The 2007 financial crisis highlighted the severe risks posed by counterparty defaults in financial markets. Assessing and addressing counterparty credit risk has consequently been a focal point of new regulations introduced in the wake of the crisis. The Central Clearing Counterparty (CCP) is at the heart of the solution, an entity dedicated to managing and mitigating counterparty risk in a market. CPPs manage risk by collecting collateral, referred to as margin, from the participants trading on the market. Appropriately sizing the margin is of utmost importance for the CCP to maintain the integrity of its operation and, by extension, protect the participants in the market. Most contemporary margin methodologies require significant resources which precludes frequent margin updates. In light of this issue, our work examines the capability of replicating the popular margin methodology Historical Simulation Value at Risk using machine-learning-based methods envisioning that an adequate such model could be used as a complement to the traditional model, providing real-time margin estimations. The experiment concerns portfolios containing stocks, bonds, and options and uses static market data and scenarios. We conclude that neither of the ensemble methods are sufficiently accurate, while both of the neural network-based models show moderate promise, warranting further development.

This thesis has investigated whether a transformer-based language model can be improved by training the model on context sequences which are input sequences with a larger window of text, by combining a transformer model with a neural network for non-text features, or by domain-adaptive pre-training. Two types of context input sequences are tested: left context and full context. The three modifications are explored by applying BERT models to the Swedish Parliamentary Corpus to classify whether a text sequence is a heading. A standard BERT model is trained for sequence classification alongside a position model which adds an additional feedforward neural network to the model. Each model is trained with- and without context sequences as well as with- and without domain-adaptive pre-training. A standard implementation of the BERT model with domain adaptation achieves an F1 score of 0.9358 on the test set and an accuracy of 0.9940. The best performing standard BERT model with a context input sequence achieves an F1 of 0.9636 and an accuracy of 0.9966 while the best performing position model achieves an F1 of 0.9550 and an accuracy of 0.9957. The best performing model which combines context input sequences with the position model achieves an F1 of 0.9908 and an accuracy of 0.9991 on the test set. Analysis of misclassified sequences suggests that the models with context input sequences and positional features are less likely to misclassify sequences which can appear both as a heading and a non-heading in the corpus. However, a McNemar's exact test indicates that only a position model with left context input sequences differs significantly from its standard BERT counterpart in terms of the number of differing misclassifications at a 5% significance level. Furthermore, there is no experimental evidence that domain-adaptive pre-training improves classification performance on this specific sequence classification task.

The Thermo-Calc software is a key tool in the research process for many material engineers. However, integrating multiple modules in Thermo-Calc requires the user to write code in a Python-based language, which can be challenging for novice programmers. This project aims to enable the generation of such code from user prompts by using existing generative AI models. In particular, we use a retrieval-augmented generation architecture applied to LLaMA and Mistral models. We use Code LLaMA-Instruct models with 7, 13, and 34 billion parameters, and a Mistral-Instruct model with 7 billion parameters. These models are all based on LLaMA 2. We also use a LLaMA 3-Instruct model with 8 billion parameters. All these models are instruction-tuned, which suggests that they have the capability to interpret natural language and identify appropriate options for a command-line program such as Python. In our testing, the LLaMA 3-Instruct model performed best, achieving 53% on the industry benchmark HumanEval and 49% on our internal adequacy assessment at pass@1, which is the expected probability of getting a correct solution when generating a response. This indicates that the model generates approximately every other answer correct. Due to GPU memory limitations, we had to apply quantisation to process the 13 and 34 billion parameter models. Our results revealed a mismatch between model size and optimal levels of quantisation, indicating that reduced precision adversely affects the performance of these models. Our findings suggest that a properly customised large language model can greatly reduce the coding effort of novice programmers, thereby improving productivity in material research.

