Diploma work deals with a converter design for an artifical power net. There is described analysis of possible solving of this converter. Next there is contained dimension of components force circumference converter, design of filtres and converter regulator. Project deals with simulation of dynamical qualities for an output changer device in settings Matlab Simulink. In this work there is even a design power board alternator, wakers and directing of alternator.

The project developed in within the framework of Magister ´s studies of field M-STM is elaborated to meet customer requirements for manufacture of transformer station, which is intended for use in photovoltaic power station. Design and drawing documentation were prepared using 3D CAD System Solid Edge. Self-supporting frame of the transformer station welded from separate parts made from alu/zinc plates is in the form of a technological unit, which is equiped with electrical technology needed for safe and failure-free operation of the power station. It deals with technology of components production, including subsequent process of hard soldering and surface finishing. As the best solution was chosen processing of forming punching and using brake press. Chosen production methods are assessed not only form point of view of used technology, including the technical pitfalls associated with processing of alu/zinc plates, but also in terms of economic costs of manufacture throughout the transformer station. The forming processes are made using punching press TRUMPF TC200R, with rated shear force of 165 kN and brake press URSVIKEN Optiflex with rated compressive force 1 300 kN. Forming tools are optimally selected according to parameters of different types of parts. Finally, it compares the technical-economic aspects of production transformer station made of alu/zinc plates compared with whole-concrete construction of the station using bell-casting method.

In this thesis is described the function and design of the transformer designed on ferromagnetic core, composed of transformer sheets. It is explained a transient inrush current of the transformer. In this work is inserted voltage and current waveforms, designed printed circuit board, serving to reduce the inrush current.

A dissertation submitted to the Faculty of Engineering, University of the Witwatersrand, Johannesburg, in fulfilment of the requirements for the degree of Master of Science in Engineering / The sudden failure of oil paper-insulated current transformers has become a problem of considerable concern due to the cost of the resulting destruction, and danger to personnel. The aim of the dissertation is to determine the most suitable method of detecting incipient faults in current transformers, test this method on an experimental current transformer, and finally implement this technique in a low cost on-line monitor. A literature survey indicated that hydrogen gas-in-oil on-line monitoring would be the most suitable technique.(Abbreviation abstract) / Andrew Chakane 2019

On-load tap-changers (OLTC) are devices in the power grid that keeps the voltage level constant for consumers, regardless of the power demand. Hitachi ABB Power Grids, producer of the OLTC family named VUC, guarantees 30 years of lifetime. Such a pledge requires high standard devices. This thesis has analyzed data from routine tests of switching times in the diverter switch of OLTC’s, performed before devices were put in service. The correlation of part switching times for all units leaving the factory during the past year was evaluated by calculating Pearson’s correlation coefficient. A linear trend was fitted to the data, realizing that the prediction errors, as well as the part switching times, were Gaussian distributed. The time while the resistor vacuum interrupter was open could be predicted within the interval of approximately 2ms with 2 standard deviations accuracy. To classify time series from the routine test as expected or unexpected, a model-based algorithm was implemented. The average switching time for all consecutive switches was used to define expected series. A moving average was implemented to neglect outliers and remove oscillating patterns. The majority of all data was classified as expected time series. The ones who did not, still preserved a good correlation between the part switching times. Examining the relationship between part switching times could be a valuable perspective in further classification of expected time series. The possibility of incorporating measurement of part switching times on OLTC’s in normal operation, to use the knowledge gained by this thesis, was investigated. Position sensors were mounted to measure the position of the lifting yokes, opening and closing the vacuum interrupters. The time while the vacuum interrupter contacts were open could be estimated with better accuracy than the position sensor provided. Unfortunately, those sensors cannot be utilized in normal operation. If other possibilities could be found, perhaps a laser position sensor, the implemented algorithm would be valuable.

京都大学 / 0048 / 新制・課程博士 / 博士(工学) / 甲第17885号 / 工博第3794号 / 新制||工||1580(附属図書館) / 30705 / 京都大学大学院工学研究科航空宇宙工学専攻 / (主査)教授 藤本 健治, 教授 泉田 啓, 教授 椹木 哲夫 / 学位規則第4条第1項該当 / Doctor of Philosophy (Engineering) / Kyoto University / DFAM

Transformer is one of the most reliable components in an electric power system, however its failure has huge opportunity costs for an electric utility. In this work, we detect transformer electrical faults promptly and accurately classify the fault types using voltage/current data from Phasor Measurement Units. Our work can also eliminate uncertainties which are inherent in traditional transformer fault diagnostic techniques like dissolved gas analysis. In this thesis, first, possible causes of transformer failures are discussed, and four common transformer electrical faults are identified. Second, a comprehensive simulation model for electrical faults is developed. Third, fast and efficient abrupt change detection algorithms are applied for fault event detection. Finally, selected supervised machine learning classifiers are trained to classify type of transformer electrical faults. Our proposed work can be used with alarms and relays to notify system operators and remove the faults, as well as for post-mortem analysis of transformer failures.

