Combining KPCA with Support Vector Regression Machine for Short-Term Electricity Load Forecasting

Abstract

Short-term electricity load forecasting is important both from the technological and the economical point of view, but it is also a difficult work because the accuracy of forecasting is influenced by many unpredicted factors whose relationships are commonly complex, implicit and nonlinear. By studying the methods proposed by other scholars, a mew method, KPCA (kernel principal component analysis) -SVRM (support vector regression machine) is proposed by this paper. The first step of this method is to apply KPCA to SVRM for feature extraction. KPCA first maps the original inputs into a high dimensional feature space using the kernel method and then calculates PCA in the high dimensional feature space. These new features are then used as the inputs of SVRM to solve the load forecasting problem. By learning and training, we use the data of this subset to get the solution and find interrelationship of input and output by the SVRM. Practical examples are cited in this paper to illustrate the process. The KPCA-SVRM method can also be used to solve other forecasting problems.