Short term electrical load forecasting using back propagation neural networks

This paper presents a new approach for short term electrical load forecasting (STLF) using artificial neural networks (ANN), and examines the feasibility of various mathematical models for STLF. To make these mathematical models to yield satisfactory and acceptable results, various system models are formulated considering various combination of parameters like base load component, day of the week, load inertia, short term trends, autocorrelation, length of the past data, etc. Various modifications of Back Propagation Algorithm (BPA) have been proposed, to explore the ideal combination that suit the forecasting need of large utilities like regional electricity grids. Further, the load dynamics are extensively studied to identify the parameters for system modeling.