With the fast development of information technology (IT) industry, the demand and market volume for off-line power supplies keeps increasing, especially those for desktop, flat-panel TV, telecommunication, computer server and datacenter. An off-line power supply normally consists of electromagnetic interference (EMI) filter, power factor correction (PFC) circuit and isolated DC/DC converter. Isolated DC/DC converter occupies more than half of the volume in an off-line power supply and takes the most control responsibilities, so isolated DC/DC converter is the key aspect to improve the overall performance and reduce the total cost for off-line power supply. On the other hand, of all the power supplies for industrial applications, those for the data center servers are the most performance driven, energy and cost conscious due to the large electricity consumption. The total power consumption of today's data centers is becoming noticeable. Moreover, with the increase in cloud computing and big data, energy use of data centers is expected to continue rapidly increasing in the near future. It is very challenging to design isolated DC/DC converters for datacenters since they are required to provide low-voltage high-current output and fast transient response. The LLC resonant converters have been widely used as the DC-DC converter in off-line power supplies and datacenters due to its high efficiency and hold-up capability. Using LLC converters can minimize switching losses and reduce electromagnetic interference. Almost all the high-end offline power supplies employs LLC converters as the DC/DC converter. But there are three major challenges in LLC converters. Firstly, the control characteristics of the LLC resonant converters are very complex due to the dynamics of the resonant tank. This dissertation proposes to implement a special LLC control method, state-trajectory control, with a low-cost microcontroller (MCU). And further efforts have been made to integrate all the state-trajectory control function into one MCU for high-frequency LLC converters, including start-up and short-circuit protection, fast transient response, light load efficiency improvement and SR driving. Secondly, the transformer in power supplies for IT industry is very bulky and it is very challenging to design. By pushing switching frequency up to MHz with gallium nitride (GaN) devices, the magnetics can be integrated into printed circuit board (PCB) windings. This dissertation proposes a novel matrix transformer structure and its design methodology. On the other hand, shielding technique can be employed to suppress the CM noise for PCB winding transformer. This dissertation proposes a novel shielding technique, which not only suppresses CM noise, but also improves the efficiency. The proposed transformer design and shielding technique is applied to an 800W 400V/12V LLC converter design. Thirdly, the LLC converters have sinusoidal current shape due to the nature of resonance, which has larger root mean square (RMS) of current, as well as larger conduction loss, compared to pulse width modulation (PWM) converter. This dissertation employs three-phase interleaved LLC converters to reduce the circulating energy by inter-connecting the three phases in certain way, and proposed a novel magnetic structure to integrated three inductors and three transformers into one magnetic core. By pushing switching frequency up to 1MHz, all the magnetics can be implemented with 4-layer PCB winding. Additional 2-layer shielding can be integrated to reduce CM noise. The proposed magnetic structure is applied to a 3kW 400V/12V LLC converter. This dissertation solves the challenges in analysis, digital control, magnetic design and EMI in high-frequency DC/DC converters in off-line power supplies. With the academic contribution in this dissertation, GaN devices can be successfully applied to high-frequency DC/DC converters with MHz switching frequency to achieve high efficiency, high power density, simplified but high-performance digital control and automatic manufacturing. The cost will be reduced and the performance will be improved significantly. / Ph. D. / With the fast development of information technology (IT) industry, the demand and market volume for off-line power supplies keeps increasing, especially those for desktop, flat-panel TV, telecommunication, computer server and datacenter. The total power consumption of today’s data centers is becoming noticeable. Moreover, with the increase in cloud computing and big data, energy use of data centers is expected to continue rapidly increasing in the near future. The efficiency of off-line power supplies is very critical for the whole human society in order to reduce the total electricity consumption. And the cost is also a key driving force for the development of novel technology in off-line power supplies due to the large market volume. An off-line power supply normally consists of electromagnetic interference (EMI) filter, power factor correction (PFC) circuit and isolated DC/DC converter. Isolated DC/DC converter occupies more than half of the volume in an off-line power supply and takes the most control responsibilities, so isolated DC/DC converter is the key aspect to improve the overall performance and reduce the total cost for off-line power supply. Among all the DC/DC converter topologies, the LLC resonant converters have been most widely used as the DC/DC converter due to its high efficiency and hold-up capability. But there are three major challenges in LLC converters. Firstly, the control characteristics are very complex due to the dynamics of the resonant tank. To achieve good control performance, very complex and expensive digital controller has to be employed. Secondly, the magnetic components are very bulky, and it is expensive to manufacture them. Thirdly, there is circulating energy in LLC converters due to the nature of resonance, which increases the total loss. To solve these challenges, this dissertation proposes to implement a special control method, state-trajectory control, with a low-cost microcontroller (MCU). All the control functions can