L'étude présentée dans ce mémoire concerne le diagnostic des machines électriques à l'aide d'une association innovante entre la modélisation à évènements discrets, la Simulation Comparative et Concurrente (SCC) et les Réseaux de Neurones Artificiels (RNAs). Le diagnostic des machines électriques est effectué à partir d'une analyse temporelle des signaux statoriques et rotoriques à l'aide de réseaux de neurones de type Feed-Forward. Afin de comparer les différentes configurations de ces réseaux de neurones, l'approche proposée dans ce document utilise la simulation comparative et concurrente implémentée grâce au formalisme à évènements discrets DEVS (Discrete EVent system Specification). L'intégration des algorithmes de la SCC et des RNAs au sein du formalisme DEVS a été effectuée de manière générique et indépendamment du simulateur en développent des extensions et une librairie de modèles dans l'environnement de modélisation et de simulation à évènements discrets DEVSimPy. L'application de cette nouvelle solution pour le diagnostic des machines électriques permet de détecter les défauts à partir d'une architecture logiciel facilement portable sur des systèmes embarqués de type FPGA. / This thesis deals with the time-domain analysis of the electrical machines fault diagnosis due to early short-circuits detection in both stator and rotor windings. It also introduces to the Discrete EVent system Specification (DEVS) a generic solution to enable concurrent and comparative simulations (CCS). The DEVS-based CCS is an extension introduced using an aspect-oriented programming (AOP) to interact with the classic DEVS simulator. A new DEVS-based artificial neural network (ANN) is also introduced with a separation between learning and calculation models. The DEVS-based CCS is validated on the proposed ANN DEVS library inside the DEVSimPy environment. The concurrent ANN contributes in the time-domains analysis for the electrical machine fault diagnosis. This new method is based on data coming directly from the sensors without any computation but with a new dedicated preprocessing technique. Later, some enhancements are brought to the artificial neural network based on a new multistage architecture reducing the training time and errors compared to the single ANN. The new architecture and techniques has been validated on real data sixteen non-destructive windings faults analysis and localization.

A classificação de imagens de satélite é muito utilizada para elaborar mapas de cobertura do solo. O objetivo principal deste trabalho consistiu no mapeamento automático da cobertura do solo no entorno da Usina de Lajeado (TO) utilizando-se o classificador SVM. Buscou-se avaliar a dimensão de áreas antropizadas presentes na represa e a acurácia da classificação gerada pelo algoritmo, que foi comparada com a acurácia da classificação obtida pelo tradicional classificador MAXVER. Esta dissertação apresentou sugestões de calibração do algoritmo SVM para a otimização do seu resultado. Verificou-se uma alta acurácia na classificação SVM, que mostrou o entorno da represa hidrelétrica em uma situação ambientalmente favorável. Os resultados obtidos pela classificação SVM foram similares aos obtidos pelo MAXVER, porém este último contextualizou espacialmente as classes de cobertura do solo com uma acurácia considerada um pouco menor. Apesar do bom estado de preservação ambiental apresentado, a represa deve ter seu entorno devidamente monitorado, pois foi diagnosticada uma grande quantidade de incêndios gerados pela população local, sendo que as ferramentas discutidas nesta dissertação auxiliam esta atividade de monitoramento. / Satellite Image Classification are very useful for building land cover maps. The aim of this study consists on an automatic land cover mapping in the domain area of Lajeados dam, at Tocantins state, using the SVM classifier. The aim of this work was to evaluate anthropic dimension areas near the dam and also to verify the algorithms classification accuracy, which was compared to the results of the standard ML (Maximum Likelihood) classifier. This work presents calibration suggestions to the SVM algorithm for optimizing its results. SVM classification presented high accuracy, suggesting a good environmental situation along Lajeados dam region. Classification results comparison between SVM and ML were quite similar, but SVMs spatial contextual mapping areas were slightly better. Although environmental situation of the study area was considered good, monitoring ecosystem is important because a significant quantity of burnt areas was noticed due to local communities activities. This fact emphasized the importance of the tools discussed in this work, which helps environmental monitoring.

