基于Bernstein多项式的自适应混沌时间序列预测算法

提出了利用Bernstein多项式对混沌时间序列的动力学方程进行建模的方法,并将该方法与递推最小二乘(RLS)算法相结合,从而可以自适应地逼近混沌时间序列的动力学特性,以达到预测的目的.理论分析和仿真实验表明该方法对一些常见的混沌时间序列具有较高的预测精度和较理想的准确预测率.由于RLS算法的收敛速度较快,因此该方法比较适合于对短混沌时间序列进行实时预测. 关键词： 混沌 预测 Bernstein多项式 RLS算法

An adaptive approach based on Bernstein polynomial to predict chaotic time series

In this paper, we propose an approach using the Bernstein polynomial to model the dynamics of chaotic time series. Combining it with RLS algorithm, we can predict the chaotic time series adaptively. Theoretical analysis and computer simulation have demonstrated that this approach can provide high precision and satisfactory percentage of prediction for some typical chaotic time series. Because of the fast convergence of RLS algorithm, this approach can be applied to predicting short record chaotic time series in real time.