Le procédé de soudage à l'arc sous atmosphère inerte (TIG) est souvent employé pour des assemblages nécessitant une grande qualité du joint soudé. Les propriétés du joint soudé dépendent essentiellement du cycle thermique imposé par l'opération de soudage, de la composition chimique du matériau métallique et des mouvements convectifs du métal fondu dans le bain de fusion. L'écoulement du métal liquide dans le bain de fusion modifie la distribution de température en son sein et à proximité, ainsi que la forme géométrique du joint. Afin d'améliorer l'opération de soudage TIG, par exemple pour accroitre la productivité ou éviter des défauts rédhibitoires, il est nécessaire de bien comprendre les phénomènes physiques mis en jeu dans le bain de fusion ainsi que l'effet des paramètres opératoires (intensité, hauteur d'arc, gaz …) sur ces phénomènes physiques. Dans le but d'appréhender les phénomènes mis en jeu au cours de l'opération TIG et dans le bain de fusion, un modèle multi-physique 2D axisymétrique a été établi et résolu par la méthode des éléments finis (MEF). Les forces telles que Lorentz (électromagnétique), Marangoni (Tension superficielle), Boussinesq et la force de cisaillement du plasma d'arc ont été prises en compte au niveau du bain de fusion. Le modèle TIG établi est utilisé pour prédire la distribution de température et la distribution des vitesses dans le bain de fusion ainsi que la forme géométrique du bain de fusion. Un protocole expérimental a été développé dans le but de valider le modèle proposé. Pour cela, une opération de soudage TIG stationnaire (pas de mouvement de la torche) a été réalisée sur un disque métallique. L'opération a été contrôlée par des mesures de température, par une observation de la formation et de l'évolution de la surface du bain de fusion avec une caméra rapide et un enregistrement des paramètres opératoires (intensité et tension). Toutes les données sont synchronisées entre elles pour permettre une analyse expérimentale pertinente. La confrontation des résultats expérimentaux avec le modèle multi-physique du soudage TIG a fait apparaître une assez bonne adéquation, mais des différences existent, essentiellement liées à la méconnaissance des paramètres décrivant le flux de chaleur utilisé dans la simulation. Le flux de chaleur a été modélisé par une fonction Gaussienne qui nécessite la connaissance du rendement du procédé TIG et la distribution spatiale (ou rayon de la Gaussienne). L'estimation de ces paramètres a été réalisée par une méthode inverse. Cette méthode inverse a consisté à estimer les paramètres inconnus à partir des données expérimentales disponibles. La méthode d'optimisation dite de Levenberg-Marquardt, associée à une technique de régularisation itérative, a été utilisée pour estimer les paramètres. La pertinence et la robustesse de cette méthode ont été validées au travers de plusieurs cas numériques ; soit des cas utilisant des données « exactes » ou des données « bruitées ». Trois types d'erreurs ont été analysés séparément : bruit de mesure, erreur sur la position du capteur et imprécision sur la valeur des propriétés thermophysiques. Les deux dernière erreurs sont celles qui impactent fortement le résultat de l'estimation, essentiellement l'estimation du rendement du procédé TIG. Enfin, une partie des données expérimentales a été utilisée pour résoudre le problème inverse. Les paramètres ont été estimés avec une marge d'erreur inférieure à 10% et ils sont en bon accord avec les valeurs trouvées dans la littérature. / Gas Tungsten Arc Welding (GTAW) process is generally used for assemblies that requires high quality weld joint. The microstructure and the weld joint relies mainly on the thermal cycle due to the welding operation, the chemical composition of the metallic material and the complex flow of molten metal in the weld pool. Moreover the fluid flow in the weld pool play a major role in the temperature distribution and the final weld pool shape. Better understanding of the physical phenomena involved in the welding operation, more exactly in the weld pool, are the fundamental step for improving the GTAW operation, for example increase the productivity or avoid defects. In the present research work, a two dimensional axi-symmetric multiphysics model was established in order to predict the weld pool shape evolution in the frame of a stationary Gas Tungsten Arc Welding using a finite element numerical approach. The weld pool model included various driving forces such as self-induced electromagnetic (Lorentz force), surface tension (Marangoni force), buoyancy and the arc plasma drag force. The stated GTAW model is used for predicting the velocity and temperature distribution in the fusion zone and the final weld pool shape. In order to validate the GTAW model, an experimental set up was defined for synchronizing the acquisition of time dependent data such as temperature, weld pool radius and welding process parameters (current and voltage). Image processing