Orientador: Dr. José L. Azcue Puma / Dissertação (mestrado) - Universidade Federal do ABC, Programa de Pós-Graduação em Engenharia Elétrica, 2016. / Este trabalho apresenta o estudo, modelagem e projeto de um sistema de conversão de energia para a conexão de um módulo fotovoltaico com a rede elétrica. Nesta pesquisa são abordados três conversores para adequar a energia gerada por um painel fotovoltaico e conectá-la à rede elétrica. O primeiro conversor é um Boost intercalado de dois níveis encarregado de implementar o MPPT (Maximum Power Point Tracking), o segundo conversor é um Ressonante LLC que realiza o isolamento galvânico e o terceiro conversor é um inversor em ponte completa que mantém a tensão de barramento em um valor determinado e realiza a conversão de corrente contínua para corrente alternada. Os três conversores são projetados, calculando os elementos passivos que os compõem e simulando os circuitos obtidos. Também, é realizada a modelagem dos conversores e o projeto dos controladores utilizando a análise de pequenos sinais. Os resultados de simulação são apresentados utilizando os softwares PSIM e MATLAB. Na parte experimental são verificados os dois primeiros conversores usando o kit High Voltage Isolated Solar MPPT fabricado pela Texas Instruments. / This work presents the study, modeling and design of a photovoltaic conversion system for the connection of a photovoltaic module with the grid. In this research three converters are approached to adequate the generated energy for a photovoltaic panel and to transfer the energy to the grid. The first converter is an interleaved two-level boost responsible for implementing the MPPT (Maximum Power Point Tracking) the second converter is a Resonant LLC that performs galvanic isolation and the third converter is a complete bridge inverter that keeps the bus voltage at a certain value and performs the conversion of continuos current to alternating current. The three converters are designed, calculating the passive elements that compose them and simulating the obtained circuits. Also, the modeling of the inverters and the design of the controllers are performed using small signal analysis. Simulation results are presented using the PSIM and MATLAB software. In the experimental part the first two inverters are verified using the High Voltage Isolated Solar MPPT kit manufactured by Texas Instruments.

Coordenação de Aperfeiçoamento de Pessoal de Nível Superior / This Master Thesis presents the study and proposes a digital control system applied to the PFC boost converter operating in mixed conduction mode. The motivation to this subject, a bibliographic review of the existing current control techniques and the objectives of this work are presented in the first chapter. The control system is composed of a current loop, which has the objectives to provide low harmonic content to the input current and high power factor to the converter, and a voltage loop, which has the goal to regulate the output voltage of the converter. Initially, the study of the conduction modes of the PFC boost converter, highlighting their operation conditions, and details of the mixed conduction mode operation are showed. Next, two predictive digital current controllers with feedforward action are approached in details, including the equations deduction, block diagram, operation process, design, simulation results and parametric variation analysis. A comparative analysis between both controllers is also accomplished. The proportional-integral voltage controller is presented following, comprising its structure, design and simulation results. After, issues related to the practical implementation of the converter and control system, as the converter design and the control device programming, are shown. Finally, the experimental results obtained through the implementation of the proposed system, including the waveforms acquired using a prototype constructed in laboratory and the quality indexes as the power factor and the total harmonic distortion, are presented with the purpose to validate and prove the content of this work. / Esta Dissertação de Mestrado apresenta o estudo e propõe um sistema de controle digital para o conversor boost PFC operando no modo de condução mista. A motivação para este assunto, uma revisão bibliográfica das técnicas de controle de corrente existentes e os objetivos do trabalho são apresentados no capítulo introdutório. O sistema de controle é composto de uma malha de corrente, que tem por objetivo proporcionar baixo conteúdo harmônico para a corrente de entrada e alto fator de potência para o conversor, e uma malha de tensão, que tem por objetivo regular a tensão de saída do conversor. Inicialmente, o estudo dos modos de condução do conversor boost PFC, destacando as suas condições de operação, e detalhes da operação no modo de condução mista são mostrados. Em seguida, dois controladores digitais preditivos de corrente com ação feedforward são abordados em detalhes, incluindo o equacionamento, diagrama de blocos, funcionamento, projeto, resultados de simulação e análise da variação paramétrica. Uma análise comparativa entre ambos também é realizada. O controlador proporcional-integral de tensão é abordado na seqüência, compreendendo sua estrutura, projeto e resultados de simulação. Posteriormente, questões relativas à implementação prática do conversor e do sistema de controle, como o projeto do conversor e a programação do dispositivo de controle, são mostradas. Por fim, os resultados experimentais obtidos com a implementação do sistema proposto, incluindo as formas de onda adquiridas em um protótipo construído em laboratório e os índices de qualidade como o fator de potência e a distorção harmônica total, são apresentados com o intuito de validar e comprovar o conteúdo deste trabalho.