L’objectif de cette thèse consiste en l’élaboration de deux méthodologies visant à révéler des effets jusqu’alors masqués en modélisation décisionnelle. Dans la première partie, nous cherchons à mettre en œuvre une méthode d’analyse locale des critères de choix dans un contexte de choix binaires. Dans une seconde partie, nous mettons en avant les effets de génération dans l’étude des comportements de choix. Dans les deux parties, notre démarche de recherche combine de nouveaux outils d’analyse prédictive (Support Vector Machines, FANOVA, PLS) aux outils traditionnels de statistique inférentielle, afin d’enrichir les résultats habituels par des informations complémentaires sur les effets masqués que constituent les effets locaux dans les fonctions de choix binaires, et les effets de génération dans l’analyse temporelle des comportement de choix. Les méthodologies proposées, respectivement nommées AEL et APC-PLS, sont appliquées sur des cas réels, afin d’en illustrer le fonctionnement et la pertinence. / The objective of this thesis is the development of two methodologies to reveal previously hidden effects in decision modeling. In the first part, we try to implement a method of local analysis in order to select criteria in the context of binary choices. In a second part, we highlight the effects of generations in the study of consumer behavior. In both parts, our research approach combines new predictive analytical tools (such as Support Vector Machines, FANOVA, PLS) to traditional tools of inferential statistics, to enrich the usual results by additional on the masked effects, which are the local effects in the binary choice functions, and the effects of generation in temporal choice behavior analysis.The proposed methodologies, respectively named AEL and APC- PLS are both applied to real cases in order to illustrate their operation and relevance.

Les travaux de cette thèse s’inscrivent dans le contexte du développement de groupes motopropulseurs électriques pour véhicules automobiles. Ces développements ont mis en lumière les problématiques liées aux phénomènes acoustiques et vibratoires des machines électriques, spécifiques à l'application automobile. L'objectif de ces travaux est la mise en place de lois de commande dédiées à la réduction d'harmoniques de courant à l'origine d'harmoniques vibratoires.Pour cela, les différents éléments du système ont été modélisés, puis les paramètres du modèle électrique de la machine ont été identifiés au moyen de mesures expérimentales.Deux stratégies de commande ont ensuite été développées puis mises en place. D'une part, un régulateur dédié à l'harmonique de courant visé a été synthétisé par optimisation H-infini. D'autre part, la modélisation de l'origine de cet harmonique comme provenant d'un signal perturbateur externe a permis la compensation de celui-ci au moyen d'un observateur.L'efficacité de ces deux stratégies sur la réduction des harmoniques de courant visés et des harmoniques vibratoires correspondants a été validée expérimentalement sur banc d'essai vibratoire. / This work takes place in the context of the development of powertrains for electric vehicles. Those developments have raised new issues such as noise and vibrations of electrical machines, which are important issues in automotive applications. The aim of this thesis is to develop control laws dedicated to the reduction of currents harmonics that are responsible for vibration harmonics.In that purpose, different parts of the system have been modelled. Parameters of the electrical model of the studied machine have then been identified thanks to experimental measurements.Two different control laws are proposed and developed. One the one hand, a controller is synthetized thanks to H-infinity optimization. On the other hand, the studied current harmonic is modelled as coming from an external disturbance. This virtual disturbance is estimated thanks to an observer and compensated.

