基于神经网络的Lyapunov指数谱的计算

The Calculation of Lyapunov Exponents Based on the Neural Network

利用 BP神经网络的非线性函数逼近能力 ,由实验观察数据列计算系统的 Lyapunov指数谱 .实例计算表明 ,此种方法精度较高且计算量较小 。

Lyapunov exponents are one of a number of parameters that characterize the nature of a chaotic dynamical system. We calculate the Lyapunov exponents from an observed time series based on the ability that a BP neural network can approximate nonlinear functions. The results show that this method needs less computing time and has higher precision ,so it has practical significance.