse d’un filtre H-/H∞ permettant la détection de défauts capteurs et actionneurs tout en assurant un niveau de robustesse vis-à-vis d'éventuelles perturbations. En tenant compte des caractéristiques du système de la PàC, les conditions d'existence et de stabilité du filtre sont données sous la forme de LMI. La seconde approche traite le problème de la détection des défauts paramétriques. Pour cet objectif, un observateur adaptatif a été proposé, ce dernier permet d'estimer la valeur du paramètre susceptible de présenter une défaillance. Cette méthode donne la possibilité d'estimer simultanément les défauts paramétriques et les états non mesurés du système. Le deuxième volet de la thèse est dédié à la commande du système PàC. Le but est de synthétiser une loi de commande par rétroaction qui permet de répondre à la sollicitation de la charge tout en respectant une contrainte de fonctionnement nominal de la PàC. La synthèse de cette loi de commande a été réalisée en prenant compte tout le caractère non linéaire du système / The fuel cell is a device that transforms the chemical energy in electricity. This device requires some auxiliaries for its operation. In order to ensure the performances in terms of security, reliability and life cycle of the fuel cell, adequate diagnostic and control systems are indispensables. This thesis is a contribution to the problem of faults diagnosis and control of PEM Fuel Cell. The first part of this work is dedicated to the development of diagnostic methods applied to the fuel cell. To do this, two approaches are proposed. The first one concerns the synthesis of a H-/H∞ filter allowing the sensors and actuators faults detection while ensuring a level of robustness towards disturbances. Taking into account the fuel cell system characteristics, the conditions for existence and stability of the filter are given in the form of Linear Matrix Inequalities (LMI). The second approach deals with the problem of parametric faults detection. For this purpose, an adaptive observer has been proposed, which makes it possible to estimate the value of the parameter likely to exhibit a failure. This method gives the possibility of simultaneously estimating the parametric faults and the unmeasured states of the system. The second part of the thesis is dedicated to the control of the fuel cell system. The purpose is to synthesize a feedback control law that allows to answer the load request while respecting a nominal operating constraint of the fuel cell. Such a functioning allows to have a better efficiency while preserving the state of health of the fuel cell. The control law is obtained by using the original nonlinear model and without any kind of linearization

The learning using privileged information paradigm has allowed support vector machine models to incorporate privileged information, variables available in the training set but not in the test set, to improve predictive ability. The consequent introduction of the knowledge transfer method has enabled a practical application of support vector machine models utilizing privileged information. This thesis describes a modified knowledge transfer method inspired by cross-validation, which unlike the current standard knowledge transfer method does not create the knowledge transfer function and the approximated privileged features used in the support vector machines on the same observations. The modified method, the robust knowledge transfer, is described and evaluated versus the standard knowledge transfer method and is shown to be able to improve the predictive performance of the support vector machines for both binary classification and regression.

Bundle Adjustment is a common fine-tuning step used in photogrammetry. It uses different types of parameters, some of which can be considered to be almost linear while others can be considered to be highly nonlinear, e.g. the rotational parameters. However, in the Bundle Adjustment process all parameters are treated equal. In concert with a poor initial estimate, this might cause Bundle Adjustment to diverge. In this report, two novel methods based on the damped Gauss-Newton with Armijo linesearch, modified by giving rotational parameters a special treatment, are tested. These methods, Clamped Alpha and Linear Exponential Search, are compared to Gauss-Newton with Armijo linesearch, as well as to the undamped Gauss-Newton method, also known as the Gauss-Markov method. Parameter sweeps over different perturbation levels for the angular parameters show that each of the three damped methods outperform the Gauss-Newton method. Notably, the Clamped Alpha method also outperforms the other two damped methods, with as much as 16 times as many convergent cases for a given perturbation level. Meanwhile, the average number of iterations is increased by only 1.8 times that of the Gauss-Newton with Armijo linesearch. The results add to existing research arguing for the use of damped methods in Bundle Adjustment. In particular, the simple and cheap Clamped Alpha method is potentially attractive for problems where the uncertainty of the camera angles is significant. While the Clamped Alpha method show promising results, it should be noted that the experiments in this study are on synthetic data. In order to solidify these results, further investigations into the performance of Clamped Alpha using real-world data should be conducted.