Neste trabalho, foram utilizadas as vantagens da ferramenta matemática de análise temporal e espectral, a transformada wavelet discreta (DWT), além dos coeficientes de predição linear (LPC) e do algoritmo de inteligência artificial, Least Squares Support Vector Machines (LS-SVM), para aplicações em análise de sinais de voz e classificação de vozes patológicas. Inúmeros trabalhos na literatura têm demonstrado o grande interesse existente por ferramentas auxiliares ao diagnóstico de patologias da laringe. Os componentes da DWT forneceram parâmetros de medida para a análise e classificação das vozes patológicas, principalmente aquelas provenientes de pacientes com edema de Reinke e nódulo nas pregas vocais. O banco de dados com as vozes patológicas foi obtido do Departamento de Otorrinolaringologia e Cirurgia de Cabeça e Pescoço do Hospital das Clínicas da Faculdade de Medicina de Ribeirão Preto (FMRP-USP). Utilizando-se o algoritmo de reconhecimento de padrões, LS-SVM, mostrou-se que a combinação dos componentes da DWT de Daubechies com o filtro LP inverso levou a um classificador de bom desempenho alcançando mais de 90% de acerto na classificação das vozes patológicas. / The main objective of this work was to use the advantages of the time-frequency analysis mathematical tool, discrete wavelet transform (DWT), besides the linear prediction coefficients (LPC) and the artificial intelligence algorithm, Least Squares Support Vector Machines (LS-SVM), for applications in voice signal analysis and classification of pathological voices. A large number of works in the literature has been shown that there is a great interest for auxiliary tools to the diagnosis of laryngeal pathologies. DWT components gave measure parameters for the analysis and classification of pathological voices, mainly that ones from patients with Reinke\'s edema and nodule in the vocal folds. It was used a data bank with pathological voices from the Otolaryngology and the Head and Neck Surgery sector of the Clinical Hospital of the Faculty of Medicine at Ribeirão Preto, University of Sao Paulo (FMRP-USP), Brazil. Using the automatic learning algorithm applied in pattern recognition problems, LS-SVM, results have showed that the combination of Daubechies\' DWT components and inverse LP filter leads to a classifier with good performance reaching more than 90% of accuracy in the classification of the pathological voices.

Les modèles géométriques de calcul permettent d’effectuer des calculs à l’aide de primitives géométriques. Parmi eux, le modèle des machines à signaux se distingue par sa simplicité, ainsi que par sa puissance à réaliser efficacement de nombreux calculs. Nous nous proposons ici d’illustrer et de démontrer cette aptitude, en particulier dans le cas de processus massivement parallèles. Nous montrons d’abord à travers l’étude de fractales que les machines à signaux sont capables d’une utilisation massive et parallèle de l’espace. Une méthode de programmation géométrique modulaire est ensuite proposée pour construire des machines à partir de composants géométriques de base les modules munis de certaines fonctionnalités. Cette méthode est particulièrement adaptée pour la conception de calculs géométriques parallèles. Enfin, l’application de cette méthode et l’utilisation de certaines des structures fractales résultent en une résolution géométrique de problèmes difficiles comme les problèmes de satisfaisabilité booléenne SAT et Q-SAT. Ceux-ci, ainsi que plusieurs de leurs variantes, sont résolus par machines à signaux avec une complexité en temps intrinsèque au modèle, appelée profondeur de collisions, qui est polynomiale, illustrant ainsi l’efficacité et le pouvoir de calcul parallèle des machines a signaux. / Geometrical models of computation allow to compute by using geometrical elementary operations. Among them, the signal machines model distinguishes itself by its simplicity, along with its power to realize efficiently various computations. We propose here an illustration and a study of this ability, especially in the case of massively parallel processes. We show first, through a study of fractals, that signal machines are able to make a massive and parallel use of space. Then, a framework of geometrical modular programmation is proposed for designing machines from basic geometrical components —called modules— supplied with given functionnalities. This method fits particulary with the conception of geometrical parallel computations. Finally, the joint use of this method and of fractal structures provides a geometrical resolution of difficult problems such as the boolean satisfiability problems SAT and Q-SAT. These ones, as well as several variants, are solved by signal machines with a model-specific time complexity, called collisions depth, which is polynomial, illustrating thus the efficiency and the parallel computational abilities of signal machines.