algorithms were developed for the time dependent weld pool size identification from the high speed camera images. Comparison between experimental and calculated data exhibited important discrepancies on the temperature field and weld pool radius. These discrepancies are due to the incoming heat flux from the arc plasma into the work piece. The heat flux was modeled with a Gaussian function itself described with few parameters;two of these required to be estimated: GTAW efficiency and Gaussian distribution.An inverse approach is used for estimating these parameters from the available experimental data: temperature, weld pool radius and macrographs. The Levenberg-Marquardt method is used to solve the inverse heat transfer problem coupled to an iterative process regularization. Afterward the inverse heat transfer problem was investigated through few numerical cases in order to verify its robustness to three sorts of error in the input data (measurement noise, sensor location error and inaccuracies associated with the thermophysical properties). The inverse approach was robust to errors introduced on measurement data. However, errors on the position of sensors or on the knowledge of material thermo-physical properties are problematic on the GTAW efficiency estimation. Finally the inverse problem was solved with experimental measurement. The estimated parameters are in good agreement with the literature. The evaluated error on the estimated parameters is less than 10%.

Submitted by Renata Lopes (renatasil82@gmail.com) on 2017-02-23T19:26:50Z No. of bitstreams: 1 francianeconceicaopeters.pdf: 4730497 bytes, checksum: 201c60342a8bf9edc9b308fa50fafa54 (MD5) / Approved for entry into archive by Adriana Oliveira (adriana.oliveira@ufjf.edu.br) on 2017-02-24T12:08:43Z (GMT) No. of bitstreams: 1 francianeconceicaopeters.pdf: 4730497 bytes, checksum: 201c60342a8bf9edc9b308fa50fafa54 (MD5) / Made available in DSpace on 2017-02-24T12:08:43Z (GMT). No. of bitstreams: 1 francianeconceicaopeters.pdf: 4730497 bytes, checksum: 201c60342a8bf9edc9b308fa50fafa54 (MD5) Previous issue date: 2010-01-27 / CAPES - Coordenação de Aperfeiçoamento de Pessoal de Nível Superior / A construção de imagens associadas à distribuição de condutividades no interior de um meio condutor a partir de injeção de corrente elétrica e medidas de potencial no contorno externo do corpo é uma técnica conhecida como tomografia por impedância elétrica. É um problema inverso que tem sido estudado visando aplicações biomédicas, monitoramento de processos industriais e investigação geofísica. Em alguns casos, é possível levar em consideração informações conhecidas sobre o domínio do corpo no processo de construção da imagem, recaindo no problema da detecção de inclusões que é o problema efetivamente tratado neste trabalho. Este problema pode ser resolvido por meio da minimização de uma função da diferença entre potenciais medidos no contorno e calculados para uma dada distribuição de condutividades. O presente trabalho desenvolve uma estratégia para a solução deste problema baseada na parametrização da geometria do contorno das inclusões cujas formas e dimensões se pretende determinar. O problema de minimização é resolvido por meio do Método de Levenberg-Marquardt e o problema direto via Método dos Elementos de Contorno. Para avaliar o desempenho da estratégia proposta são apresentados resultados numéricos envolvendo contornos definidos por splines, problemas com a presença de ruído nas medidas, avaliação de protocolos de injeção de corrente e medição de potencial elétrico e ainda uma aplicação voltada ao monitoramento cardíaco. / The images reconstruction of the conductivity distribution inside a conductive body based on electrical current injection and potential measurements on the outer boundary of this body is a technique known as electrical impedance tomography. This is an inverse problem that has been studied in biomedical applications, industrial process monitoring and geophysics investigation. In some cases, it is possible to take into account in the reconstruction process, informations about the body, leading to the problem of identifying inclusions, that is the problem actually treated in this work. This inverse problem can be solved by the minimization of a function, defined as the difference between the measured potentials and the computed ones for a given conductivity distribution. The present work describes a strategy to solve this problem based on the parametrization of the inclusions boundary, whose shape and size is intended to be determined. The minimization problem is solved via Levenberg-Marquardt Method and the forward one is solved via Boundary Elements Method. In order to evaluate the performance of the proposed strategy, numerical experiments with inclusions of boundaries defined by splines, problems with noisy data, current injection and potential measurement protocols and an application of the strategy to the cardiac function monitoring are presented.

