Integrating KPCA and locally weighted support vector regression for short-term load forecasting

This paper proposes a new approach to solve the short term load forecasting problem that considers electricity price as one of the main characteristics of the system load. The proposed method is derived by integrating the kernel principal component analysis (KPCA) method with locally weighted support vector regression (LWSVR). LWSVR can be derived by modifying the risk function of the support vector regression algorithm with use of locally weighted regression while keeping the regularization term in its original form. In the proposed model, the first stage is using KPCA to extract features and obtain kernel principal components which used to construct the phase space of the multivariate time series of inputs. LWSVR is employed in the second stage to solve the load forecasting problem. In addition, to optimize the weighting function´s bandwidth, the weighted distance algorithm is presented. The performance of the proposed model is evaluated with the historical load, temperature and price data from the Victorian electricity market in Australia. The results show that the proposed method provides a relatively better forecasting performance in comparison with other published models employing the same data.