Power system short-term load forecasting based on KPCA-LSSVM

• 1、School of Electrical Engineering and Automation，Harbin Institute of Technology

Abstract：Although SVM have the remarkable advantages of non-linear regression, high and small time complexity, its training efficiency will reduce when it dealing with massive data of load. According to this problem, load forecasting based on least squares support vector machines is introduced to short-term load forecasting. Load forecasting based on least squares support vector machines will reduce when the Dimension of sample data is too high. principle component analysis(PCA) is one of linear dimensionality reduction measures. The load forecasting based on least squares support vector machines and PCA is established in this paper. The results of practical example shows that the workload of the load forecasting model is reduced and the prediction accuracy is improved by this method.PCA can obtain the linear relationship of the original characteristics. kernel principle component analysis(KPCA) can reflect the nonlinear relationship of the original characteristics. So Load forecasting based on least squares support vector machines and KPCA is put forward in this paper. The results of practical example shows that the prediction accuracy of this method is higher than the load forecasting based on least squares support vector machines and PCA.

Keywords： power systems short-term load forecasting least squares support vector machine principle component analysis kernel principle component analysis