Desenvolvimento de um Classificador Inteligente para Variações de Tensão de Curta Duração

Gabriel Humberto Andrade Freitas; Danilo Hernane Spatti

doi:10.20906/sbai.v1i1.2564

Gabriel Humberto Andrade Freitas Departamento de Engenharia Elétrica e de Computação, Escola de Engenheria de São Carlos - Universidade de São Paulo, SP
Danilo Hernane Spatti Departamento de Sistemas de Computação, Instituto de Ciências Matemáticas e de Computação - Universidade de São Paulo, SP

DOI: https://doi.org/10.20906/sbai.v1i1.2564

Keywords: Power Quality (PQ), Electrical Disturbances, Classification, Artificial Neural Networks (ANN), Electrical Signal Processing

Abstract

Nowadays, keep Power Quality (PQ) has become a necessary factor in many social and economic environments, becoming a concern not only of engineers and regulatory agencies, but also of system agents, consumers and companies. Besides, the power quality disturbances presence it´s more often due to the addition of new kinds of loads in the electric system, those with non-linear features, from converters, new kinds of generation (photovoltaic and eolic), etc. In this bias, the developed work presents an identification and classification method for the PQ issues using an Artificial Neural Network (ANN) Multilayer Perceptron (MLP); whose inputs are a reduced set of electrical features extracted from these signal´s samples. The system, trained and validated, presented an excellent classification performance for four signal patterns (reference signal plus three disturbances) using five attributes as inputs.