Coordenação de Aperfeiçoamento de Pessoal de Nível Superior / This Ph.D. Dissertation presents contributions to the study of hybrid multilevel systems, analyzing several issues that compose this system, such as: topological variations of the output inverter stage and input rectifier stage, design methodologies, modulation strategies and control system. This extensive analysis is carried out because these issues are closely related, so that a modification in any element can affect the overall system performance. Initially, due to the enormous flexibility and large complexity to design hybrid multilevel converters, this work proposes a unified analysis for these converters, which are composed of several series-connected cells with distinct voltage levels, modulation strategies, topologies and/or semiconductor technologies. This unified analysis includes a detailed comparative analysis among distinct topologies and a new generalized design methodology. Different configurations of hybrid converters are proposed from this new design methodology, depending on the specifications imposed to the multilevel system. Therefore, this work can be used as a valuable tool to define an adequate hybrid configuration for a specific application. After, this work also proposes a new arrangement to implement the isolated voltage sources of the series-connected cells that compose a hybrid multilevel inverter. This new arrangement is based on the multipulse connection of uncontrolled rectifiers processing distinct power levels, unlike conventional multipulse converters. A new design methodology for this input stage is proposed in this work, defining the phase shifts among the secondaries of the isolation transformer to eliminate dominant harmonics from the current drawn from the utility grid even when the active power levels processed by rectifiers are different. This work also investigates the impact of the hybrid multilevel modulation strategy on the harmonic contents of the output voltages and input currents. After to demonstrate the negative impact of the modulation strategy on the input harmonic performance for some operating points, this Ph.D. Dissertation proposes a new modulation strategy that makes possible, together with an asymmetrical multipulse connection of rectifiers, to eliminate undesired harmonics from the input current in any operating point without affecting the output harmonic performance. Beyond these issues, this work also includes a qualitative analysis about hybrid multilevel inverters operating as actuators in closed-loop systems, to motivate a discussion about this new subject. Finally, several experimental results are presented to demonstrate the practical feasibility of some proposals of this Ph.D. Dissertation. / Esta Tese de Doutorado apresenta contribuições ao estudo de sistemas multiníveis híbridos, investigando vários temas que compõem esse sistema, tais como: variações topológicas do estágio inversor de saída e do estágio retificador de entrada, metodologias de projeto, estratégias de modulação e sistema de controle. Isso se deve principalmente ao fato que esses temas estão intimamente relacionados, de tal forma que uma alteração em qualquer elemento do sistema pode modificar o desempenho do todo. Inicialmente, devido à enorme flexibilidade e grande complexidade para projetar conversores multiníveis híbridos, esse trabalho realiza uma abordagem unificada desses conversores, que são compostos de várias células em série, com valores de tensão, estratégias de modulação, topologias e/ou tecnologias de semicondutores diferentes. Essa abordagem inclui uma análise comparativa entre várias topologias e uma nova metodologia de projeto generalizada. Assim, esse trabalho pode ser usado como uma importante ferramenta para definir um conversor híbrido adequado para uma determinada aplicação. Posteriormente, esse trabalho também propõe um novo arranjo para implementar as fontes de tensão isoladas das células que compõem um inversor multinível híbrido. Esse novo arranjo é baseado na conexão multipulso de retificadores não controlados que, ao contrário dos conversores multipulso convencionais, processam níveis distintos de potência. Uma nova metodologia de projeto generalizada é proposta nesse trabalho para esse estágio de entrada, determinando os ângulos de defasagem entre os secundários do transformador de isolação para eliminar harmônicas dominantes da corrente drenada da rede pública de energia, mesmo quando os retificadores processam níveis distintos de potência ativa. Esse trabalho também investiga o impacto da estratégia de modulação híbrida tanto no conteúdo harmônico das tensões de saída quanto no conteúdo harmônico das correntes de entrada. Após demonstrar o impacto negativo da estratégia de modulação no desempenho harmônico de entrada em alguns pontos de operação, essa Tese propõe uma nova estratégia de modulação que torna possível, juntamente com uma conexão multipulso assimétrica de retificadores, eliminar harmônicas dominantes da corrente de entrada em qualquer ponto de operação sem prejudicar o conteúdo harmônico das tensões de saída. Além desses assuntos, esse trabalho também realiza uma breve análise qualitativa do desempenho de inversores multiníveis híbridos em sistemas de controle em malha fechada, com o intuito de iniciar uma discussão sobre esse novo tema. Por fim, diversos resultados experimentais, tanto do estágio inversor de saída quanto do estágio retificador de entrada, são apresentados para demonstrar a viabilidade prática de algumas propostas dessa Tese de Doutorado.