Les hémicryptophanes, molécules constituées d’une unité cyclotriveratrylène (CTV) reliée à un autre groupement de symétrie C3, sont des molécules cages qui trouvent de nombreuses applications dans le domaine de la chimie hôte-invité. Dans une première partie, ce manuscrit présente une revue des développements récents de la chimie des hémicryptophanes et définit les objectifs de ce travail de thèse.Nos travaux concernent principalement le design d'hémicryptophanes spécifiquement conçus pour la reconnaissance moléculaire de substrats d'intérêt. Ainsi, des structures variées d'hémicryptophanes ont été conçues comme capteur fluorescent du phosphate de choline, ou pour la reconnaissance de paire d'ions. Des hémicryptophanes hétéroditopiques originaux portant des unités tris(2-pyridylmethyl)amine (TPA) ont été synthétisés et présentent un grand intérêt pour des applications ultérieures en reconnaissance. Nous avons étudié la reconnaissance stéréosélective de saccharides avec des hémicryptophanes énantiopurs qui associent trois types de chiralité sur sept centres stéréogènes. Enfin, nous décrivons le mouvement de "respiration" moléculaire de cages énantiopures, enrichissant ainsi les rares applications des hémicryptophanes comme machines moléculaires.Des complexes du vanadium(V) des hémicryptophanes ont été développés comme catalyseurs efficaces dans des réactions de sulfoxydation et pour l'oxydation catalytique de lignines. Des dérivés azaphosphatrane d’hémicryptohanes ont été développés comme organocatalyseurs de la réaction de polymérisation par ouverture du lactide. La dernière partie de ce manuscript est consacrée à des développements nouveaux dans le domaine de la formation de cages par auto-assemblage. Nous avons réussi à former par coordination des tétraèdres dont les quatre faces comportent une sous-unité azaphosphatrane. Nous montrons pour la première fois que dans ce type de cage l’unité azaphosphatrane joue un rôle prépondérant dans la complexation d’anions. / In the wide area of host-guest chemistry, hemicryptophanes, a type of molecular cages combining a cyclotriveratrylene (CTV) unit with another different C3 symmetrical moiety, have received increasing attention. In a first part of this work, the advances in hemicryptophane chemistry have been thoroughly reviewed and the objectives of the thesis have been postulated. Our work mainly concerns the targeted molecular recognition by rational design of hemicryptophanes. Various hemicryptophane structures have been designed as fluorescent sensors for choline phosphate, or for ion-pairs recognition. Original heteroditopic hemicryptophanes bearing tris(2-pyridylmethyl)amine (TPA) units have been synthesized that present a great interest for further applications in molecular recognition. We investigated the stereoselective recognition of carbohydrates using enantiopure hemicryptophanes combining three classes of chirality on seven stereogenic units. At last, we described the breathing motion of a series of enantiopure cages, complementing the rare application of hemicryptophanes as molecular machines.Hemicryptophane vanadium(V) complexes, have been developed as efficient supramolecular catalysts for sulfoxidation and for the catalytic lignin oxidation. Azaphosphatrane-functionalized hemicryptophanes were developed as hydrogen-bonding organocatalysts for the ring-opening polymerization of lactide. Finally, my attention has opened to a more prospective view focusing on cages constructed by self-assembly, and we have demonstrated the feasibility of introducing azaphosphatrane moieties into tetrahedron capsules using subcomponent self-assembly, and also proved for the first time the utility of azaphosphatranes as anion binding moieties.

Dans beaucoup de pays du monde, on observe une importante augmentation du nombre de personnes âgées vivant seules. Depuis quelques années, un nombre significatif de projets de recherche sur l’assistance aux personnes âgées ont vu le jour. La plupart de ces projets utilisent plusieurs modalités (vidéo, son, détection de chute, etc.) pour surveiller l'activité de la personne et lui permettre de communiquer naturellement avec sa maison "intelligente", et, en cas de danger, lui venir en aide au plus vite. Ce travail a été réalisé dans le cadre du projet ANR VERSO de recherche industrielle, Sweet-Home. Les objectifs du projet sont de proposer un système domotique permettant une interaction naturelle (par commande vocale et tactile) avec la maison, et procurant plus de sécurité à l'habitant par la détection des situations de détresse. Dans ce cadre, l'objectif de ce travail est de proposer des solutions pour la reconnaissance des sons de la vie courante dans un contexte réaliste. La reconnaissance du son fonctionnera en amont d'un système de Reconnaissance Automatique de la Parole. Les performances de celui-ci dépendent donc de la fiabilité de la séparation entre la parole et les autres sons. Par ailleurs, une bonne reconnaissance de certains sons, complétée par d'autres sources informations (détection de présence, détection de chute, etc.) permettrait de bien suivre les activités de la personne et de détecter ainsi les situations de danger. Dans un premier temps, nous nous sommes intéressés