Journal of Xi'an Jiaotong University

Multi-Scale Short Term Load Prediction Model Using Least Square Support Vector Machine
Liu Zunxiong¹,Zhong Hualan²,Zhang Deyun¹
(1.School of Electronics and Information Engineering,Xi'an Jiaotong University,Xi'an 710049,China;2. School of Electronics and Electric Engineering, East China Jiaotong University,Nanchang 330013,China)

Abstract:A revised wavelet network model for short-term electric load prediction was proposed by using LS-SVM (least square support vector machine) to achieve the combined prediction of wavelet decomposition coefficients,in which the wavelet decomposition of short-term load time series is performed by à trous algorithm firstly,the approximate coefficients at the specified scale and the wavelet coefficients at the related scales can be obtained,then LS-SVM is utilized to perform the multi-scale combined prediction for the coefficients of predicted points,the corresponding predicted value is yielded by the wavelet reconstruction.The simulation experiments are carried out with short-term load demand data from a local region,and the correlation between the predicted points and history data is explored.The results show that the predicted accuracy with the proposed model is better than that of traditional neural-network-wavelet approaches for short term load prediction.
Keywords:short term electric load;multi-scale prediction; à trous;least square support vector machine