混沌时间序列的混合预测方法

提出了一种基于小波变换、粒子群优化的最小二乘支持向量机(PSO-LSSVM)和广义自回归条件异方差模型(GARCH)的混沌时间序列的混合预测方法.首先利用小波变换将混沌时间序列分解和重构成概貌时间序列和细节时间序列; 然后利用PSO-LSSVM模型预测概貌时间序列的未来值,采用GARCH模型预测细节时间序列的未来值;最后将概貌时间序列和细节时间序列的未来值求和作为最终的预测结果.采用该方法对Mackey-Glass和变参数Logistic混沌时间序列进行预测. 结果表明该方法能精确地预测混沌时间序列,验证了文中所提方法的有效性.

Prediction of the chaotic time series using hybrid method

A chaotic time series hybrid forecasting method based on wavelet transform, least squares support vector machine (LSSVM) optimized by particle swarm optimization (PSO) and generalized autoregressive conditional heteroscedasticity (GARCH) is proposed. Firstly, the chaotic time series is decomposed and reconstructed into approximate series and detailed series. Secondly, the approximate series future values are predicted by PSO-LSSVM model; while the detailed series future values are forecasted by GARCH model. Finally, the sum of the approximate and detailed series future values is used as the final forecasting values. Two chaotic time series, namely, Mackey-Glass and variable-parameter Logistic, are used to evaluate the performance. The results imply that the proposed method can provide more accurate results, which demonstrates the validity of this method.