Previsor neural de carga elétrica baseado em análise de multiresolução via wavelets e técnicas de reconstrução do espaço-fase

Abstract

The importance of short-term load forecasting has been increasing lately. With deregulation and competition, energy price forecasting has become a big business. Bus-load forecasting is essential to feed analytical methods utilized for determining energy prices. The variability and non-stationarity of loads are becoming worse due to the dynamics of energy prices. Besides, the number of nodal loads to be predicted does not allow frequent interactions with load forecasting experts. More autonomous load predictors are needed in the new competitive scenario. This thesis deals with two main research lines. In the first one, two different strategies for successfully embedding the Discrete Wavelet Transform into Artificial Neural Networksbased short-term load forecasting is presented. The first strategy is new. It consists of creating a model for load forecasting whose inputs are based on information from the original load sequence and from wavelet domain subseries, as well. The second alternative predicts the load’s future behavior by independently forecasting each subseries in the wavelet domain. The other research line evaluates the feasibility of a nonlinear criterion based on the method of delay coordinates for determining the best set of input variables for a neural forecaster. This criterion is fully compared to another linear criterion based on the autocorrelation function. The main goal of this work is to develop more robust load forecasting algorithms. Hourly load and temperature data for a North-American electric utility are used to test the proposed methodologies.