Short-term load forecasting using wavelet transform and support vector machines

This paper presents a new technique in short-term load forecasting (STLF.) The proposed method consists of the discrete wavelet transform (DWT) and support vector machines (SVMs.) The DWT splits up load time series into low and high frequency components to be the features for the SVMs. The SVMs then forecast each component separately. At the end we sum up all forecasted components to produce a final forecasted load. The data from Bangkok-Noi area in Bangkok, Thailand, is used to verify on the one-day ahead load forecasting. The performance of the algorithm is compared with that of the SVM without DWT, and neural networks with and without DWT. The experimental results show that the proposed algorithm yields more accuracy in the STLF than the others.