Submitted by Renata Lopes (renatasil82@gmail.com) on 2017-05-15T14:59:59Z No. of bitstreams: 1 marcoshenriquefonsecaribeiro.pdf: 12873424 bytes, checksum: 2b2b91fd2a9726856a0486afa760fe2c (MD5) / Approved for entry into archive by Adriana Oliveira (adriana.oliveira@ufjf.edu.br) on 2017-05-17T16:01:12Z (GMT) No. of bitstreams: 1 marcoshenriquefonsecaribeiro.pdf: 12873424 bytes, checksum: 2b2b91fd2a9726856a0486afa760fe2c (MD5) / Made available in DSpace on 2017-05-17T16:01:12Z (GMT). No. of bitstreams: 1 marcoshenriquefonsecaribeiro.pdf: 12873424 bytes, checksum: 2b2b91fd2a9726856a0486afa760fe2c (MD5) Previous issue date: 2016-10-03 / A Tomografia por Impedância Elétrica (TIE) consiste em uma técnica onde imagens são construídas a partir da injeção de uma corrente elétrica em determinado meios, seguida da leitura de valores de potencial elétrico em pontos do contorno externo de tal domínio. Desta maneira, conhecendo-se ou estimando-se a condutividade elétrica de regiões internas ao meio, pode-se inferir aspectos geométricos da composição do mesmo. Trabalhos na literatura aplicam esta técnica ao contexto de obtenção de imagens do tórax humano, com objetivo de estimar a geometria das cavidades cardíacas de um determinado paciente. O objetivo final de estudo deste trabalho, dentro do contexto de aplicação da TIE à obtenção de cavidades cardíacas, é propor uma metodologia para a estimação da Fração de Ejeção Cardíaca, ou simplesmente Fração de Ejeção (FE), que consiste em medir o percentual de volume de sangue expulso dos ventrículos ao final de um ciclo de batimento do coração. Este trabalho visa evoluir outros trabalhos já existentes que modelam o problema acima descrito como sendo um problema inverso, de otimização, onde se pretende minimizar a diferença entre valores de potencial elétrico medidos e valores simulados por modelos computacionais. A evolução se dá em níveis diferentes. No primeiro nível, é feito um avanço sobre as técnicas de otimização para a resolução do problema inverso, em sua formulaçãobidimensional. Paratal, épropostaumametaheurísticaqueauxiliamétodosde buscanaobtençãodevaloresmaisacurados. Estametaheurísticaéapresentadaemversões sequencial e paralela. São apresentados resultados computacionais de testes realizados para este primeiro nível. Em um segundo nível, é feita a modelagem em três dimensões das mesmas abordagens já encontradas na literatura, que, para a aplicação específica da determinação da FE, até então estão limitadas a modelos bidimensionais. Assim, todo o problema é revisto para uma nova proposta de modelagem, que inclui a criação de modelos geométricos tridimensionais para as regiões de interesse do problema. Como principal contribuição do trabalho neste segundo nível, encontra-se um esquema de parametrização das malhas de polígonos que modelam ventrículos do coração, de forma que se tenha uma maneira compacta de representar as mesmas e, ao mesmo tempo, diminuindo o custo computacional do método de otimização por meio de drástica redução do número de variáveis do problema. Por fim, também é realizado um estudo preliminar da sensibilidade da técnica à presença de ruídos nos dados de entrada. / The Electrical Impedance Tomography (EIT) consists in a technique where images are constructed from the measurements of the electrical potential in some points on the external boundary of some specific domain, caused by the injection of an electrical current in such domain. This way, knowing or estimating the electrical conductivity of some regions inside the domain, geometric aspects of the composition of that domain can be inferred. Works in literature apply this technique to the context of obtaining images from the human thorax, with the objective of estimating the geometry of some cardiac cavities of a specific patient. The final goal of this work, inside the context of the obtention of cardiac cavities, is to propose a methodology for estimating the Cardiac Ejection Fraction, orsimplyEjectionFraction(EF),whichconsistsinmeasuringthepercentualofthevolume of blood expelled from the ventricles at the end of a heart beat cicle. This work intends to evolute previous works, that models the above mentioned problem as an inverse problem, an optimization problem, where the intention is to minimize the difference between the values of measured electrical potentials and the values obtained through simulation using computational models. This evolution occurs in different levels. In the first level, is performedanimprovementoverthepre-existentoptimizationtechniquesforthesolutionof theinverseproblem,inatwodimensionalversion. Forthis,isproposedametaheuristicthat assistssearchmethodstowardstheobtentionofmoreaccuratedvalues. Suchmetaheuristic is presented in sequential and parallel versions. Computational results for performed tests for this level are presented. In a second level, a three dimensional modeling of the same approaches found in literature is done. Those approaches, for the specific application of determining the EF, are so far limited to two dimensional models. Therefore, the whole problem is reviewed in order to propose a new model, which includes the creation of three dimensional geometric models for the regions of interest of the problem. As the main contribution of this work in that second level, there is a parameterization schema of the polygon meshes that model heart ventricles, so that it provides a compact way of representing such meshes, and, at the same time, a way of reducing the computational cost of the optimization method by means of a drastic reduction of the number of variables of the problem. Finally, a preliminary study of the sensibility of the technique to the presence of noise in the input data is also performed.