aux méthodes en provenance de la Reconnaissance et Vérification du Locuteur. Dans cet esprit, nous avons testé des méthodes basées sur GMM et SVM. Nous avons, en particulier, testé le noyau SVM-GSL (SVM GMM Supervector Linear Kernel) utilisé pour la classification de séquences. SVM-GSL est une combinaison de SVM et GMM et consiste à transformer une séquence de vecteurs de longueur arbitraire en un seul vecteur de très grande taille, appelé Super Vecteur, et utilisé en entrée d'un SVM. Les expérimentations ont été menées en utilisant une base de données créée localement (18 classes de sons, plus de 1000 enregistrements), puis le corpus du projet Sweet-Home, en intégrant notre système dans un système plus complet incluant la détection multi-canaux du son et la reconnaissance de la parole. Ces premières expérimentations ont toutes été réalisées en utilisant un seul type de coefficients acoustiques, les MFCC. Par la suite, nous nous sommes penchés sur l'étude d'autres familles de coefficients en vue d'en évaluer l'utilisabilité en reconnaissance des sons de l'environnement. Notre motivation fut de trouver des représentations plus simples et/ou plus efficaces que les MFCC. En utilisant 15 familles différentes de coefficients, nous avons également expérimenté deux approches pour transformer une séquence de vecteurs en un seul vecteur, à utiliser avec un SVM linéaire. Dans le première approche, on calcule un nombre fixe de coefficients statistiques qui remplaceront toute la séquence de vecteurs. La seconde approche (une des contributions de ce travail) utilise une méthode de discrétisation pour trouver, pour chaque caractéristique d'un vecteur acoustique, les meilleurs points de découpage permettant d'associer une classe donnée à un ou plusieurs intervalles de valeurs. La probabilité de la séquence est estimée par rapport à chaque intervalle. Les probabilités obtenues ainsi sont utilisées pour construire un seul vecteur qui remplacera la séquence de vecteurs acoustiques. Les résultats obtenus montrent que certaines familles de coefficients sont effectivement plus adaptées pour reconnaître certaines classes de sons. En effet, pour la plupart des classes, les meilleurs taux de reconnaissance ont été observés avec une ou plusieurs familles de coefficients différentes des MFCC. Certaines familles sont, de surcroît, moins complexes et comptent une seule caractéristique par fenêtre d'analyse contre 16 caractéristiques pour les MFCC / In many countries around the world, the number of elderly people living alone has been increasing. In the last few years, a significant number of research projects on elderly people monitoring have been launched. Most of them make use of several modalities such as video streams, sound, fall detection and so on, in order to monitor the activities of an elderly person, to supply them with a natural way to communicate with their “smart-home”, and to render assistance in case of an emergency. This work is part of the Industrial Research ANR VERSO project, Sweet-Home. The goals of the project are to propose a domotic system that enables a natural interaction (using touch and voice command) between an elderly person and their house and to provide them a higher safety level through the detection of distress situations. Thus, the goal of this work is to come up with solutions for sound recognition of daily life in a realistic context. Sound recognition will run prior to an Automatic Speech Recognition system. Therefore, the speech recognition’s performances rely on the reliability of the speech/non-speech separation. Furthermore, a good recognition of a few kinds of sounds, complemented by other sources of information (presence detection, fall detection, etc.) could allow for a better monitoring of the person's activities that leads to a better detection of dangerous situations. We first had been interested in methods from the Speaker Recognition and Verification field. As part of this, we have experimented methods based on GMM and SVM. We had particularly tested a Sequence Discriminant SVM kernel called SVM-GSL (SVM GMM Super Vector Linear Kernel). SVM-GSL is a combination of GMM and SVM whose basic idea is to map a sequence of vectors of an arbitrary length into one high dimensional vector called a Super Vector and used as an input of an SVM. Experiments had been carried out using a locally created sound database (containing 18 sound classes for over 1000 records), then using the Sweet-Home project's corpus. Our daily sounds recognition system was integrated into a more complete system that also performs a multi-channel sound detection and speech recognition. These first experiments had all been performed using one kind of acoustical coefficients, MFCC coefficients. Thereafter, we focused on the study of other families of acoustical coefficients. The aim of this study was to assess the usability of other acoustical coefficients for environmental sounds recognition. Our motivation was to find a few representations that are simpler and/or more effective than the MFCC coefficients. Using 15 different acoustical coefficients families, we have also experimented two approaches to map a sequence of vectors into one vector, usable with a linear SVM. The