Thesis submitted in fulfilment of the requirements for the degree Doctor of Technology: Electrical Engineering in the Faculty of Engineering at the Cape Peninsula University of Technology / This research focuses on the predictive current control of multilevel converters with the aim of providing an optimized system for three-phase, multilevel inverters (MLIs) so that the load current and the voltage of the capacitors can be controlled. A model predictive current control algorithm is proposed, specifically directed at the utilisation of power obtained from renewable energy systems (RESs). The model was developed for three-phase, multilevel voltage source inverters (MLVSIs), three-phase, three-level diode-clamped converters (DCCs) and flying capacitor converters (FCCs). In this study the renewable energy systems model is used to investigate system performance when power is supplied to a resistiveinductive load (RL-load). The proposed control method was split into two different control algorithms. Firstly, a finite set-model predictive current control (FS-MPCC) method was developed to control the output current of three-phase, MLIs. This control method was selected to reduce the calculation effort for model predictive control (MPC) and to increase the possible prediction horizon. Secondly, to solve the flying capacitor voltage balance problem in an FCC, as well as to solve the DC-link capacitor voltage balance problem in a DCC, a hysteresis-voltage alancing algorithm based on predictive control, was designed—this algorithm was used to keep the flying capacitor voltages and DC-link capacitor voltages within their hysteresis bands. Finally, for some classes of power converters, a performance evaluation of the FS-MPCC method for three-phase, three-level MLIs was investigated in terms of power quality and dynamic response. The improvement was assessed in terms of total harmonic distortion (THD) of the output voltage for the RL-load. The modelling and co-simulation were carried out using MATLAB/Simulink with PSIM software. The co-simulation results indicated that the proposed control algorithms achieved both high performance and a high degree of robustness in RESs applications.

Doctor of Philosophy / Department of Electrical and Computer Engineering / Behrooz Mirafzal / Corrective schemes for facilitating continued operation of dc-ac converters during internal and external abnormalities are presented in this dissertation. While some of the developed techniques are suited for any dc-ac converter topology, most of the presented methodologies are designed specifically for cascaded H-bridge (CHB) multilevel converters. While CHB provide increased scalability and efficiency compared to traditional topologies, these converters are more likely to experience internal faults due to the additional components required. Realizing the full potential of CHB converters requires fault tolerant techniques, such as those demonstrated in this dissertation. Adaptive sinusoidal pulse width modulation (ASPWM) is introduced in this dissertation as a method which enables CHB to directly utilize time-variant dc sources, increasing CHB flexibility when compared to traditional pulse width modulation (PWM) methods which require dc sources with equal magnitudes or with magnitudes existing in specific ratios. Two alternative algorithms are presented to enable ASPWM implementation, providing a trade off between system performance and required sensor circuitry. This dissertation also introduces a load independent analytical approach for identifying discordant operating points, i.e. operating points where some cells in an asymmetric CHB leg regenerate power while the overall leg delivers power, or vice-versa. Identification of these points is essential due to the deleterious effects which can result from extended discordant operation, for instance overcharging of batteries leading to lifespan degradation or even catastrophic failures such as fires or explosions. Additionally, a method for rapidly identifying, isolating, and verifying internal IGBT open-circuit and gate-driver faults is presented in this dissertation to address the increased probability of switch failures occurring within CHB. The proposed approach enables converter operation to continue in the event of gate-driver or open-circuit faults, but avoids unnecessary converter reconfiguration due to gate-misfiring faults. For a CHB leg with M cells, the proposed technique identifies and isolates open-circuit switch faults in less than 2M measurement (sampling) cycles, and verification is completed in less than one full fundamental cycle. Furthermore, this dissertation introduces a real-time implementable atypical PWM technique which enables increased dc bus utilization under a wide range of non-ideal operating conditions. While this approach is suitable for a wide range of converters operating under external abnormalities, for instance maximizing dc bus utilization for converters providing auxiliary services such as negative-sequence compensation, this approach also facilitates operation of CHB with faulty cells. The proposed method can be used with any control technique and any carrier-based PWM method, enabling its implementation in both symmetric and asymmetric CHB. In addition to these fault tolerant techniques, a novel approach for analyzing the active- and reactive-power deliverable by grid-interactive converters is proposed. This method facilitates performance comparisons for various converter configurations, simplifying the process for selecting filter components, dc bus voltages, and other system parameters. This analytical approach also enables converter performance to be analyzed during internal and external fault events, allowing assessment of converter robustness. The efficacy of the developed techniques are supported by MATLAB/Simulink simulations as well as experimental data obtained using a laboratory-scale cascaded H-bridge multilevel converter.

