Neural Nework-Based Time Series Methods for Load Forecasting

Abstract

Load forecast plays an important role in power system operation and control. Significant contribution in power system economics may also be observed. Many decisions of the power system depend on the future load demand. The accuracy of STLF is necessary for the optimal and economical operation of the power systems. This paper presents a new approach to STLF. In this paper, time series methods are presented on the basis of neural networks. The time series methods are included autoregressive, nonlinear autoregressive, and non-linear autoregressive with external inputs (narx). The comparative results are presented with the ANN. In this paper, the narx method gives more efficient and accurate results than other methods.