Modelo predictivo de fallas en alimentadores primarios de concesión de la Empresa Eléctrica Regional Centro Sur usando aprendizaje profundo de máquina

Abstract

This work proposes the forecasting of failures in primary feeders of Empresa Eléctrica Regional Centro Sur C.A. (CENTROSUR) using Deep Learning. The model, based on neural networks, has not been wide explored in the electricity distribution industry, both locally and internationally. For the development of the model, the company’s historical data from the last five years about failures was collected and analyzed. The validation of the forecasting model was performed using data from the #521 primary feeder of the Substation (S/E) 5 of CENTROSUR, where the highest incidence of faults is evidenced.