be integrated into one simple, low-cost MCU to replace the previous complex and expensive controller. By pushing switching frequency up to MHz with next generation power devices, this dissertation proposes a novel magnetics structure that can be integrated into printed circuit board (PCB) windings to achieve low-cost and automatic manufacturing. Furthermore, this dissertation employs three-phase interleaved LLC converters topology to reduce the circulating energy, and proposed a novel magnetic structure to integrated three inductors and three transformers into one magnetic core with simple 4-layer PCB winding. All the proposed technologies have been verified on hardware prototypes, and significant improvements over industrial state-of-art designs have been demonstrated. To sum up, this dissertation solves the challenges in analysis, digital control, magnetic design and EMI in DC/DC converters for off-line power supplies. With the academic contribution in this dissertation, the cost can be reduced due to the simplified control and automatic manufactured magnetics, and the efficiency can be improved with proper utilization of next generation power devices. This dissertation will improve future DC/DC converter for IT industrial in the three most important aspects of efficiency, power density and cost.

Object counting is a process of determining the quantity of specific objects in images. Accurate object counting is key for various applications in image understanding. The common applications are traffic monitoring, crowd management, wildlife migration monitoring, cell counting in medical images, plant and insect counting in agriculture, etc. Occlusions, complex backgrounds, changes in scale, and variations in object appearance in real-world settings make object counting challenging. This dissertation explores a progression of techniques to achieve robust localization and counting under diverse image modalities. The exploration initiates with addressing the challenges of vehicular target localization in cluttered environments using infrared (IR) imagery. We propose a network, called TCRNet-2, that processes target and clutter information in two parallel channels and then combines them to optimize the target-to-clutter ratio (TCR) metric. Next, we explore class-agnostic object counting in RGB images using vision transformers. The primary motivation for this work is that most current methods excel at counting known object types but struggle with unseen categories. To solve these drawbacks, we propose a class-agnostic object counting method. We introduce a dual-branch architecture with interconnected cross-attention that generates feature pyramids for robust object representations, and a dedicated feature aggregator module that further improves performance. Finally, we propose a novel framework that leverages vision-language models (VLM) for zero-shot object counting. While our earlier class-agnostic counting method demonstrates high efficacy in generalized counting tasks, it relies on user-defined exemplars of target objects, presenting a limitation. Additionally, the previous zero-shot counting method was a reference-less approach, which limits the ability to control the selection of the target object of interest in multi-class scenarios. To address these shortcomings, we propose to utilize vision-language models for zero-shot counting where object categories of interest can be specified by text prompts.

Modern data center power architecture developing trend is analyzed, efficiency improvement method is also discussed. Literature survey of high frequency isolated power conversion system which is also called solid state transformer is given including application, topology, device and magnetic transformer. Then developing trend of this research area is clearly shown following by research target. State of art wide band gap device including silicon carbide (SiC) and gallium nitride (GaN) devices are characterized and compared, final selection is made based on comparison result. Mostly used high frequency high power DC/DC converter topology dual active bridge (DAB) is introduced and compared with novel CLLC resonant converter in terms of switching loss and conduction loss point of view. CLLC holds ZVS capability over all load range and smaller turn off current value. This is beneficial for high frequency operation and taken as our candidate. Device loss breakdown of CLLC converter is also given in the end. Medium voltage high frequency transformer is the key element in terms of insulation safety, power density and efficiency. Firstly, two mostly used transformer structures are compared. Then transformer insulation requirement is referred for 4160 V application according to IEEE standard. Solid insulation material are also compared and selected. Material thickness and insulation distance are also determined. Insulation capability is preliminary verified in FEA electric field simulation. Thirdly two transformer magnetic loss model are introduced including core loss model and litz wire winding loss model. Transformer turn number is determined based on core loss and winding loss trade-off. Different core loss density and working frequency impact is carefully analyzed. Different materials show their best performance among different frequency range. Transformer prototype is developed following designed parameter. We test the developed 15 kW 500 kHz transformer under 4160 V dry type transformer IEEE Std. C57.12.01 standard, including basic lightning test, applied voltage test, partial discharge test. 500 kHz 15 kW CLLC converter gate drive is our design challenge in terms of symmetry propagation delay, cross talk phenomenon elimination and shoot through protection. Gate drive IC is carefully selected to achieve symmetrical propagation delay and high common mode dv/dt immunity. Zero turn off resistor is achieved with minimized gate loop inductance to prevent cross talk phenomenon. Desaturation protection is also employed to provide shoot through protection. Finally 15 kW 500 kHz CLLC resonant converter is developed based on 4160V 500 kHz transformer and tested up to full