Implementación de una Red Neuronal Artificial para la predicción de la Demanda Eléctrica a corto plazo

Abstract

The prognostication of electricity demand is a major problem for the electricity business, the regulators agencies can make appropriate decisions based on the results obtained in a demand prognostic. A good prediction produces great technical and financial benefits, It for this reason that nowadays is necessary to develop predictive models with minimal error rate.

The results presented have been obtained by using an artificial neural network for forecasting electricity demand in the short term developed using Matlab, the database used corresponds to those recorded by the SCADA system measurements on a feeder primary of of LojaEcuador city.

An artificial neural network was implemented with the minimum of layers and neurons, without losing forecast accuracy, to train the neural network an adequate selection and classification of input variables was performed.