Nonlinear Models for Short-time Load Forecasting

AbstractShort Term Load Forecasting (STLF) is essential for planning the day-to-day operation of an electric power system. As this forecasting leads to increased security operation's conditions and economic cost savings, numerous techniques have been used to improve the STLF. We propose in this paper the comparison of two nonlinear regression techniques namely Gaussian Process (GP) regression models and Neural Network (NN) models. While the Bayesian approach to NN modelling offers significant advantages over the classical NN learning methods, it will be shown that the use of GP regression models will improve the performances of the forecasting. The proposed techniques are applied to real load data