This dissertation presents voltage stabilization techniques for flying capacitors used in soft-switching multilevel active power filters. The proposed active filter has proved to be a solution for power system harmonics produced by static high power converters. However, voltage unbalance of the clamping capacitors in the active filter in practical applications was observed due to its unequal parameters. Thus, the fundamentals of flying capacitors were characterized dealing with voltage balancing between flying capacitors and dc capacitors under practical operation, rather than ideal conditions. The study of voltage balancing provides the fundamental high-level solutions to flying capacitor based multilevel converter and inverter applications without additional passive balancing circuits. The use of proposed voltage balancing techniques made it possible to have a simple structure for solving the problems associated with the conventional bulky passive resistors and capacitor banks. Furthermore, the proposed control algorithms can be implemented with a real time digital signal processor. It can achieve the high performance of the active filter by compensating an adaptive gain to the controller. The effectiveness of the proposed controller was confirmed through various simulations and experiments. The focus of this study is to identify and develop voltage stabilization techniques for flying capacitors used in a proposed active filter. The voltage unbalance is investigated and characterized to provide safe operations. After having defined the problems associated with the voltage unbalance, the most important voltage stabilization techniques are proposed to solve this problem, in conjunction with an instantaneous reactive power (IRP) control of an active filter. In order to reduce the switching losses and improve the efficiency of the active filter, the proposed soft-switching techniques were evaluated through simulation and experimentation. Experimental results indicate that the proposed active filter achieved zero-voltage conditions in all of the main switches and zero-current turn-off conditions to the auxiliary switches during commutation processes. Also, various studies on soft-switching techniques, multilevel inverters, control issues and dynamics of the proposed active filter are discussed and analyzed in depth. / Ph. D.

Over the past two decades, power consumption has increased exponentially worldwide, posing new challenges to power grids to meet the load requirements. With this growing power demand, the need for efficient high-density medium-voltage (MV) power converters has increased to support flexible power distribution grids. The modular multilevel converters (MMC) became the most typical MV power converters in applications from 2010. This topology has many advantages, such as voltage scalability, excellent output performance, and low voltage ratings for switching devices. However, without the excellent reliability of the MMC, applications cannot reap these benefits. The MMC topology comprises several series-connected submodules (typically a half-bridge or a full-bridge inverter). As a result of increased switching devices, the converter becomes vulnerable since a single device fault can disrupt the whole converter operation. Therefore, fault-tolerant strategies to replace faulty SM with a redundant SM are developed using additional bypass switches. Conventionally TRIACs and vacuum switches are employed as bypass switches that operate in the range of 2-10 microseconds. Despite having performance advantages, MMCs are still not fully employed in aerospace and naval industries due to their enormous size. Many Power Electronics Building Blocks (PEBB) are proposed, with size optimization, as submodules for modular converters. The PEBB1000, a 1000 V- PEBB proposed by Dr. Jun Wang, achieved a significant size reduction of 80% with a novel switching cycle control (SCC) scheme. This novel control scheme requires high switching frequency and high di/dt-currents for MMC operation. Due to di/dt-rate limitations, TRIAC-based switch cannot perform bypass operation. Therefore, research work has been conducted on bypass switches for PEBB1000 using wide-bandgap SiC devices. This thesis presents the design of a SiC MOSFET-based bypass switch for PEBB1000 in MMC application. A detailed fault case analysis is presented to show the feasibility of the bypass operation for 90% PEBB-level faults. Significant variations in PEBB1000 bypass requirements are observed through SCC-based MMC simulations. Accordingly, a 1700 V, 100 A bypass switch has been designed using the anti-series topology of MOSFETs. Various specifications, such as 142 nanoseconds operation time, 500 nanoseconds bypass commutation time, and 277A transient current conduction capability, are validated through practical tests. Results prove that SiC-MOSFETs work better than TRIACs in high di/dt-current conduction and operation times. For future work, false-triggering endurance has to be analyzed for 1000 V switching voltage. / Master of Science / When a building is on fire, the safety of people inside depends on the timely arrival of the fire rescue departments. Similarly, for an electrical fault, the safety of electrical systems depends on fast and secure fault protection devices. This thesis presents work on one such fault-protection device used in the power distribution grid: solid-state bypass switch. Distribution grids supply power majorly to households and industries at the city or state level. They employ medium-voltage (MV) converters to step down the voltages to meet the distribution requirements. In MV converters, several low-voltage modules are connected in series to achieve the high-voltage power conversion. When a fault occurs at one of the low-voltage modules in MV converters, power flow gets disrupted due to a series connection like a chain. Therefore, bypass switches are connected in parallel to low-voltage modules for an alternate power flow path. Conventionally used bypass switches have 2-10 microseconds operation time. Recent advancements in semiconductor devices, SiC MOSFETs, allow operation times less than one microsecond. Therefore, research work has been conducted on bypass switches using SiC MOSFETs. Finally, the SiC-MOSFET based bypass switch is built and tested according to converter requirements. Results proved that the designed switch operates in 142 nanoseconds, ten times faster than a conventional switch.