power level with 98% peak efficiency. / Master of Science / Modern data center power architecture developing trend is analyzed, efficiency improvement method is also discussed. At the same time high frequency operation is preferred to reduce reactive component size like transformer and capacitor. To achieve better trade-off between high efficiency and high frequency in our research. Literature survey of high frequency isolated DC/DC power converter is given including application, circuit topology, power electronics device and magnetic transformer. Then developing trend of this research area is clearly shown following by research target. State of art advance material based power electronics devices are characterized and compared, final selection is made based on comparison result. Mostly used high frequency high power DC/DC converter topology dual active bridge (DAB) is introduced and compared with novel CLLC resonant converter in terms of converter loss. CLLC holds smaller converter loss. This is beneficial for high frequency operation and taken as our candidate. Medium voltage high frequency transformer is the key element in terms of insulation safety, power density and efficiency. Firstly, two mostly used transformer structures are compared. Then transformer insulation requirement is referred for 4160 V application according to IEEE standard. Solid insulation material are also compared and selected. Material thickness and insulation distance are also determined. Thirdly transformer loss model are introduced including core loss model and winding loss model. Transformer turn number is determined based on transformer loss and volume trade-off. Transformer prototype is developed following designed parameter. We test the developed transformer under IEEE standard requirement and pass all the test. Converter gate drive is one of our design challenge. We need to achieve symmetrical propagation delay between command signal and final drive circuit output, suppress interference from other high frequency switching devices, and protect device under short circuit condition. Gate drive IC is carefully selected to achieve symmetrical propagation delay and suppress other’s interference. Device conduction voltage is employed to compare with threshold value to determine whether it is under short circuit condition. Finally 15 kW 500 kHz CLLC resonant converter is developed based on 4160V 500 kHz transformer and tested up to full power level with 98% peak efficiency.

The introduction of synchrophasor technology to the realm of power systems has presented a myriad of novel approaches to age-old problems. In this thesis, the questions of instrument transformer calibration and transmission line parameter estimation have been examined. With synchrophasors offering real-time data for analysis, a solution to each individual problem seems feasible. A quandary however arises due to the fact that calibration methods depend on accurate knowledge of line parameters, and estimation of these parameters depend on calibrated measurements. Traditional methods of determining the parameters may not be the most accurate due to a variety of fluctuations possible on the system, which is why real-time estimation could prove beneficial. This work analyzes each problem and a feasible solution and proposes a method to achieve transducer calibration as well as parameter estimation together, while employing synchronized phasor measurements. / Master of Science / Synchrophasor Measurement Units (PMUs) provide the magnitude and angle of the quantity being measured, along with GPS time synchronization. Voltage, current and frequency data can be sent to a central control centre at the rate of 30 or 60 times per second. With a sufficient number of PMUs deployed on the electric grid, system operators now have available essentially a ‘snapshot’ of the system, which aids to monitor the grid, predict abnormal conditions as well as quickly identify troubled areas and accordingly take remedial actions. In order to facilitate the safe and reliable operation of the electric power grid, there are numerous devices that monitor quantities such as voltage, current, frequency etc. Most of these devices however are incapable of handling high levels of voltage and currents that are common to the power network. Instrument transformers (IT) are used to step down the measured quantities to much lower magnitudes that can then be analyzed by downstream devices. Each instrument transformer has a specified transformation ratio. For example, a voltage transformer with a transformation ratio of 100:1 would step down 500V to 5V. With time, these ITs may experience wear which might lead to degradation of its ratio, which would in turn be detrimental for applications relying on accurate measurements. Therefore routine calibration of ITs is desired. Traditional methods of calibration however involve taking the device out of service temporarily. As one can imagine, this is cost, labour and time intensive. With the availability of PMU data, it is now possible to perform calibration of these devices without having to take the device offline, provided we have accurate knowledge of the transmission line parameters. The parameters of a transmission line include the resistance, reactance and susceptance of the line and depend on the type of conductor used, the length and ambient temperature. Therefore seasonal and daily temperature variations can cause changes in the line parameters. With PMU data, we now have the capability to estimate these parameters, so that we have the most accurate idea of the present parameters. However for this, calibrated voltages and currents are required. Herein we face a quandary: we need to calibrate the ITs, which require accurate line parameters, but to estimate the current line parameters we need calibrated voltages and currents. This is the problem this thesis addresses. First, methods to perform both tasks, i.e. instrument transformer calibration as well as line parameter estimation using PMU measurements are analyzed. Finally an iterative method is proposed that can be applied to solve both problems together.

