Redes neurais artificiais aplicadas ao problema de variação de tensão de longa duração em sistemas de distribuição de energia elétrica

Abstract

This study presents a tool that uses Artificial Neural Networks (ANNs) to identify different patterns of problems causing long-term voltage variations on energy supply, having regard to the financial impact associated with the lack of quality in electricity, from the point of view of both the consumer and the supplier. The proposal is based in train and test a Multilayer Perceptron (MLP), with different patterns of problems within real data provided by the Companhia Paulista de Força e Luz (CPFL). Based on the generalization problems from the ANNs, it will be capable of recognize different sources of problems related to voltage level. This proposal will enable the suppliers with a fast re-establishment of electricity supply, decreasing the costs of financial compensation and dissatisfaction among clients.