The modular multi-level converter (MMC) is the most prominent interface converter used between the HVDC grid and the HVAC grid. One of the important design challenges in MMC is to reduce the capacitor size. In the current practice, a rather large capacitor bank is required to store line-frequency related circulating energy, even though a number of control strategies have been introduced to reduce the capacitor voltage ripples. In the present paper, a novel control strategy is proposed by means of harmonic injections in conjunction with gain control to completely eliminate both the line frequency and the second-order harmonic of the capacitor voltage ripple. Ideally, the proposed method works with the full bridge topology. However, the concept also works with half bridge topology with a significant reduction of line frequency related ripple. To gain a better understanding of the nature of circulating energy and the means of reducing it, the method of state plane analysis is employed to offer visual support. In addition, the design trade-off between full bridge MMC and half bridge MMC is presented and a novel control strategy for a hybrid MMC is proposed. Finally, the work is supported with a scaled down hardware demonstration. / Master of Science / The modular multi-level converter (MMC) is the most prominent interface converter used between the HVDC grid and the HVAC grid. One of the important design challenges in MMC is to reduce the capacitor size. In the current practice, a rather large capacitor bank is required by the commonly used control strategy. The large capacitor bank increases the cost and the space of the MMC system. In the present paper, a novel control strategy is proposed to significantly reduce the capacitor bank in the system. Ideally, the proposed method works with the full bridge topology. However, the concept also works with half bridge topology with a significant reduction of capacitor bank. To gain a better understanding of the nature of the operating principles of capacitors, the method of state plane analysis is employed to offer visual support. In addition, the design trade-off between full bridge MMC and half bridge MMC is presented and a novel control strategy for a hybrid MMC is proposed. Finally, the work is supported with a scaled down hardware demonstration.

Recent research into industrial applications of electric power conversion shows an increase in the use of renewable energy sources and an increase in the need for electric power by the loads. The Medium-Voltage DC (MVDC) concept can be an optimal solution. On the other hand, the Modular Multilevel Converter (MMC) is an attractive converter topology choice, as it has advantages such as excellent harmonic performance, distributed energy storage, and near ideal current and voltage scalability. The fault response, on the other hand, is a big challenge for the MVDC distribution systems and the traditional MMCs with the Half-Bridge submodule configuration, especially when a DC short circuit fault happens. In this study, the fault current behavior is analyzed. An alternative submodule topology and a fault operation control are explored to achieve the fault current limiting capability of the converter. A three-phase SiC-based MMC prototype with the Full-Bridge configuration is designed and built. The SiC devices can be readily adopted to take advantage of the wide-bandgap devices in MVDC applications. The Full-Bridge configuration provides additional control and energy storage capabilities. The full in-depth design, controls, and testing of the MMC prototype are presented, including among others: component selection, control algorithms, control hardware implementation, pre-charge and discharge circuits, and protection scheme. Systematical tests are conducted to verify the function of the converter. The fault current behavior and the performance of the proposed control are verified by both simulation and experiment. Fast fault current clearing and fault ride-through capability are achieved. / Master of Science / Recent research into industrial applications of electric power conversion shows an increase in the use of renewable energy sources and an increase in the need for electric power by the loads. The Medium-Voltage DC (MVDC) concept can be an optimal solution. On the other hand, the Modular Multilevel Converter (MMC) is an attractive converter topology choice. The fault response, on the other hand, is a big challenge for the MVDC distribution systems and the traditional MMCs, especially when a DC short circuit fault happens. In this study, the fault current behavior is analyzed. An alternative submodule topology and a fault operation control are explored to achieve the fault current limiting capability of the converter. A three-phase SiC-based MMC prototype with the Full-Bridge configuration is designed and built. The full in-depth design, controls, and testing of the MMC prototype are presented. Systematical tests are conducted to verify the function of the converter. The fault current behavior and the performance of the proposed control are verified by both simulation and experiment. Fast fault current clearing and fault ride-through capability are achieved.