<p dir="ltr">Reinforcement learning algorithms have traditionally been implemented with the goal</p><p dir="ltr">of maximizing a reward signal. By contrast, Decision Transformer (DT) uses a transformer</p><p dir="ltr">model to predict the next action in a sequence. The transformer model is trained on datasets</p><p dir="ltr">consisting of state, action, return trajectories. The original DT paper examined a small</p><p dir="ltr">number of environments, five from the Atari domain, and three from continuous control,</p><p dir="ltr">and one that examined credit assignment. While this gives an idea of what the decision</p><p dir="ltr">transformer can do, the variety of environments in the Atari domain are limited. In this</p><p dir="ltr">work, we propose an extension of the environments that decision transformer can be trained</p><p dir="ltr">on by adding support for the VizDoom environment. We also developed a faster method for</p><p dir="ltr">offline RL dataset generation, using Sample Factory, a library focused on high throughput,</p><p dir="ltr">to generate a dataset comparable in quality to existing methods using significantly less time.</p><p dir="ltr"><br></p>

As Next Generation (NextG) networks become more complex, the need to develop a robust, reliable network traffic prediction framework for intelligent network management increases. This study compares the performance of machine learning models in network traffic prediction using a custom Service-Level Agreement (SLA) - based loss function to ensure SLA violation constraints while minimizing overprovisioning. The proposed SLA-based parametric custom loss functions are used to maintain the SLA violation rate percentages the network operators require. Our approach is multivariate, spatiotemporal, and SLA-driven, incorporating 20 Radio Access Network (RAN) features, custom peak traffic time features, and custom mobility-based clustering to leverage spatiotemporal relationships. In this study, five machine learning models are considered: one recurrent neural network (LSTM) model, two encoder-decoder architectures (Transformer and Autoformer), and two gradient-boosted tree models (XGBoost and LightGBM). The prediction performance of the models is evaluated based on different metrics such as SLA violation rate constraints, overprovisioning, and the custom SLA-based loss function parameter. According to our evaluations, Transformer models with custom peak time features achieve the minimum overprovisioning volume at 3% SLA violation constraint. Gradient-boosted tree models have lower overprovisioning volumes at higher SLA violation rates. / Master of Science / As the Next Generation (NextG) networks become more complex, the need to develop a robust, reliable network traffic prediction framework for intelligent network management increases. This study compares the performance of machine learning models in network traffic prediction using a custom loss function to ensure SLA violation constraints. The proposed SLA-based custom loss functions are used to maintain the SLA violation rate percentages required by the network operators while minimizing overprovisioning. Our approach is multivariate, spatiotemporal, and SLA-driven, incorporating 20 Radio Access Network (RAN) features, custom peak traffic time features, and mobility-based clustering to leverage spatiotemporal relationships. We use five machine learning and deep learning models for our comparative study: one recurrent neural network (RNN) model, two encoder-decoder architectures, and two gradient-boosted tree models. The prediction performance of the models was evaluated based on different metrics such as SLA violation rate constraints, overprovisioning, and the custom SLA-based loss function parameter.

Software plays an integral role in powering numerous everyday computing gadgets. As our reliance on software continues to grow, so does the prevalence of software vulnerabilities, with significant implications for organizations and users. As such, documenting vulnerabilities and tracking their development becomes crucial. Vulnerability databases addressed this issue by storing a record with various attributes for each discovered vulnerability. However, their contents suffer several drawbacks, which we address in our work. In this dissertation, we investigate the weaknesses associated with vulnerability descriptions in public repositories and alleviate such weaknesses through Natural Language Processing (NLP) approaches. The first contribution examines vulnerability descriptions in those databases and approaches to improve them. We propose a new automated method leveraging external sources to enrich the scope and context of a vulnerability description. Moreover, we exploit fine-tuned pretrained language models for normalizing the resulting description. The second contribution investigates the need for uniform and normalized structure in vulnerability descriptions. We address this need by breaking the description of a vulnerability into multiple constituents and developing a multi-task model to create a new uniform and normalized summary that maintains the necessary attributes of the vulnerability using the extracted features while ensuring a consistent vulnerability description. Our method proved effective in generating new summaries with the same structure across a collection of various vulnerability descriptions and types. Our final contribution investigates the feasibility of assigning the Common Weakness Enumeration (CWE) attribute to a vulnerability based on its description. CWE offers a comprehensive framework that categorizes similar exposures into classes, representing the types of exploitation associated with such vulnerabilities. Our approach utilizing pre-trained language models is shown to outperform Large Language Model (LLM) for this task. Overall, this dissertation provides various technical approaches exploiting advances in NLP to improve publicly available vulnerability databases.