Electricity Load Forecast using Neural Network Trained from Wavelet-Transformed Data

With accurate electricity load forecasting important information is provided that helps to build up cost effective risk management plans for any electric utility such as electricity generators and retailers in the electricity market. In this article, we propose a wavelet based multilayer perceptron (MLPw) approach for the prediction of one-hour and one-day ahead load trained from Haar a trous wavelet-transformed historical electricity load data. We assess results produced by the MLPw method, with multiple resolution autoregressive (MAR), single resolution autoregressive (AR), multilayer perceptron (MLP), and the general regression neural network (GRNN) model. Experimental results are based on the New South Wales (Australia) electricity load data that is provided by the National Electricity Market Management Company (NEMMCO)