Conselho Nacional de Desenvolvimento Científico e Tecnológico / This study proposes a modulation strategy based on geometric approach for a hybrid multilevel converter with single dc voltage source. This multilevel converter is composed of a three-phase three-level multilevel converter with neutral-point-clamped (NPC) connected in series with single-phase three-level full-bridge cells. The NPC converter operates at the fundamental frequency of the reference signals, and the switching angle of the switches is calculated to ensure that the fundamental component of the NPC converter output voltage is equal to the fundamental component of the reference voltage of the entire hybrid multilevel converter. Therefore, the full-bridge cells, which operate with pulse width modulation (PWM), do no process active power and they can be supplied by capacitors. However, to guarantee that the full-bridge cells operate in the linear region a common mode voltage is added on its reference signals. This work uses a systematic methodology for obtaining the region of possible values of common mode voltage to ensure the operation of the full-bridge cells in the linear region, it being possible to maintain the DC bus voltages regulated without distorting the output line-to-line voltages. Moreover, this study applies the hybrid multilevel converter with single DC source as input stage (rectifier) of medium-voltage variable speed drive systems, becoming an alternative for applications that require energy regeneration, high efficiency and input power factor as well as a satisfactory dynamic response. Simulation and experimental results are included to validate the analyzes presented in this work. / Este estudo propõe uma estratégia de modulação baseada em abordagem geométrica para um conversor multinível híbrido com única fonte de tensão contínua. Esse conversor multinível é formado por um conversor multinível trifásico três níveis com ponto neutro grampeado (NPC Neutral Point Clamped) conectado em série com células em ponte completa monofásicas de três níveis. O conversor NPC opera na frequência fundamental dos sinais de referência, e o ângulo de comutação dos interruptores é calculado para garantir que a componente fundamental da tensão de saída do conversor NPC seja igual à componente fundamental da tensão de referência de todo o conversor multinível híbrido. Dessa forma, as células em ponte completa, que operam com modulação por largura de pulso, não processam potência ativa e podem ser alimentadas por capacitores. Contudo, para que as células em ponte completa operem na região linear é adicionada uma tensão de modo comum nos seus sinais de referência. Esse trabalho utiliza uma metodologia sistemática para obtenção da região de possíveis valores de tensão de modo comum que garantem a operação das células em ponte completa na região linear, sendo possível manter as tensões nos capacitores dos barramentos CC reguladas sem distorcer as tensões de linha de saída. Ainda, esse trabalho aplica o conversor multinível híbrido com uma única fonte de tensão contínua como estágio de entrada (retificador) de sistemas de acionamento de velocidade variável para motores de média tensão, sendo uma alternativa para aplicação que demandam regeneração de energia, elevado rendimento e fator de potência de entrada, assim como uma resposta dinâmica satisfatória. Resultados de simulação e experimentais são incluídos para validar as análises apresentadas nesse trabalho.

