支持向量机在混沌系统预测中的应用

Application of Support Vector Machine to Prediction of Chaotic Systems

提出用支持向量机回归方法解决混沌系统预测问题。阐述了支持向量机回归算法,对四阶混沌时间序列进行预测,在此基础上结合城市交通的混沌性,对珠海市迎宾大道的交通流量进行预测。仿真实验表明,支持向量机泛化能力好、学习速度快,对混沌时间序列具有很好的预测效果,对城市交通流量预测也是切实可行的。

Support Vector Machine （SVM） regression method is proposed for solving the prediction problem of chaotic systems. The algorithm of SVM regression is formulated. One application is given to the prediction of four-order chaotic time series. Based on this application and in conjunction with the chaotic characteristics of urban traffic, another application is given to the real-time prediction of the traffic flow in Zhuhai city. Simulation experiments show that SVM regression has fast learning ability and good generalization. It is very suitable for the prediction of chaotic time series, and is also effective to forecast urban traffic flow.