Previsão de demanda: um estudo de caso para o Sistema Interligado Nacional

Abstract

At about the middle of the 90s decade, Brazilian government starteda reformulation process on the electricity energy market. The new Brazilian regulation model has defined the rules to promote the markethorizontalization through the disaggregation of the segments involved onthe service provision, the break of state monopole and competition. Facing the new reality of competition and market growth trend, the companies must be able to manage the inherited risks of the process, in order to map the market behavior and foresee the need for intervention at critical parts of the process, and to keep or improve the quality and cost of service. The purpose of this work is to present the application of innovative technologies to address energy demand forecast through the identification of market behavior and correlation with exogenous and endogenous variables. The models proposed are based on statistical methods, artificial neural networks, algorithms for decomposing time series into trend, cyclical and seasonal components and algorithms for independent component analysis. The methods proposed were implemented and applied to forecast the future demand of SIN - Sistema Interligado Nacional