Coordenação de Aperfeiçoamento de Pessoal de Nível Superior / This Master Thesis presents a methodology to compare different topologies and configurations of multilevel systems applied to drive medium voltage induction motors. Initially, it presents the input and output parameters that must be supplied by the power drivesystems manufacturer. Then, the mathematical models of the performance indexes used are presented, being, the total harmonic distortion, first order distortion factor and the power losses of the main semiconductors devices. After, it is presented a methodology that allows the amplitude selection of DC sources for the inverter with a specific number of cascaded Hbridge cells. A study regarding the hybrid multilevel modulation strategies and the impact of their variations on the semiconductors devices power losses of the output inverter is developed. Based on this study, it is proposed a new hybrid multilevel modulation technique, it minimizes the semiconductors power losses and allow the use of front-end uncontrolled rectifiers without modifying the total harmonic distortion of the output voltage inverter. After, it is developed a comparative study between the asymmetrical hybrid multilevel inverter and the symmetrical multilevel inverter, which enables the determination of the switching frequency for the one determined performance and the maximum switching frequency of each converter. This Master Thesis presents complementary methods of design of the hybrid multilevel inverter and a methodology that will make possible to choose the drive system that presents high efficiency and reduced cost, neither penalizing the energy quality of the utility grid nor the induction motor constraints. / Esta Dissertação de Mestrado apresenta uma metodologia para comparar diferentes topologias e configurações de sistemas multiníveis aplicados ao acionamento de motores de indução de média tensão. Inicialmente são apresentados os parâmetros de entrada e saída que devem ser fornecidos pelos fabricantes de sistemas de acionamento de alta potência. Então, são apresentados os modelos matemáticos dos índices de desempenho utilizados, sendo eles, distorção harmônica total, fator de distorção de primeira ordem e perdas nos dispositivos semicondutores principais. Posteriormente, é apresentada uma metodologia que permite selecionar a amplitude das fontes CC para um inversor com um número específico de células H-bridge conectadas em série. Um estudo a respeito das estratégias de modulação multinível híbrida e o impacto de suas variações sobre as perdas dos dispositivos semicondutores do inversor de saída é apresentado. Fundamentado neste estudo, é proposta uma nova técnica de modulação multinível hibrida, que minimiza as perdas nos semicondutores e permite o uso de retificadores não controlados no estágio de entrada sem modificar a distorção harmônica total da tensão de saída do inversor. Depois, é desenvolvido um estudo comparativo entre o inversor multinível híbrido assimétrico e o inversor multinível simétrico, que permite a determinação da freqüência de comutação para um rendimento específico e a máxima freqüência de comutação de cada conversor. Esta Dissertação de Mestrado apresenta métodos complementares de projeto de inversores multiníveis híbridos e uma metodologia que possibilitará escolher sistemas de acionamento que apresentem alta eficiência e custo reduzido, sem por isso prejudicar a qualidade da energia drenada da rede pública e fornecida ao motor de indução.

La topologie modulaire multiniveaux est une structure d'électronique de puissance construite par la mise en série de sous-modules identiques, composés chacun d'une cellule de commutation et d'un condensateur. Un tel système de conversion pouvant comporter un grand nombre de cellules permet d'augmenter le rendement global et la qualité des signaux en sortie. De plus, il permet d'utiliser des composants basse tension présentant un meilleur comportement dynamique et un rapport qualité-prix bien supérieur aux composants moyenne tension. Il permet également, par rapport aux structures conventionnelles, une grande souplesse pour la conception et la fabrication du fait de son aspect modulaire, tout en s'affranchissant d'un transformateur volumineux et onéreux en entrée. Comparé aux autres types de topologies, avantageuses avec un nombre limité de niveaux, le convertisseur modulaire multiniveaux semble être mieux adapté aux applications en moyenne et haute tensions, qui sont tributaires de l'association des composants de puissance. Néanmoins, pour la variation de vitesse, un certain nombre de défis technologiques ont été mis en évidence, compte tenu du fonctionnement particulier de l'onduleur modulaire multiniveaux et des contraintes imposées par l'opération en très basse fréquence. En le fonctionnement normal, la forme d'onde des courants internes, contrairement aux autres types de topologies, n'est pas symétrique en raison de la distribution du courant direct dans chaque bras. Cela entraîne une disparité significative en termes de dissipation thermique parmi les interrupteurs constituant un sous-module. Avec le choix d'une technologie de packaging 3D, la possibilité de refroidir les puces semi-conductrices en double-face offre une meilleure capacité de refroidissement et une nouvelle perspective de conception des modules pour cette application. Un nouveau concept de report de puces est présenté et un prototype de tel module a été réalisé, modélisé et caractérisé. Il permet d'équilibrer globalement la chaleur dissipée par les puces sur les deux faces du module, problème inhérent à l'emploi de structure 3D. Conjugué à la mutualisation d'un interrupteur par deux puces en