Um Algoritmo para Detecção e Classificação de Faltas em Linhas de Transmissão Baseado em Transformada Wavelet e RNA

Abstract

A method for fault diagnosis in transmission lines is presented, which to accomplish the detection step and fault duration time by a set of rules established from the current waveform analysis at the time and wavelet domains. A fault is distinguished from others power quality issues which are common in power systems operation, such as sags and transients. The fault is classified by an artificial neural network, responsible for the pattern recognition of voltages and currents related to each specific type. The method is evaluated for real and simulated faults in transmission lines of Chesf's electrical system and good results in both cases are achieved.