L'émergence des transformateurs piézoélectriques coïncident avec le développement dans les années 1950 des céramiques ferroélectriques appartenant à la famille cristalline des pérovskites qui n'ont cessé de s'améliorer depuis. Outre la compacité dont bénéficie ces structures, les transformateurs piézoélectriques offrent des performances remarquables en terme de gain en tension et rendement utiles pour des applications nécessitant une adaptation de tension ou une isolation galvanique, parfaitement dédiés aux applications de faibles puissances à haut rendement. Toutefois, les transformateurs piézoélectriques peuvent être déviés de leurs applications premières. En effet, la dernière décennie a été marquée par l'apparition de générateur de plasma par effet piézoélectrique utilisant principalement des architectures de type transformateur. Pourtant, si quelques applications usuelles illustrent parfaitement cette interaction, la compréhension des phénomènes physiques qui en sont à l'origine reste à approfondir. L'objectif de cette thèse est d'en expliquer les fondements par une approche méthodique. Ce travail s'articule autour de plusieurs étapes comprenant la mise en oeuvre d'une méthode systématique de la modélisation analytique d'un transformateur piézoélectrique, de l'étude de la carte de champ produit par un transformateur ainsi qu'une étude expérimentale vue des bornes en guise de premières investigations. La modélisation analytique est basée sur l'exploitation du Principe de Moindre Action (PMA). A partir de la théorie linéaire de la piézoélectricité, un modèle général applicable à toutes les géométries de transformateur, exploitant des modes de couplage piézoélectrique multiples, est proposé. Son caractère multimodal est par ailleurs mis en exergue. Cette modélisation est appliquée à une structure classique de transformateur piézoélectrique de type Rosen et les résultats obtenus sont validés d'une part par une identification numérique, issue d'un logiciel de calcul par éléments finis, et d'autre part par une caractérisation expérimentale. La modélisation analytique précédente ne tient pas compte dans sa mise en oeuvre de l'influence de l'environnement dans lequel évolue le transformateur piézoélectrique. Afin de caractériser le potentiel électrique produit, un modèle numérique 2D du champ électrique environnant est proposé selon la méthode des différences finies. Ce modèle est basé sur une extension du modèle analytique précédemment développé incluant les pertes mécaniques afin de quantifier le potentiel électrique de surface. Même si l'influence du plasma est négligée en première approximation, la modélisation permet de mettre en lumière les zones de fort champ correspondant aux zones de décharges luminescentes observées expérimentalement. Finalement, afin de valider le concept de générateur de plasma piézoélectrique, une caractérisation vue des bornes du transformateur piézoélectrique de type Rosen a été entreprise. Une étude systématique du déclenchement de la décharge plasma en fonction du niveau de tension et de la pression environnante a été menée. Cette part expérimentale de l'étude constitue une approche pionnière pour qualifier le comportement électromécanique du transformateur et a ainsi permis de mettre en évidence des comportements non linéaires issus de ce mode de fonctionnement atypique qu'est la génération de décharges de surface par effet piézoélectrique. / The emergence of piezoelectric transformers coincides with the development in the 1950s of ferroelectric ceramics belonging to the perovskites crystalline family. In addition to providing small size and weight, piezoelectric transformers offer outstanding performances in terms of galvanic insulation, voltage ratio and efficiency. Furthermore, compared with conventional electromagnetic transformers, piezoelectric transformers are free from electromagnetic interference. They are consequently more suitable for low power and high efficiency applications for small embedded systems. However, piezoelectric transformers can be deviated from their initial applications. Indeed, the emergence of plasma generator by piezoelectric effect, using mainly piezoelectric transformers, made its mark on the last decade. Nevertheless, if a few typical applications perfectly exemplify this interaction, the understanding of instigated physical phenomena remains to go into detail. The aim of this thesis is to explain the fundaments by a methodical