采用BP神经网络进行混沌运动控制

利用4层BP神经网络对非线性非连续函数的无穷逼近特性，设计了控制方法，对非线性系统的混沌运动进行控制。在系统发生混沌运动时，对于不同类型的期望信号，可以将不同自由度的混沌运动分别控制到各自的目标信号上。目标函数可以是周期函数，非线性函数。对Lorenz方程进行了仿真计算，结果显示可以将混沌运动控制到锯齿波信号、正弦信号和直流信号上；对Rossler方程的仿真计算，可以将系统的混沌运动控制到方波信号与直流信号上，控制所用的时间很短。

Controlling Chaos Motion to Approach Arbitrary Objective Orbit by BP Neural Network

The chaos control of nonlinear dynamic system is investigated. A 4-layers BP neural network was designed and trained to approximate the nonlinear functions. Based on the BP neural network, a control method was presented and proven. For Lorenz and Rossler system, the method was applied to control their chaos motion respectively. The results show that the control method can control the chaos motion in X, Y and Z directions to approach respectively different objective orbits and the control effects can be realized in a short period. The method was also applied to the SFD rotor, which shows it can control the chaos running to move towards the expected periodic motion directly.