基于混沌机制的神经网络预测控制及仿真研究

Simulation Research of Neural Network Predictive Control Based on the Chaos Mechanism

为解决一大类未知、时滞、非线性系统的预测控制问题,提出了一种基于改进的El-man动态回归神经网络预测控制算法.首先,在一般的Elman动态回归神经网络算法中加入了混沌机制,利用混沌机制固有的全局游动有效地消除了Elman网络易陷入局部极值的缺点,提高了系统的辨识速度;然后,结合广义预测控制(GPC)的反馈校正、滚动优化来完成非线性系统的预测控制.仿真结果表明:将本算法应用于非线性系统预测控制,对未建模动态具有较强的鲁棒性和良好的控制跟踪能力.

In order to solve the unknown delay and nonlinear system, the paper presents a predictive prediction based on the improved Elman dynamic recurrent neural network. Firstly chaos searching is imported to the Elman algorithm, make Elman algorithm skip the local minimum and can find the global minimum. And then the improved Elman dynamic recurrent neural network is used to the Generalized Predictive Control, Simulation results show the efficiency of the new neural network and the strategy.