Estimação e aplicação de sincrofasores em redes de distribuição de energia

• Ano de defesa: 2021
• Autor(a) principal: Grando, Flavio Lori
• Tipo de documento: Tese
• Tipo de acesso: Acesso aberto
• Idioma: por
• Instituição de defesa: Universidade Tecnológica Federal do Paraná; Curitiba; Brasil; Programa de Pós-Graduação em Engenharia Elétrica e Informática Industrial; UTFPR
• Palavras-chave em Português: Aprendizado de máquinas; Energia elétrica - Distribuição; Redes elétricas - Análise; Harmônicos (Ondas elétricas); Métodos de Simulação; Erros - Medição; Falhas de energia elétrica; Machine learning; Electric power distribution; Electric network analysis; Harmonics (Electric waves); Simulation methods; Errors - Measuremen; Electric power failures; CNPQ::CIENCIAS EXATAS E DA TERRA::CIENCIA DA COMPUTACAO; Engenharia Elétrica
• Link de acesso: http://repositorio.utfpr.edu.br/jspui/handle/1/27892

This thesis is contextualized on the monitoring of electrical power systems, more specifically on the employment of PMUs (Phasor Measurement Units) in distribution systems. PMUs have been successfully applied in the transmission network, characterizing large-area dynamic phenomena. However, their use in distribution networks is not yet a reality. The main questions involve: (i) what are the data benefits and (ii) what are the measurement requirements. To contribute to the advancement of PMU technology for distribution systems, this thesis sought to answer such questions whose contributions are organized in two parts: application and measurement of synchrophasors. In the application scope, a study of the impact of the quality of phasor measurements on the power grid event classification problem was addressed. The potential of the data to recognize system events was evaluated, while the precision and accuracy requirements that meters and, consequently, algorithms need to have in order to enable distribution applications were investigated. For this purpose, an extensive event classification study was conducted under different uncertainty conditions in the data, with various distributions and error levels. 2,160 classification results are presented, involving six classifiers (based on machine learning), three distinct groups of events, with 4, 8, and 62 classes per group, and various phasor error characteristics. The results support the instrumentation area, bringing analysis of multiple sources of errors and demonstration of how precision and accuracy affect distribution system event recognition. These results and conclusions support the second part of the study with the estimation of phasor data, taking into account the particularities of the distribution network. In the measurement scope, three new estimation algorithms were proposed, with distinct operating principles: (i) a hardware based approach, (ii) sparse approach and (iii) parametric approach. The contributions involve, but are not limited to: harmonic, interharmonic and damping factor estimation. They also involve dynamic estimations, using adaptation strategies to power system oscillations. Additionally, the methods are designed to provide noise tolerance of distribution systems and computationally efficient for use in general purpose embedded systems. Tests were conducted in a controlled simulation environment and experimental tests using an embedded system. For that, a parallel development to this work was used, which consisted of a test platform capable of applying standardized tests in an automated manner. Finally, one of the methods was employed in a PMU prototype, which was used to monitor the electrical grid, registering harmonics and system events.