Application of a Hybrid Model to Short-Term Load Forecasting

Short-term load forecasting has been viewed as an important problem for its wide application. Grey forecasting model is tested by using electric load data sampled from SA for short-term load forecasting in this paper. Then by regarding the electric load residual series obtained from grey forecasting model as the original data, the grey forecasting model and the support vector machine (SVM) are applied to forecast the follow-up residual series respectively, by adding this forecasted residual series to the original forecasted electric load by single grey forecasting model, the mean absolute percentage error is reduced from 11.97% to 11.71% when using grey forecasting model and a significant reduce to 5.45% while using SVM in residual forecasting.