混沌时间序列的Volterra级数多步预测研究  

Research on Multi-Step Prediction of Chaotic Time Series using Volterra Series

作　　者：白建东[1] 叶德谦[1] 李春兴[1] 

机构地区：[1]青岛理工大学计算机工程学院,山东青岛266033

出　　处：《计算机仿真》2008年第6期274-276,280,共4页Computer Simulation

基　　金：教育部留学回国人员科研基金(2002498)

摘　　要：针对混沌时间序列在多步预测中自适应预测方法的预测器系数无法调节的问题,根据混沌时间序列的短期可预测性及自适应算法的自适应跟踪混沌运动轨迹的特点,并基于混沌序列产生的确定性和非线性机制、混沌动力系统相空间延迟坐标的重构及二阶Volterra自适应滤波模型,给出了一种混沌时间序列的Volterra级数多步预测方法。在多步预测中,根据已知的样本得到对将来值的预测。仿真结果表明,能够对混沌时间序列进行多步预测,具有较好的预测效果。Based on the definiteness and nonlinear mechanism of the production of chaotic series, the reconstruction of delayed coordination for chaotic - dynamic systems in phase - space and the model of the second - order volterra self- adaptation filtering, also based on the short - term predictability of chaotic time series and the adaptive tracking chaotic trajectory of adaptive algorithm, this paper proposes a multi - step prediction method of chaotic time series using volterra series. This method resolves the problem of adjusting filter＇s parameters of the adaptive prediction method during multi - step prediction. During multi - step prediction this method can get the multi - step prediction value using the known data. The experimental results show that this method is available for predicting chaotic time series by multi -step, and the effect is good.

关 键 词：多步预测 混沌时间序列 重构 相空间 

分 类 号：TP311[自动化与计算机技术—计算机软件与理论]