The diploma thesis is focused on an adaptive control, especially in a self-tuning controller area. The thesis is divided into two main parts. The first part deals with identification methods of the adaptive control. The recursive least squares algorithm and the neural network method are the most popular identification methods. These methods are contrasted and the identification quality evaluation is done. The second part compares various types of the adaptive controllers with a non-adjustable PSD controller. The adaptive controllers are investigated from the various identification method point of view. The user graphic interface was realized for the identification and regulatory part. The time behavior of system responses after incoming disturbance and dynamic process changes during simulation systems control is compared with a physical model connected to B&R PLC. The comparison is done with the application of Matlab/Simulink program.

In this thesis we consider constrained systems of equations. The focus is on local Newton-type methods for the solution of constrained systems which converge locally quadratically under mild assumptions implying neither local uniqueness of solutions nor differentiability of the equation function at solutions. The first aim of this thesis is to improve existing local convergence results of the constrained Levenberg-Marquardt method. To this end, we describe a general Newton-type algorithm. Then we prove local quadratic convergence of this general algorithm under the same four assumptions which were recently used for the local convergence analysis of the LP-Newton method. Afterwards, we show that, besides the LP-Newton method, the constrained Levenberg-Marquardt method can be regarded as a special realization of the general Newton-type algorithm and therefore enjoys the same local convergence properties. Thus, local quadratic convergence of a nonsmooth constrained Levenberg-Marquardt method is proved without requiring conditions implying the local uniqueness of solutions. As already mentioned, we use four assumptions for the local convergence analysis of the general Newton-type algorithm. The second aim of this thesis is a detailed discussion of these convergence assumptions for the case that the equation function of the constrained system is piecewise continuously differentiable. Some of the convergence assumptions seem quite technical and difficult to check. Therefore, we look for sufficient conditions which are still mild but which seem to be more familiar. We will particularly prove that the whole set of the convergence assumptions holds if some set of local error bound conditions is satisfied and in addition the feasible set of the constrained system excludes those zeros of the selection functions which are not zeros of the equation function itself, at least in a sufficiently small neighborhood of some fixed solution. We apply our results to constrained systems arising from complementarity systems, i.e., systems of equations and inequalities which contain complementarity constraints. Our new conditions are discussed for a suitable reformulation of the complementarity system as constrained system of equations by means of the minimum function. In particular, it will turn out that the whole set of the convergence assumptions is actually implied by some set of local error bound conditions. In addition, we provide a new constant rank condition implying the whole set of the convergence assumptions. Particularly, we provide adapted formulations of our new conditions for special classes of complementarity systems. We consider Karush-Kuhn-Tucker (KKT) systems arising from optimization problems, variational inequalities, or generalized Nash equilibrium problems (GNEPs) and Fritz-John (FJ) systems arising from GNEPs. Thus, we obtain for each problem class conditions which guarantee local quadratic convergence of the general Newton-type algorithm and its special realizations to a solution of the particular problem. Moreover, we prove for FJ systems of GNEPs that generically some full row rank condition is satisfied at any solution of the FJ system of a GNEP. The latter condition implies the whole set of the convergence assumptions if the functions which characterize the GNEP are sufficiently smooth. Finally, we describe an idea for a possible globalization of our Newton-type methods, at least for the case that the constrained system arises from a certain smooth reformulation of the KKT system of a GNEP. More precisely, a hybrid method is presented whose local part is the LP-Newton method. The hybrid method turns out to be, under appropriate conditions, both globally and locally quadratically convergent.

