Short-term load forecasting using Support Vector Regression-based Local Predictor

This paper introduced a novel forecasting method, Support Vector Regression with Local Predictor (SVRLP), which aims to forecast the short-term load distribution function. To increase the forecast accuracy, the conventional Support Vector Regression (SVR) is combined with a phase space reconstruction technique, called local predictor. This proposed forecast method can be applied to forecast the load demand using smaller training samples and overcome the local optimal solution problem. Therefore, the SVRLP is able to provide more reliable forecast results to the system operators. In the experimental studies, the SVRLP is evaluated on the load data of collect from Guangzhou on the China Southern Grid (CSG), and is compared with the predictors based on conventional SVR, the Auto-Regressive Moving Average (ARMA) and the Artificial Neural Network (ANN), respectively. The results demonstrate that the proposed method can achieve a better performance than the other methods.