Sistema para detecção de causas em variação de tensão de longa duração utilizando redes neurais

Abstract

The sensitivity of electronic equipment to variations in voltage levels makes them susceptible to permanent damage or a reduction in their useful life. The growing demand for this equipment has also raised concerns about the quality of electricity supplied by distribution systems, with a view to diagnostic and analysis efforts. This document proposes a long-term voltage variation problems detecting system, which is based on the methodology of artificial neural networks, and its main objective is to assist in the detection and speed up the solution of problems related to Power Quality. The implementation of the neural network was made using the Keras library, and developed in the Python programming language. The neural network has as input three-phase voltage values, and provides as an answer the classification or not of each sample in the problems proposed in this work. The achieved results show that the proposed system is efficient and can help in the detection and speed up the maintenance process to resume normal voltage levels.