基于BEMD的染噪混沌时间序列预测方法

基于经验模态分解(EMD)方法对染噪混沌时间序列进行预测时,模态混叠会降低预测精度和最大可预测时间.针对这一问题,将复数据经验模态分解(BEMD)引入到染噪混沌时间序列的预测,在BEMD过程中以高斯白噪声分解的内禀模态函数(IMF)为基函数来驱动染噪混沌信号的分解,从而减小模态混叠对混沌预测的影响.Lorenz混沌时间序列和Henon混沌时间序列的预测实例表明,本方法相对于EMD方法在预测精度和最大可预测时间上都有一定程度的提高.

Prediction of Noisy Chaotic Time Series Based on BEMD

The prediction accuracy and reliable prediction time will be decreased because of mode mixing when the empirical mode decomposition (EMD) is used to predict noisy chaotic time series,a method based on bivariate empirical mode decomposition (BEMD) is proposed to tackle this problem,in which the intrinsic mode functions (IMF) of Gaussian white noise are used as the data-driven basis functions for the decomposition of noisy chaotic time series,thus the effects of mode mixing are weakened.The simulation results of Lorenz chaotic time series and Henon chaotic time series both show that this method has more performance advantages on prediction accuracy and reliable prediction time than the method based on EMD.