Sistemas Híbridos Aplicados à Filtragem de Sinais de Alarme de Proteção de uma Subestação Telecomandada

Abstract

In this work an Artificial Inteligent Hybrid System is applied to the filtering process of an alarm signal set, generated at a substation. This combined system consists in an Artificial Neural Network, using MLP backpropagation algorithm and a Fuzzy Logic Inference System of the Mamdani type. The results show that the Hybrid System is able to interpret correctly the alarm information, resulting in a better performance than the Neural System alone.