神经网络混沌运动的控制--CM方法

基于压缩映射的混沌控制方法——CM方法被应用到小的离散神经网络,通过一个外部输入的小干扰,稳定混沌轨道嵌入在混沌吸引子内的某一不稳周期轨上。利用闭回路对技术估计欲稳定周期轨的近似位置。给出二维和三维神经网络的典型例子,通过数值模拟显示CM方法控制离散神经网络混沌行为的简单和有效性。

Using the CM method to control chaotic neural networks

The CM(Contraction Mapping) method of chaos control based on contraction mappings is applied to discrete neural networks for stabilizing any unstable desired periodic orbit embedded in the chaotic attractor by a small external input.The close return pairs technique is utilized to estimate an approximate desired periodic orbit.The CM method utilizes only the approximate location of the desired periodic orbit,and does not require the linearization of the system around the stabilized orbit as required in OGY method.The desired periodic orbit can be detected automatically in control process.This approach is suitable for discrete neural networks of arbitrary dimension.To maintain the neural networks system onto the desired orbit,the necessary external input value is very small.Typical examples of neural networks are given,and the numerical simulations show validity and simplicity of the method for controlling the chaotic behavior of discrete neural networks.