Em problemas de tomada de decisão sequencial modelados como Processos de Decisão Markovianos (MDP) pode não ser possível obter uma medida exata para as probabilidades de transição de estados. Visando resolver esta situação os Processos de Decisão Markovianos com Probabilidades Imprecisas (Markov Decision Processes with Imprecise Transition Probabilities, MDP-IPs) foram introduzidos. Porém, enquanto estes MDP-IPs se mostram como um arcabouço robusto para aplicações de planejamento no mundo real, suas soluções consomem muito tempo na prática. Em trabalhos anteriores, buscando melhorar estas soluções foram propostos algoritmos de programação dinâmica síncrona eficientes para resolver MDP-IPs com uma representação fatorada para as funções de transição probabilística e recompensa, chamados de MDP-IP fatorados. Entretanto quando o estado inicial de um problema do Caminho mais Curto Estocástico (Stochastic Shortest Path MDP, SSP MDP) é dado, estas soluções não utilizam esta informação. Neste trabalho será introduzido o problema do Caminho mais Curto Estocástico com Probabilidades Imprecisas (Stochastic Shortest Path MDP-IP, SSP MDP-IP) tanto em sua forma enumerativa, quanto na fatorada. Um algoritmo de programação dinâmica assíncrona para SSP MDP-IP enumerativos com probabilidades dadas por intervalos foi proposto por Buffet e Aberdeen (2005). Entretanto, em geral um problema é dado de forma fatorada, i.e., em termos de variáveis de estado e nesse caso, mesmo se for assumida a imprecisão dada por intervalos sobre as variáveis, ele não poderá ser mais aplicado, pois as probabilidades de transição conjuntas serão multilineares. Assim, será mostrado que os SSP MDP-IPs fatorados são mais expressivos que os enumerativos e que a mudança do SSP MDP-IP enumerativo para o caso geral de um SSP MDP-IPs fatorado leva a uma mudança de resolução da função objetivo do Bellman backup de uma função linear para uma não-linear. Também serão propostos algoritmos enumerativos, chamados de RTDP-IP (Real-time Dynamic Programming with Imprecise Transition Probabilities), LRTDP-IP (Labeled Real-time Dynamic Programming with Imprecise Transition Probabilities), SSiPP-IP (Short-Sighted Probabilistic Planner with Imprecise Transition Probabilities) e LSSiPP-IP (Labeled Short-Sighted Probabilistic Planner with Imprecise Transition Probabilities) e fatorados chamados factRTDP-IP (factored RTDP-IP) e factLRTDP-IP (factored LRTDP-IP). Eles serão avaliados em relação aos algoritmos de programação dinâmica síncrona em termos de tempo de convergência da solução e de escalabilidade. / In sequential decision making problems modelled as Markov Decision Processes (MDP) we may not have the state transition probabilities. To solve this issue, the framework based in Markov Decision Processes with Imprecise Transition Probabilities (MDP-IPs) is introduced. Therefore, while MDP-IPs is a robust framework to use in real world planning problems, its solutions are time-consuming in practice. In previous works, efficient algorithms based in synchronous dynamic programming to solve MDP-IPs with factored representations of the probabilistic transition function and reward function, called factored MDP-IPs. However, given a initial state of a system, modeled as a Stochastic Shortest Path MDP (SSP MDP), solutions does not use this information. In this work we introduce the Stochastic Shortest Path MDP-IPs (SSP MDP-IPs) in enumerative form and in factored form. An efficient asynchronous dynamic programming solution for SSP MDP-IPs with enumerated states has been proposed by Buffet e Aberdeen (2005) before which is restricted to interval-based imprecision. Nevertheless, in general the problem is given in a factored form, i.e., in terms of state variables and in this case even if we assume interval-based imprecision over the variables, the previous solution is no longer applicable since we have multilinear parameterized joint transition probabilities. In this work we show that the innocuous change from the enumerated SSP MDP-IP cases to the general case of factored SSP MDP-IPs leads to a switch from a linear to nonlinear objectives in the Bellman backup. Also we propose assynchronous dynamic programming enumerative algorithms, called RTDP-IP (Real-time Dynamic Programming with Imprecise Transition Probabilities), LRTDP-IP (Labeled Real-time Dynamic Programming with Imprecise Transition Probabilities), SSiPP-IP (Short-Sighted Probabilistic Planner with Imprecise Transition Probabilities) and LSSiPP-IP (Labeled Short-Sighted Probabilistic Planner with Imprecise Transition Probabilities), and factored algorithms called factRTDP-IP (factored RTDP-IP) and factLRTDP-IP (factored LRTDP-IP). There algorithms will be evaluated with the synchronous dynamic programming algorithms previously proposed in terms of convergence time and scalability.