first approach consists of computing a set of a fixed number of statistical coefficients and use them instead of the whole sequence. The second one, which is one of the novel contributions of this work, makes use of a discretization method to find, for each feature within an acoustical vector, the best cut points that associates a given class with one or many intervals of values. The likelihood of the sequence is estimated for each interval. The obtained likelihood values are used to build one single vector that replaces the sequence of acoustical vectors. The obtained results show that a few families of coefficients are actually more appropriate to the recognition of some sound classes. For most sound classes, we noticed that the best recognition performances were obtained with one or many families other than MFCC. Moreover, a number of these families are less complex than MFCC. They are actually a one-feature per frame acoustical families, whereas MFCC coefficients contain 16 features per frame

<p> </p><p>MTsim (Mobile Traffic Simulator) is an Ericsson AB internal software application that is part of 2Gsim. It is used to simulate elements of a GSM (Global System for Mobile communications) network for feature testing and automated testing. It is written in the programming language TSS Language, also known as SMILE which is a proprietary Ericsson programming language. SMILE is based on the principles of state matrix programming which in essence means that each program is on its own a finite state machine. The language is old and was originally intended as a macro language for smaller test programs, not for applications the size of MTsim.</p><p>It is of interest to evaluate the feasibility of performing an automatic conversion of applications written in SMILE, with special interest in converting MTsim, to a Java framework since Java has many advantages compared to SMILE. Java, as a language, is well suited for larger applications, there are numerous well supported tools and there is a much wider spread competence than there is for SMILE.</p><p>It is clear that in order to do a full conversion of a SMILE program to a Java framework two applications must be implemented. First a Java framework, which acts as a run time environment, must be designed which can host the translated programs. The other part is an actual translator which takes a SMILE program as input and ouputs a translated Java program. A more sophisticated framework is preferred since it makes the actual translated programs more light weight and easy to read which means higher degree of maintainability.</p><p>There are different ways to implement state machines in Java but the most flexible and versatile is to implement it as a black-box framework in an object oriented way where the framework has sophisticated mechanisms for message and event handling which is central to any state machine framework.</p><p>The translation for SMILE can easily be done by using a AST (abstract syntax tree) representation, which is a full representation of the SMILE program in tree-form. The AST is obtained from an intermediate state of the SMILE program compiler.</p><p> </p>

<p>Electromagnetic transients impinging on electromagnetic power devices - such as electric machines, transformers and reactors - can stress the design severely. Thus the magnitudes of the transients are often decisive for the design of the devices. Further, the operation of a device can be transient in itself. This is the case for the explosive magnetic flux compression generator (EMG) and a ferromagnetic actuator. </p><p>Models are presented that are mainly intended for transients in the millisecond range and faster. Hence, eddy currents and the related skin and proximity effect become significant in windings, magnetic cores and in the armatures of the devices. These effects are important for, e.g., the damping of the transients. Further, the displacement current in the insulation of the winding is significant. It changes the response of the windings dramatically, as it manifests the finite velocity of propagation of the electromagnetic fields. Under such circumstances, reflections and excited resonances can make the transient voltage and current distribution highly irregular. </p><p>Induced voltages are modelled with self and mutual inductances or reluctances combined with winding templates. The displacement currents are modelled with capacitances or coefficients of potential. Cauer circuits and their dual form are used to model eddy currents in laminated cores and in conductors. The Cauer circuit enables one to consider hysteresis and the non-linear response of a magnetic core. It is also used to model the eddy currents in the moving armature of an EMG. </p><p>A set-up is presented that can be used to study the transient voltage and the current distribution along a coil. </p><p>The transient response of coaxially insulated windings is analysed and modelled in detail. A lumped circuit model is developed for a coil, Dryformer^TM - the new high-voltage transformer - and Powerformer^TM, the new high-voltage generator. An alternative model, a combined lumped circuit and FEM model, is presented for a coaxially insulated winding in two slot cores.</p>