Detecção, segmentação e classificação de afundamentos de tensão em sistemas elétricos de potência

• Autor(a) principal: Nagata, Erick Akio
• Data de Publicação: 2018
• Tipo de documento: Dissertação
• Idioma: por
• Título da fonte: Repositório Institucional da UFLA
• Texto Completo: http://repositorio.ufla.br/jspui/handle/1/28529
• topic: Energia elétrica - Qualidade; Afundamentos de tensão - Detecção e segmentação; Afundamentos de tensão - Classificação; Sistema elétrico - Geração distribuída; Power quality; Voltage sag classification; Voltage sag segmentation; Distributed generation; Engenharia Elétrica

Technology advancement and the increased use of electronic equipment and the decentralized energy generation made the electrical power quality to become a growing concern factor for both the energy consumers and power distribution companies. Thus, researches in this area aiming to develop new methods for disturbance detection, classification and mitigation have become more frequent and important in the scientific environment. According to this scenery, this work proposes an innovative approach to detect, segment and classify voltage sags according to their causes. To detect and segment, it is used the statistical signal processing technique known as Independent Component Analysis, with the advantage of being fast and with low computational cost in the operational stage, once it uses only 1/8 cycle of the fundamental component and works in real time. For classification purposes, Higher-Order Statistics are used to extract features and the classifiers are based on Support Vector Machine and Neural Networks. It was tested signal windows of 1, 1/2, 1/4 and 1/8 cycle, and the best results were achieved for signal windows with 1/2 cycle and using a Neural Network as classifier. For both detection/segmentation project and feature selection, it was used the metaheuristics Teaching-Learning-Based Optimization. Suitable results were achieved for simulated signals. In addition, real signals were used to evaluate the detection and segmentation method and good results were achieved for the detection task, presenting 0.86% of error for the data set analyzed.