parallèle, la nouvelle architecture a aussi pour objectif d'équilibrer le refroidissement double-face dans le temps. En effet, pour les opérations en basse fréquence, les interrupteurs fonctionnent en régime instationnaire avec de forte variation de température, il n'est donc plus possible de compenser les effets thermomécaniques de chaque composant l'un par l'autre, comme en régime stationnaire et avec un positionnement planaire des puces. D'autre part, d'un point de vu systémique, la stratégie de commande et le dimensionnement des condensateurs flottants de l'onduleur modulaire multiniveaux sont deux aspects intimement liés. En effet, les condensateurs flottants sont le siège d'ondulations de tension de très forte amplitude. Cela a pour effet de déstabiliser l'onduleur, voire de provoquer la destruction des composants en atteignant des niveaux de tension trop élevés. Ainsi, des contrôleurs judicieusement conçus permettent de réduire les ondulations indésirables, et a fortiori, d'embarquer des capacités moins importantes dans le système, tant que ces dernières sont inversement proportionnelles à l'ondulation de la tension. Afin d'avoir une compréhension approfondie sur les dynamiques régissant le convertisseur modulaire multiniveaux, un modèle dynamique global basé sur la représentation d'état a été établi. Bien que cette représentation soit limitée à l'harmonique 2 des grandeurs caractéristiques, elle permet une fidèle interprétation du mécanisme de conversion sans passer par des modèles énergétiques bien plus complexes à exploiter, et de proposer des lois de commande montrant leur efficacité notamment autour de la fréquence nominale. Cela a été vérifié sur une maquette de puissance réalisée dans le cadre de cette thèse. / Multilevel modular topology converts energy between two direct and alternative endings. This structure is constructed by the series connection of identical sub-modules, composed of a switching cell and a floating capacitor, and with arm inductors. Such a conversion system may reach a large number of levels increases the overall efficiency and quality of the output signals. In addition, it allows the use of low voltage components with better dynamics and cost effectiveness above the high voltage components. It also allows flexibility in the work of design and manufacture due to its modularity, while avoiding a bulky and expensive input transformer, regarding the conventional technology. Compared with other types of topologies, advantageous with a limited number of levels, the modular multilevel converter seems to be more suited for medium and high voltage applications, which are dependent on the association of power components. However, for variable speed drive application, a certain number of technological challenges have been highlighted, given the specific functional characteristics of the modular multilevel inverter and the constraints imposed by the very low frequency operation. On the one hand, for the normal operation of a multilevel modular converter, the waveform of the internal currents, in contrast to other types of topologies, is not symmetrical due to the distribution of the direct current in each phase leg. This may entail a significant disparity in terms of heat dissipation within the switching devices constituting a sub-module. Therefore, the problem of thermal management of active components is emphasized in the use of a modular multilevel converter. With the choice of a 3D packaging technology, interconnection by bumps, the ability to cool the semiconductor chips through the both sides of a module offers better cooling effects and a new perspective to design the power module for the studied structure. The concept of laying chips on both the two substrates of module without facing each other provides overall balanced dissipation in the space and permit to overcome the unbalanced heat distribution induced by bumps. Combined with the sharing of a switch by two chips in parallel, the proposal of the new architecture for 3D power module also aims to balance the double-sided cooling in the time range. Indeed, for the very low frequency operation, the switches operate in unsteady state where each switch has its own thermal behavior, it is no longer possible to compensate the thermo-mechanical constraints over each component with the help of the others, as in steady state and with a planar chips positioning scheme. On the other hand, from a systemic point of view, the control strategy and the dimensioning of floating capacitors of modular multilevel inverter are two interrelated aspects. Because the floating capacitors, having the role of energy sources, are loaded / unloaded through the modulation period, which causes very high voltage ripples across those capacitors with a very low frequency. This will destabilize the inverter and even provoke the destruction of components by approaching too high voltage levels. Thus, wisely designed controllers reduce unwanted ripples and, furthermore, allow embarking much smaller capacity in the system, as they are inversely proportional to the voltage ripple. In order to have a thorough understanding on the dynamics governing the modular multilevel converter, a comprehensive dynamic model based on state-space representation was established. Although this representation is limited to the second harmonic of characteristic variable, it allows a faithful interpretation of the conversion mechanism without using energy models, more complex to operate, and control laws can also be proposed and their effectiveness around the nominal frequency has been underlined. Concerning the very low frequency operations, another solution has been proposed and is ongoing patent pending.