从观察数据确定李雅普诺夫指数谱算法的一种改进

Improvement on Determining Lyapunov Spectrum from Observated Data

由观察数据计算李雅普诺夫指数谱，一个重要的方法是用多项式拟合系统内在的动力学映射，该法得到的结果精度高，但计算量很大。本文通过简化公式降低了计算中所用矩阵的阶数，有效地减少了计算量，对混沌信号数据的实时处理有实际意义。

In this paper we propose an approach which can improve the computation of Lyapunov spectrum from observated date. The conventional method in computing Lyapunov spectrum is to fit the underlying dynamical mapping of a real system with polynomials and the computation is very expensive. The heavy computation burden has been reduced effectively by our approach with decreasing orders of the matrices required in computing Lyapunov spectrum. The restult is valuable in real time processing of chaotic signals.