THE RESEARCH OF DAILY LOAD FORECASTING MODEL BASED ON WAVELET DECOMPOSING AND CLIMATIC INFLUENCE
Xie Hong · Proceedings of the CSEE · 2001
In this paper, the wavelet transform is applied to decompose daily load data into wavelet components. Each component is considered as the sum of two parts, one of which is influenced by climatic factor and a polynomial regressive model is built for this part. The other part, however, is not influenced by climatic factors. When its variance is bigger, a recurrent neural networks forecasting model is built. Otherwise a ARMA( p,q ) model is done. In this way, the precision of forecast can be improved and the efficiency of building model is increased.