fator de potÃncia e recurso para operar com dois valores de tensÃo de entrada. O mesmo à aplicÃvel a sistemas ininterruptos de energia do tipo dupla conversÃo ou on-line com caracterÃsticas de tensÃo de entrada bivolt (110 Vca e 220 Vca) e desta maneira à descartada a utilizaÃÃo de um autotransformador com seletor de tensÃo. O conversor em estudo à composto por dois conversores CA-CC boost clÃssicos, em que, para uma tensÃo de entrada de 110 Vca as entradas sÃo conectadas em paralelo e para uma tensÃo de entrada de 220 Vca as entradas sÃo conectadas em sÃrie. A ideia à fazer com que se tenha uma divisÃo equilibrada na entrada de cada conversor quando a tensÃo da rede elÃtrica for 220 Vca. Assim cada conversor boost clÃssico recebe metade da tensÃo total de alimentaÃÃo do conversor proposto. A estratÃgia de controle à baseada no controle por modo corrente mÃdia aplicada a ambos os conversores para proporcionar a correÃÃo do fator de potÃncia e a regulaÃÃo da tensÃo de saÃda. Para verificar o estudo teÃrico foi desenvolvido o projeto do circuito de potÃncia e controle validando atravÃs de resultados de simulaÃÃo e experimentais para um protÃtipo de 2,4 kW. Para a conexÃo paralelo e sÃrie das entradas, os resultados obtidos foram satisfatÃrios e o conversor operou adequadamente. / This work presents a study of a Double Boost AC-DC Converter with power factor correction and dual input voltage operation capability via a selector switch. Such converter can be applied to on-line uninterruptible power supplies with dual voltage input characteristics, this way avoiding the usage of a low frequency autotransformer. The studied structure is composed by two AC-DC classical boost converters, in which for input voltage of 110 Vac both its inputs are connected in parallel, and, for 220 Vac, they are connected in series. The control strategy is based in the average current mode control applied to both converters, in order to provide the power factor correction and output voltage regulation. Simulation and experimental results for 2.4 kW are presented, and so are validate the theoretical study and design. Connecting the inputs in parallel and series, the results were satisfactory and the converter operated properly.

Les systèmes de transmission de l’information sans fil connaissent un essor considérable et sont intégrés dans la plupart des systèmes électroniques modernes. De manière plus spécifique, la consommation énergétique de la fonction amplification de puissance RF, qui constitue le cœur de ce travail de recherche, est un enjeu économique et écologique de premier plan. Dans ce sens, ce travail présente une architecture de polarisation de drain dynamique permettant de maintenir un rendement énergétique élevé sur une large dynamique de puissance de sortie. La conception et la réalisation d’un amplificateur de puissance RF large bande, d’un modulateur de polarisation de drain haute fréquence et d’un pilote de grille en technologie GaN sont présentés. L’architecture proposée démontre une amélioration du rendement énergétique global. Une focalisation sur la problématique de couplage non-linéaire entre l’amplificateur de puissance RF et le module d’alimentation agile met en évidence les répercussions de cette méthode sur l’intégrité du signal. Une étroite impulsion de polarisation de grille est appliquée afin d'atténuer l’impact de la polarisation dynamique de drain sur les formes d'onde de l'enveloppe du signal RF amplifié. Une validation expérimentale du démonstrateur proposée est effectuée pour un signal impulsionnel RF multi-niveaux de test. Cette méthode permet de maintenir un facteur de forme de l’enveloppe du signal de sortie RF quasi-rectangulaire sans impact majeur sur les performances globales énergétiques. / Wireless communications are experiencing tremendous growth and are integrated into most modern electronic systems. More precisely, saving energy consumption of RF power amplifier is the core of this thesis work. This work presents a dynamic drain bias architecture used to keep a high efficiency over a large output power range. Design and implementation of a wideband RF power amplifier, a drain supply modulator and a gate driver circuit in GaN technology are presented. The built-in prototype demonstrates an overall efficiency improvement. A specific focus on non-linear interaction between the RF power amplifier and the drain supply modulator highlights the effects of this technique on the output envelope signal shape. A narrow pulse gate bias peaking preceding drain bias voltage variations is applied in order to mitigate drain bias current, voltage overshoot and power droop, thus improving pulse envelope waveforms of the RF output signal. An experimental validation of the proposed demonstrator is performed for a RF pulsed test sequence having different power levels. This way enables to keep rectangular pulse envelope shape at the RF output signal without any major impact on overall efficiency performances.

