Analysis and Prediction of Requirements About Offshore Patrol Oil Support Based on WT-SVM
XIE Chong;GUO Jikun;LIU Qiang;LI Zhongguo;Army Logistical University of PLA;
According to the time characteristics of the offshore patrol oil requirements,a combined prediction model based on wavelet analysis and support vector machine is built in this paper. This prediction model makes full use of the ability of multi scale extraction of information from wavelet analysis.Firstly the time series are extracted by wavelet de-noise and its multi-resolution,making the sample data that meets the normal operation of each layer falls in a certain fluctuation interval,which can improve the prediction ability of each level data. Then support vector machine is used for learning each level after threshold de-noising. Through the fusion of the training results at all levels,the predicted results are finally obtained. The experimental results show that the model can present the trend of the series accurately and predict the sequence more precise than the single SVM model in time series prediction.
【CateGory Index】： E144;E91