Master’s thesis describes adaptive optimal controller design which change parameters of algorithm based on the system information regard for optimal criterion. Generally, the optimal controller solves the problem of minimum states vector. Problems of desired value and steady-state error are solved by variation in optimization algorithm.

In this thesis we consider constrained systems of equations. The focus is on local Newton-type methods for the solution of constrained systems which converge locally quadratically under mild assumptions implying neither local uniqueness of solutions nor differentiability of the equation function at solutions. The first aim of this thesis is to improve existing local convergence results of the constrained Levenberg-Marquardt method. To this end, we describe a general Newton-type algorithm. Then we prove local quadratic convergence of this general algorithm under the same four assumptions which were recently used for the local convergence analysis of the LP-Newton method. Afterwards, we show that, besides the LP-Newton method, the constrained Levenberg-Marquardt method can be regarded as a special realization of the general Newton-type algorithm and therefore enjoys the same local convergence properties. Thus, local quadratic convergence of a nonsmooth constrained Levenberg-Marquardt method is proved without requiring conditions implying the local uniqueness of solutions. As already mentioned, we use four assumptions for the local convergence analysis of the general Newton-type algorithm. The second aim of this thesis is a detailed discussion of these convergence assumptions for the case that the equation function of the constrained system is piecewise continuously differentiable. Some of the convergence assumptions seem quite technical and difficult to check. Therefore, we look for sufficient conditions which are still mild but which seem to be more familiar. We will particularly prove that the whole set of the convergence assumptions holds if some set of local error bound conditions is satisfied and in addition the feasible set of the constrained system excludes those zeros of the selection functions which are not zeros of the equation function itself, at least in a sufficiently small neighborhood of some fixed solution. We apply our results to constrained systems arising from complementarity systems, i.e., systems of equations and inequalities which contain complementarity constraints. Our new conditions are discussed for a suitable reformulation of the complementarity system as constrained system of equations by means of the minimum function. In particular, it will turn out that the whole set of the convergence assumptions is actually implied by some set of local error bound conditions. In addition, we provide a new constant rank condition implying the whole set of the convergence assumptions. Particularly, we provide adapted formulations of our new conditions for special classes of complementarity systems. We consider Karush-Kuhn-Tucker (KKT) systems arising from optimization problems, variational inequalities, or generalized Nash equilibrium problems (GNEPs) and Fritz-John (FJ) systems arising from GNEPs. Thus, we obtain for each problem class conditions which guarantee local quadratic convergence of the general Newton-type algorithm and its special realizations to a solution of the particular problem. Moreover, we prove for FJ systems of GNEPs that generically some full row rank condition is satisfied at any solution of the FJ system of a GNEP. The latter condition implies the whole set of the convergence assumptions if the functions which characterize the GNEP are sufficiently smooth. Finally, we describe an idea for a possible globalization of our Newton-type methods, at least for the case that the constrained system arises from a certain smooth reformulation of the KKT system of a GNEP. More precisely, a hybrid method is presented whose local part is the LP-Newton method. The hybrid method turns out to be, under appropriate conditions, both globally and locally quadratically convergent.