Le Syndrome Dysgénésique et Respiratoire Porcin (PRRS) est la maladie infectieuse endémique la plus couteuse en élevage porcin dont l'agent responsable est un Arterivirus, le PRRSV, qui présente une grande diversité génétique. L'infection par le PRRSV est fréquemment associée à l'infection par les virus influenza. La vaccination est une méthode de lutte adaptée contre ces virus. Dans le cas du PRRSV, les vaccins les plus utilisés sont des virus vivants modifiés (MLV) qui induisent une immunité protectrice peu efficace contre les variants viraux. Dans le cas du virus influenza, les vaccins inactivés utilisés présentent la même insuffisance.Dans ce travail de thèse, j'ai évalué des stratégies vaccinales visant à induire une immunité efficace contre des variants viraux, en utilisant des antigènes conservés entre souches, adressés aux cellules présentatrices d'antigènes (APC), et j'ai analysé l'effet de différentes voies et modes d'administration.Dans le cas du virus grippal, le ciblage d'antigènes conservés (HA2, M2e, NP) au CD11c a permis d'augmenter la réponse T uniquement lors d'administration par voie intramusculaire (IM) et fut sans effet sur la réponse anticorps. La vaccination par voie intradermique s'est traduit par une exacerbation de la pathologie lors d'une épreuve virale, alors que la vaccination par voie IM a réduit les symptômes, la durée d'excrétion virale en corrélation avec une meilleure réponse anticorps anti-HA2 et M2e.Dans le cas du virus PRRSV qui fut mon sujet principal d'étude, j'ai cherché à optimiser des réponses lymphocytaires T IFNγ en employant une stratégie vaccinale ADN codant des antigènes contenant des épitopes T conservés entre souches, ciblés aux APC. En effet, alors que les mutations virales conduisent à un échappement aux anticorps neutralisants, la réponse lymphocytaire T IFNγ a été proposée impliquée dans la protection croisée. J'ai montré que l'immunogénicité optimale de vaccins ADN PRRSV, conduisant à la réponse T la plus large, est obtenue par l'administration intradermique associée aux nanoparticules de PLGA (NP), suivi d'une électroporation (EP), par rapport à EP seul ou délivrance intradermique ou transcutanée avec des patches à micro-aiguilles résorbables. Cette immunogénicité optimale est associée à une bonne transfection des cellules de la peau, à une accumulation de cellules inflammatoires, et à une mobilisation des cellules dendritiques. J'ai ensuite utilisé ce mode d'administration EP+NP pour immuniser des porcs avec des plasmides codant des antigènes conservés du PRRSV adressés ou non aux APC via CD11c ou XCR1. Les porcs ont été immunisés soit avec des injections répétées d'ADN seul soit en prime-boost ADN-MLV. Le régime ADN-MLV s'est montré supérieur pour l'induction de réponse B et T à celui de l'ADN ou du MLV seuls, et le ciblage aux APC a nettement augmenté la réponse anticorps mais pas la réponse T IFNγ. Dans une expérience suivante à visée d'application sur le terrain, j'ai utilisé le régime ADN-MLV (sans NP cette fois), délivré avec EP ou avec jet sous pression (PJ). Dans ces conditions, la primo-vaccination avec ADN n'a pas significativement augmenté la réponse T IFNγ induite par le MLV, mais elle a clairement augmenté la réponse anticorps avec un bénéfice du ciblage des APC. L'immuno-potentialisation induite par la primo-vaccination ADN n'a pas conduit à l'amélioration de la protection contre une épreuve avec un virus hétérologue et a montré que cette protection n'est au final pas corrélée avec la réponse lymphocytaire T IFNγ et opère en l'absence d'anticorps neutralisants détectables. Enfin, l'ensemble de ce travail montre que l'effet du ciblage des APC chez le porc est influencé par la voie d'administration et par le régime d'administration comme le prime-boost ADN-MLV. / The Porcine Reproductive and Respiratory Syndrome (PRRS) is the most damaging infectious disease in pigs worldwide. The etiologic agent is an Arterivirus, the PRRSV, which presents a large genetic diversity. PRRSV infection is frequently associated with influenza virus co-infection. Vaccination is a highly suitable way to control these viruses. In the case of PRRSV, the most effective commercial vaccines are modified live vaccines (MLV) which induce only a partial protection against heterologous strains. In the case of the influenza virus, the available inactivated vaccines show the same weakness.With the goal to control emerging influenza and PRRSV variants, I evaluated vaccine strategies involving conserved viral antigens between strains which were targeted to antigen-presenting cells (APC) and delivered by different routes and methods.In the case of influenza virus, the targeting of conserved antigens (HA2, M2e and NP) to CD11c led to increased IFNγ T cell responses only when vaccines were delivered by the intramuscular (IM) route and had no effect on the humoral response. The intradermal route exacerbated disease following challenge whereas the IM route reduced the symptoms, the duration of viral excretion in correlation with higher anti-HA2 and anti-M2e antibody responses.In the case of PRRSV, which was my main subject, I sought to optimize the IFNγ T cell responses by using DNA vaccines encoding antigens with conserved T-epitopes between strains, and targeted to APC. Indeed, whereas viral mutants escape neutralizing antibodies, it has been proposed that the IFNγ T cell responses are instrumental for cross-protection. I showed that the broadest T cell responses were induced by DNA vaccines combined to nanoparticles PLGA (NP) injected by the intradermal route, followed by electroporation (EP) compared with EP-only, intradermal route-only or transcutaneous dissolvable microneedles. This optimal immunogenicity was associated with a high transfection level of skin cells, an accumulation of inflammatory cells, and dendritic cells mobilisation. Next I used the EP+NP method to immunize pigs with plasmids encoding conserved PRRSV antigens targeted or not to APC via CD11c or XCR1. Pigs were immunized either with repeated injections of DNA alone or with a prime-boost DNA-MLV. The DNA-MLV regimen induced improved humoral and IFNγ T cell responses compared to DNA alone or MLV alone and the APC-targeting significantly increased the humoral response but not the IFNγ T cell response. Finally, I evaluated the DNA-MLV regimen efficacy, with an applied perspective, using naked DNA without NP and delivered by EP or by a convenient needle free injection technology (PJ). In these conditions, the DNA prime did not significantly increase the IFNγ T cell response induced by the MLV, but clearly increased the humoral response with a benefit of the APC-targeting. However, the immune potentiation induced by the DNA prime did not lead to an improved protection following a heterologous challenge. The heterologous protection was not correlated to the measured humoral and IFNγ T cell responses, and neutralizing antibodies were undetectable. Thus cross-protective effectors have not been sufficiently activated by our DNA-MLV strategy and the immune correlates of protection against heterologous PRRSV are still to be identified to develop cross-protective vaccines. Finally, this work shows that the effect of APC-targeting in pigs is influenced by delivery routes and methods and by vaccine regimen such as the prime-boost DNA-MLV.

