NeuroDem-a neural network based short term demand forecaster

Abstract

The application of Neural Network (NN) based short-term load forecasting (STLF) has developed to sophisticated practical systems over the years. However, the question of how to maximize the generalization ability of such machines, together with the choice of architecture, activation functions, training set data and size, etc. makes up a huge number of possible combinations for the final NN design, whose optimal solution has not been figured yet. This paper describes a STLF system (NeuroDem) which has been used by Brazilian electric utilities for 3 years. It uses a nonparametric model based on a linear model coupled with a polynomial network, identified by pruning/growing mechanisms. NeuroDem has special features of data pre-processing and confidence intervals calculations, which are also described. Results of load forecasts are presented for one year with forecasting horizons from 15 min to 168 hours ahead.