Em geral, quando modelamos problemas de planejamento probabilístico do mundo real, usando o arcabouço de Processos de Decisão Markovianos (MDPs), é difícil obter uma estimativa exata das probabilidades de transição. A incerteza surge naturalmente na especificação de um domínio, por exemplo, durante a aquisição das probabilidades de transição a partir de um especialista ou de dados observados através de técnicas de amostragem, ou ainda de distribuições de transição não estacionárias decorrentes do conhecimento insuficiente do domínio. Com o objetivo de se determinar uma política robusta, dada a incerteza nas transições de estado, Processos de Decisão Markovianos com Probabilidades Imprecisas (MDP-IPs) têm sido usados para modelar esses cenários. Infelizmente, apesar de existirem diversos algoritmos de solução para MDP-IPs, muitas vezes eles exigem chamadas externas de rotinas de otimização que podem ser extremamente custosas. Para resolver esta deficiência, nesta tese, introduzimos o MDP-IP fatorado e propomos métodos eficientes de programação matemática e programação dinâmica que permitem explorar a estrutura de um domínio de aplicação. O método baseado em programação matemática propõe soluções aproximadas eficientes para MDP-IPs fatorados, estendendo abordagens anteriores de programação linear para MDPs fatorados. Essa proposta, baseada numa formulação multilinear para aproximações robustas da função valor de estados, explora a representação fatorada de um MDP-IP, reduzindo em ordens de magnitude o tempo consumido em relação às abordagens não-fatoradas previamente propostas. O segundo método proposto, baseado em programação dinâmica, resolve o gargalo computacional existente nas soluções de programação dinâmica para MDP-IPs propostas na literatura: a necessidade de resolver múltiplos problemas de otimização não-linear. Assim, mostramos como representar a função valor de maneira compacta usando uma nova estrutura de dados chamada de Diagramas de Decisão Algébrica Parametrizados, e como aplicar técnicas de aproximação para reduzir drasticamente a sobrecarga computacional das chamadas a um otimizador não-linear, produzindo soluções ótimas aproximadas com erro limitado. Nossos resultados mostram uma melhoria de tempo e até duas ordens de magnitude em comparação às abordagens tradicionais enumerativas baseadas em programação dinâmica e uma melhoria de tempo de até uma ordem de magnitude sobre a extensão de técnicas de iteração de valor aproximadas para MDPs fatorados. Além disso, produzimos o menor erro de todos os algoritmos de aproximação avaliados. / When modeling real-world decision-theoretic planning problems with the framework of Markov Decision Processes(MDPs), it is often impossible to obtain a completely accurate estimate of transition probabilities. For example, uncertainty arises in the specification of transitions due to elicitation of MDP transition models from an expert or data, or non-stationary transition distributions arising from insuficient state knowledge. In the interest of obtaining the most robust policy under transition uncertainty, Markov Decision Processes with Imprecise Transition Probabilities (MDP-IPs) have been introduced. Unfortunately, while various solutions exist for MDP-IPs, they often require external calls to optimization routines and thus can be extremely time-consuming in practice. To address this deficiency, we introduce the factored MDP-IP and propose eficient mathematical programming and dynamic programming methods to exploit its structure. First, we derive eficient approximate solutions for Factored MDP-IPs based on mathematical programming resulting in a multilinear formulation for robust maximin linear-value approximations in Factored MDP-IPs. By exploiting factored structure in MDP-IPs we are able to demonstrate orders of magnitude reduction in solution time over standard exact non-factored approaches. Second, noting that the key computational bottleneck in the dynamic programming solution of factored MDP-IPs is the need to repeatedly solve nonlinear constrained optimization problems, we show how to target approximation techniques to drastically reduce the computational overhead of the nonlinear solver while producing bounded, approximately optimal solutions. Our results show up to two orders of magnitude speedup in comparison to traditional at dynamic programming approaches and up to an order of magnitude speedup over the extension of factored MDP approximate value iteration techniques to MDP-IPs while producing the lowest error among all approximation algorithm evaluated.