This diploma thesis is focused on object detections through general-purpose computing on graphics processor units. There is an explanation of graphics adapters work and basics of their architecture in this thesis. Based on the adapters, there is the effective work in libraries for general-purpose computing on graphics processor units demonstrated in this thesis. Further, the thesis shows the available algorithms for object detection and which ones from them are possible to be effectively parallelized. In conclusion of this thesis, there is a comparison of the object detections speeds to common implementations on classical processors.

<p>Implantable wireless devices are used to treat a variety of diseases that are not able to be treated with pharmaceuticals or traditional surgery, These implantable devices have use in the treatment of neurological disorders like epilepsy, optical disorders such as glaucoma, or injury related issues such as targeted muscle reinnervation. These devices can rely upon harvesting power from an inductive wireless power source and batteries. Improvements to how well the devices utilize this power directly increase the efficacy of the device operation as well as the device's lifetime, reducing the need for future surgeries or implantations. </p> <p> I have designed an improvement to cavity resonator based wireless power by designing a dynamic impedance matching implantable power supply, capable of tracking with device motion throughout a changing magnetic field and tracking with changing powering frequencies. This cavity resonator based system presents further challenges practically in the turn-on cycle of the improved device. </p> <p> I further design a coil-to-coil based wireless power system, capable of dynamically impedance matching a high quality factor coil to optimize power transfer during steady state, while also reducing turn-on transient power required in dynamic systems by utilizing a second low quality factor coil. This second coil has a broadband response and is capable of turning on at lower powers than that of the high quality factor coil. The low quality factor coil powers the circuitry that dynamically matches the impedance of the high quality factor coil, allowing for low power turn on while maintaining high power transfer at all operating frequencies to the implantable device. </p> <p> Finally, an integrated circuit is designed, fabricated, and tested that is capable of smoothly providing regulated DC power to the implantable device by stepping up from wireless power to a reasonable voltage level or stepping down from a battery to a reasonable voltage level for the device. The chip is fabricated in 0.18um CMOS process and is capable of providing power to the "Bionode" implantable device. </p>

