Online prediction of time series using incremental wavelet decomposition and support vector machine

Abstract

Time series prediction is widely used in industry engineering, finance, economy, traffic and many other fields. For power system, prediction is often concerned, and online prediction has significance to the system operation safely and steadily. An efficient method for online prediction of time series using wavelet decompositions and support vector machine is presented, which can improve the prediction accuracy. For online application, sliding window model and incremental algorithms for wavelet decompositions are used. This method has low cost in memory and run time, it can predict time series in high accuracy and less time. Simulation experiment using gas furnace time series dataset show the effectiveness of proposed method.