混沌时间序列预测模型参数的联合优化

Parameters Joint Optimization of Chaotic Time Series Prediction Model

为了提高混沌时间序列预测模型的预测精度,提出了一种相空间重构和最小二乘支持向量机(LSSVM)参数的联合优化方法.联合优化方法的核心思想是首先采用均匀设计对相空间重构和LSSVM参数进行联合设计,然后采用自调用LSSVM进行参数联合优化,最后利用混沌时间序列对联合优化方法进行验证性测试.实验结果表明,联合优化方法预测精度明显优于其它优化方法,且优化速度更快.

To improve the prediction accuracy of the chaotic time series prediction model,a joint optimization method is proposed for phase space reconstruction and least square support vector machine（LSSVM） parameters.The main idea of the joint optimization method is that phase space reconstruction and LSSVM parameters are jointly designed using uniform design firstly,then the parameters are jointly optimized by a self-calling LSSVM,and the joint optimization method is tested by chaotic time series lastly.The experiment results show that the proposed method obtains better prediction accuracy and higher optimization speed than other prediction methods.