The translational self-diffusion, the librations, and the reorientational motions of guest water molecules in the zeolite chabazite are examined by Molecular Dynamics (MD) computer simulations at different temperatures and loadings, including at room temperature, at which the experiments are carried out. Satisfactory agreement is found between the computed and measured translational self-diffusion coefficients. It is, however, furthermore found that the way in which the long-range electrostatic interactions are computed has an effect on the self-diffusion at high loadings and temperatures. The spectral densities of the librational motions of water are found to be similar to those in aqueous salt solutions. The reorientations of the water molecules, on the other hand, are much slower than in the liquids, and very anisotropic. The vector in direction of the molecular dipole moment reorients only very slowly, at the time scale of the simulations, due to the attraction to the almost immobile Ca++-ions and the walls of the zeolite. The other two vectors seem to undergo jump-reorientations rather than reorientations by a diffusion process. Hyper dynamics boost potential method has been applied to the MD simulations to estimate the self-diffusion coefficients of Ca++ ions in dehydrated chabazite. Because of our system is very complicated, the self-diffusion of Ca++ ions can only be roughly estimated. The Ca++ ions diffusion is small enough to confirm that the cation motion can be neglected in the normal MD simulation.

L'intégration actuelle des systèmes photovoltaïques dans les systèmes d'alimentation conventionnels a montré une croissance importante, ce qui a favorisé l'expansion rapide des micro-réseaux du terme anglais microgrid. Cette intégration a cependant augmenté la complexité du système d'alimentation qui a conduit à de nouveaux défis de recherche. Certains de ces défis de recherche encouragent le développement d'approches de modélisation innovantes en temps réel capables de faire face à cette complexité croissante. Dans ce contexte, une méthodologie innovante est proposée et basée sur les composants pour la modélisation et l'émulation de systèmes photovoltaïques en temps réel integers aux microgrids. L'approche de modélisation proposée peut utiliser le langage de modélisation des systèmes (SysML) pour décrire la structure et le comportement des systèmes photovoltaïques intégrés en tenant compte de leurs caractéristiques multidisciplinaires. De plus, cette étude présente le cadre de spécification de haut niveau des systèmes embarqués (HiLeS) pour transformer les modèles SysML développés en code source destinés à configurer le matériel intégré. Cette caractéristique de la generation automatique de code permet de profiter de dispositifs avec un haut degré d'adaptabilité et de performances de traitement. Cette méthodologie basée sur HiLeS et SysML est axée sur l'étude des systems photovoltaïques partiellement ombragés ainsi que des architectures flexibles en électronique de puissance en raison de leur influence sur les microgrids actuels. En outre, cette perspective de recherche est utilisée pour évaluer les stratégies de contrôle et de supervision dans les conditions normales et de défauts. Ce travail représente la première étape pour développer une approche innovante en temps réel pour modéliser et émuler des systèmes photovoltaïques complexes en tenant compte des propriétés de modularité, de haut degré d'évolutivité et des conditions de travail non uniformes. Les résultats expérimentaux et analytiques valident la méthodologie proposée. / Nowadays, the integration of photovoltaic systems into electrical grids is encouraging the expansion of microgrids. However, this integration has also increased the power system complexity leading to new research challenges. Some of these research challenges require the development of innovative modeling approaches able to deal with this increasing complexity. Therefore, this thesis is intended to contribute with an innovative methodology component-based for modeling and emulating in real-time photovoltaic systems integrated to microgrids. The proposed modeling approach uses the Systems Modeling Language (SysML) to describe the structure and behavior of integrated photovoltaic systems. In addition, this study presents the High Level Specification of Embedded Systems (HiLeS) to transform automatically the developed SysML models in embedded code and Petri nets. These characteristics of automatic code generation and design based on Petri nets allow taking advantage of FPGAs for application of real-time emulation of photovoltaic systems. This dissertation is focused on partially shaded photovoltaic systems and flexible power electronics architectures because of their relevant influence on current microgrids. Furthermore, this research perspective is intended to evaluate control and supervision strategies in normal and fault conditions. This work represents the first step to develop an innovative real-time approach to model and emulate complex photovoltaic systems considering properties of modularity, high degree of scalability, and non-uniform working conditions. Finally, experimental and analytical results validate the proposed methodology.

No description available.