Cette thèse porte sur l’optimisation du contrôle des Machines Synchrones à Réluctance Variable (MSRV) et en particulier sur la compensation des phénomènes vibratoires. Elle comprend trois points forts : le développement d’une méthode de minimisation des pulsations de couple et sa validation expérimentale, l’études des facteurs qui peuvent influencer la méthode et l’influence de cette méthode sur les autres performances de la MSRV.Tout d’abord, une méthode de contrôle permettant de compenser les ondulations de couple d’une MSRV existante a été développée. Premièrement, une équation analytique du couple est proposée et analysée afin d’exprimer la relation analytique harmonique entre le couple et les courants d’alimentation. La notion de « fonction de couple » est ensuite introduite. Une stratégie basée sur l’utilisation de cette fonction pour minimiser les ondulations du couple est présentée. Deux méthodes de réduction des ondulations de couple utilisant les différents harmoniques de la fonction de couple sont mises en évidence en. Elles ont été analysées et comparées pour répondre aux différents objectifs. Par la suite, la méthode a été validée par les résultats des simulations pour trois technologies de rotor de MSRV parmi les plus répandues. La méthode est aussi validée par la modélisation analytique et la simulation dynamique à l’aide du logiciel Matlab/Simulink ainsi que par les résultats d’expérimentation avec l’aide du banc d’essai.Ensuite, les facteurs qui peuvent influencer la méthode proposée pour réduire les ondulations de couple et les performances dynamiques ont analysés. Dans un premier temps, une amélioration sensible de l’aptitude au démarrage de la MSRV lorsque la compensation des ondulations de couple est mise en œuvre est mise en évidence. D’autre part la sensibilité de la méthode aux erreurs de mesure de position est évaluée afin de quantifier sa fiabilité dans le cas de l’utilisation d’estimateurs lors du contrôle sans capteur. Enfin, l’influence de la saturation sur la méthode proposée est aussi étudiée à l’aide d’une analyse par éléments finis du comportement magnétique de la MSRV.Finalement, l’influence de la méthode de compensation des ondulations de couple sur d’autres performances de la MSRV est analysée. Les courants optimaux ont plus harmoniques que les courants originaux. Par conséquent, les pertes dans le cuivre, dans le fer et dans les semi-conducteurs de l’onduleur sont modélisées analytiquement, calculées et comparées. Les conclusions montrent que les pertes dans le cuivre sont les plus sensibles à la compensation des harmoniques de couple tandis que les pertes dans le fer et dans l’onduleur sont faiblement affectées. D’autre part, la réduction des ondulations de couple peut changer le comportement vibro-acoustique de la MSRV. La dernière partie est consacrée à l’étude de la relation entre les ondulations de couple et le bruit. Une équation est proposée pour évaluer la variation du bruit produit par la compensation des ondulations de couple. Dans cette partie, les simulations dans Flux 2D sont effectuées pour calculer la variation du bruit. En outre, le logiciel professionnel Manatee réalisant l’analyse des vibrations et de l’acoustique est utilisé dans le but de conforter les résultats obtenus par la modélisation analytique. / This thesis aims to study the control and optimization of a synchronous reluctance machine for the purpose of improving the vibrational performance. The main works of the thesis can be classified into three parts: the proposed torque ripple reduction method, the factors which can influence the proposed method and the influence of the proposed method.At first, the torque ripple of synchronous reluctance machine is reduced by a control method. Firstly, a torque equation is proposed in order to present the relationship between torque ripple and the optimal currents. Then a new parameter, torque function, is put forward. Based on the torque function, the torque ripple reduction strategy is presented. Two different torque ripple minimizations are proposed by applying different torque function harmonics. They are analyzed and compared in order to define the optimal method. In order to test the proposed method further, the selected torque ripple minimization approach is applied to three SynRMs. The results of finite element simulations imply that the proposed method is effective to decrease the torque ripples of these three SynRMs. The proposed torque ripple reduction method is verified according to the models built in MATLAB/Simulink and the experiment results respectively.Then the factors which could influence the proposed torque ripple reduction method are analyzed. Firstly, torque function is a function of rotor position, current angle and saturation. Based on the model in Simulink, the influence of different starting position on the performance of the studied SynSR is analyzed. Besides, the estimated position errors produced by senserless control could also affect the toque ripple minimization by changing torque function. At last, the influence of saturation on the proposed torque ripple reduction method is introduced because the amplitudes of the optimal currents are increased.In addition, the influence of torque ripple reduction on the other perfomances of SynRM is analyzed. The optimal currents have more harmonics than the original sinusoidal currents. So three losses (copper losses, iron losses and inverter losses) are modeled, calculated, analyzed and compared. According to the results, the copper losses are the most sensible losses. The iron losses and the inverter losses are a little increased and the increased parts can be neglected. Besides, reducing torque ripple by adding stator currents could influence the vibro-acoustic of the studied SynRM. Thus this section aims to explain the relationship between torque ripple reduction and acoustic noise. An analytical equation is proposed in order to evaluate the variation of noise produced by torque ripple reduction. Simulations in Flux 2D have been performed in order to calculate the variation of noise resulted by torque ripple reduction. At last, the software Manatee which is professional in studying the vibration and noise is applied for the purpose of comparing the results with those of the finite element analysis.

