An efficient method for peak load forecasting

Abstract

This paper describes the peak load characteristics of power systems. A new method for the peak load forecasting (PLF) is proposed by means of an integration technique, which combines fuzzy clustering with feed-forward neural network (FNN) of the Levenberg-Maruardt (LM) training algorithm. Compared with the traditional back propagation (BP) algorithm, the proposed method proves more efficient to the predicted accuracy during the peak load period by using actual data of a power grid in China