approach. This work is based on several steps including the implementation of piezoelectric transformer analytical modeling, the study of the electrical field generated by a piezoelectric transformer and a first experimental investigation from piezoelectric transformer terminals. The analytical modeling is based on the utilization of the least action principle. From linear piezoelectric theory, a general model applicable to all transformer geometries, using multiple piezoelectric coupling modes, is put forward. Its multimodal characteristic is in addition underlined. This modeling is applied to a classical Rosen type transformer and the obtained results are confirmed on one hand by a numerical identication, and on the other hand by an experimental characterization. The previous analytical modeling does not take into account in its application the influence of the environment in which the piezoelectric transformer evolves. In order to qualify the produced electrical potential, a 2D numerical model of surrounding electrical field is put forward according to nite difference method. This model is based on the extension of previously developed analytical model including mechanical losses in order to quantify the surfacic electrical potential. Even if plasma influence is ignored in first approximation, the modeling allows to highlight high electrical field areas matching glow discharges areas experimentally observed. Finally, in order to validate the concept of piezoelectric plasma generator, a characterization from Rosen type piezoelectric transformer terminals has been undertaken. A systematic study of plasma discharge ignition in function of the input voltage level and the surrounding pressure has been carried out. This experimental part of the study constitutes a pioneering approach in order to qualify transformer electromechanical behavior. It has consequently allowed to give rise to nonlinear behaviors from untypical operation mode which is the surfacic discharge generation by piezoelectric effect.

Desde seu lançamento, em 2002, a Norma IEC 61850 vem evoluindo para se tornar o padrão adotado nos Sistemas de Automação de Subestações. Dentre seus vários aspectos, destacam- se os serviços de tempo real, que permitem a implementação de funções de automação e de proteção dentro da subestação através da troca de mensagens específicas entre Dispositivos Eletrônicos Inteligentes através de um barramento digital de rede de dados. O objetivo central deste trabalho é explorar algumas das questões que envolvem a implementação de uma classe de serviços de tempo real: a transmissão de valores amostrados através de Serviços SMV, definidos pela Norma IEC 61850-9. Primeiramente, apresenta-se um breve resumo das principais características da Norma IEC 61850 que possibilitam o atendimento dos três requisitos por ela estabelecidos como base: a interoperabilidade entre dispositivos de diferentes fabricantes, a versatilidade na configuração e reconfiguração do Sistema de Automação da Subestação, e a possibilidade de implementação de novas tecnologias. Em seguida, explora-se com maior profundidade todos os aspectos relevantes à implementação dos Serviços SMV. Devido à complexidade deste assunto, o autor propõe abordá-lo sob a ótica de cinco tópicos interdependentes: variações da Norma IEC 61850-9, confiabilidade do barramento de processo, sincronismo de tempo, análise da qualidade da medição e segurança cibernética. Com base nos resultados apresentados neste estudo, propõem-se duas plataformas, um protótipo de Transformador de Potencial Óptico e um protótipo de Relé de Proteção Diferencial para transformadores de potência, com o objetivo de explorar alguns dos aspectos pertinentes à implementação de um barramento de processo de acordo com a Norma IEC 61850-9. Também foram realizados testes de geração e transmissão de mensagens contendo valores de amostras de tensão/corrente do sistema elétrico (denominadas de SV Messages) com a finalidade de implementá-las de fato e avaliar as ferramentas de mercado disponíveis. Por fim foi proposto um modelo para a simulação do sistema de potência em conjunto com a rede de comunicação utilizando o programa Matlab/Simulink. O autor espera que este trabalho contribua para esclarecer os vários conceitos envolvidos na implementação do barramento de processo definido pela Norma IEC 61850-9, auxiliando na pesquisa e no desenvolvimento de novas ferramentas e dispositivos, e no