No description available.

Submitted by SBI Biblioteca Digital (sbi.bibliotecadigital@puc-campinas.edu.br) on 2017-08-10T11:55:42Z No. of bitstreams: 1 FABIANA DA SILVA PODELESKI.pdf: 1801192 bytes, checksum: 27ac2ce1c17ed0cdfce39e602146bdc7 (MD5) / Made available in DSpace on 2017-08-10T11:55:42Z (GMT). No. of bitstreams: 1 FABIANA DA SILVA PODELESKI.pdf: 1801192 bytes, checksum: 27ac2ce1c17ed0cdfce39e602146bdc7 (MD5) Previous issue date: 2017-06-29 / Coordena??o de Aperfei?oamento de Pessoal de N?vel Superior - CAPES / This document proposes a new approach for the restoration of electric power distribution systems by optimal current flow model (OCF). The importance of working with proposals for restoration using OCF is to allow analyzing the problem of restoration by a multiobjective mathematical programming model with linear or quadratic objective function and constraints that represent the network structure of the distribution system. Two objectives are evaluated for the restoration, losses reduction and recomposition time, resulting in a multiobjective programming problem. The proposed restoration action consists of opening and closing of branches in order to transfer loads to areas that are affected by interrupting the power supply. The proposition is directed to the primary distribution networks, characterized by presenting a radial topology and being in a restorative state, when there is a permanent fault. It is also suitable for systems with distributed generation (DG) when the power flow in the branches is no longer unidirectional. The resolution of the problem starts from the prior knowledge of the distribution system (topology and operational levels), the affected region and the possible recomposition resources for restoring the network through OCF model. The objective function of losses can be represented by a linear or a quadratic function. The linear representation results in a problem with linear equations and inequalities, that is, in a linear programming problem. The use of a quadratic objective function (minimization of losses) implies a more complex model for execution, since it results in a set of linear and non-linear equations and inequalities, when it is a multiobjective problem. The quadratic model may become unsuitable for applications in smart grid technologies due to longer algorithm execution time. The results attested the importance of applying a multiobjective proposal, because when individually evaluated the criteria of loss minimization and shorter recomposition time, different recomposition options were obtained. / O presente documento prop?e um novo enfoque para a recomposi??o de sistemas de distribui??o de energia el?trica resolvido por modelo de Fluxo de Corrente ?timo (FCO). A import?ncia de se trabalhar com propostas para recomposi??o utilizando FCO ? possibilitar a an?lise do problema de recomposi??o por um modelo de programa??o matem?tica multiobjetivo, com fun??o objetivo linear ou quadr?tica e restri??es que representem a estrutura da rede do sistema de distribui??o. S?o avaliados dois objetivos para a recomposi??o, minimiza??o de perdas e menor tempo de recomposi??o, resultando em um problema de programa??o multiobjetivo. A a??o de recomposi??o proposta compreende manobras para transfer?ncia de carga ?s ?reas que se encontram ilhadas devido ? interrup??o de fornecimento de energia. A proposi??o est? dirigida ?s redes prim?rias de distribui??o, caracterizadas por apresentarem topologia radial e se encontrarem em um estado restaurativo, quando h? presen?a de uma falha permanente. Tamb?m ? adequada a sistemas com gera??o distribu?da (GD) quando os fluxos nos ramos deixam de ser unidirecionais. A resolu??o do problema parte do conhecimento pr?vio do sistema de distribui??o (topologia e n?veis operacionais), da regi?o afetada e dos poss?veis recursos restauradores para restaura??o da rede por meio de FCO. A fun??o objetivo pode ser representada por uma fun??o linear ou quadr?tica para as perdas. A representa??o linear resulta em um problema com equa??es e inequa??es lineares, ou seja, em um problema de programa??o linear. A utiliza??o de uma fun??o objetivo quadr?tica (minimiza??o de perdas) implica em um modelo mais complexo para execu??o, uma vez que re?ne um conjunto de equa??es e inequa??es lineares e n?o lineares, quando se tratar de um problema multiobjetivo. O modelo quadr?tico pode se tornar impr?prio para aplica??es em tecnologias de redes inteligentes devido ao maior tempo de execu??o de algoritmo. Os resultados atestaram a import?ncia de aplica??o de uma proposta multiobjetivo, pois quando avaliados individualmente os crit?rios de minimiza??o de perdas e de menor tempo de recomposi??o, foram obtidas diferentes op??es de recomposi??o.