A Novel Hybrid Approach of KPCA and SVM for Building Cooling Load Prediction

Abstract

In this paper, a novel building cooling load forecasting approach combining kernel principal component analysis (KPCA) and support vector machine (SVM) is proposed. KPCA is an improved PCA, which possesses the property of extracting optimal features by adopting a nonlinear kernel function method. The original inputs are firstly transformed into nonlinear principal components using KPCA. These new features are then used as the inputs of SVR to solve the load forecasting problem. The theoretical analysis and the simulation results show that KPCA can efficiently extract the nonlinear feature of initial data. KPCA-SVR has powerful learning ability, good generalization ability and low dependency on sample data compared with PCA-SVR and SVR. It also indicates that the integration of KPCA and SVR forecast cooling load effectively and can be used in building cooling load prediction.