aprimoramento da Norma IEC 61850. / Since its introduction, in 2002, the IEC 61850 Standard has been evolved to become the standard adopted in Substation Automation Systems. Among its various aspects it can be highlighted the real-time services, which allows the implementation of automation and protection functions within the substation by exchanging specific messages between Intelligent Electronic Devices via a digital data bus network. The main objective of this text is to explore some issues about the implementation of one class of real-time services: the transmission of sampled values through SMV Services, defined by IEC 61850-9 Standard. First, it is presented a brief summary of the main features of IEC 61850 Standard which allows it to meet the three main requirements on this standard: interoperability between devices from different manufacturers, versatility in configuration and reconfiguration of Substation Automation System, and possibility of implementation of new technologies. Next, it is explored, with more details, all aspects involved in the implementation of SMV Services. Due to the complexity of this subject, the author proposes approaches it from the perspective of five interdependent topics: variations of IEC 61850-9 Standard, process bus reliability, time synchronism, analysis of measurement quality, and cyber security. Based on results presented two platforms are proposed, an Optical Voltage Transformer prototype and a Differential Protection Relay prototype for power transformers, in order to explore some of aspects related to the implementation of a process bus according to IEC 61850-9 Standard. It was executed also some tests of generation and transmission of messages containing samples values of voltage/current from power system (referred as SV Messages) in order to implement them in fact, and to evaluate available market tools. At the end it is propoused a model for simulation of power system and network communication using Matlab/Simulink software. The author hopes this work may contribute to clarify the various concepts involved in the implementation of a process bus according to IEC 61850-9 Standard, aiding in research and development of new tools and devices, and improvement of IEC 61850 Standard.

Nesta pesquisa, apresentam-se os principais eventos relacionados com a proteção de transformadores e sua correlação com os distúrbios de qualidade da energia elétrica (QEE). Neste sentido, foi desenvolvido um algoritmo que utiliza a transformada Wavelet (TW) e a lógica Fuzzy (LF) para classificar os eventos transitórios associados à proteção de transformadores. Estes eventos foram observados em um sistema elétrico de potência (SEP) simulado com a utilização do software Alternative Transients Program (ATP). Importa ressaltar que o sistema modelado apresenta transformadores ligados em paralelo, possibilitando o estudo de eventos decorrentes desta situação, como a energização solidária (Sympathetic Inrush). Por este SEP, modelado sobre parâmetros reais, foram simuladas várias situações transitórias, que provocam o aparecimento de correntes diferenciais, sendo estas direcionadas para análise do algoritmo desenvolvido. Afirma-se que, nos testes realizados, o algoritmo proposto apresentou um desempenho satisfatório perante as mais variadas situações a que foi submetido, identificando as causas das correntes diferenciais, sejam proporcionadas por defeitos ou por outras condições de operação aplicadas. / In this research, the main events related to the transformer protection and its correlation with the power quality disturbances (PQ) are presented. In this context, an algorithm based on Wavelet transform (WT) and Fuzzy logic (FL) was developed to classify the transient events associated with the transformer protection. These events were observed in an electrical power system (EPS) simulated using the Alternative Transients Program (ATP) software. It should be emphasized that the modeled system presents transformers connected in parallel, allowing the study of events of this situation, such as sympathetic inrush. For the simulated EPS, modeled based on real parameters, various transients situationswere simulated, causing the appearance of differentials currents which were directed to the analysis. The proposed algorithm showed a satisfactory performance tomany situations, identifying the causes of the differentials currents, either provided by faults or other operation conditions.