摘要
针对非线性系统时滞问题,给出了一种新型的单神经元Smith预测控制算法。神经网络的预测控制器由不完全微分的单神经元自适应PID控制器和神经网络的Smith预估器组成。预估器对输出进行多步预测,控制器超前动作以消除时滞对系统的影响。不完全微分的单神经元自适应PID控制器通过改进的Hebb学习规则实现其权值调节,通过权系数的在线调整实现自适应控制。仿真实验证明了该方法具有较快的响应速度和较好的响应性能。
To solve time - delay problems of nonlinear system, a new single neural Smith predictive control algorithm is presented. The predictive controller based on neural network is composed of adaptive incomplete derivative adaptive PID controller in neuron unit and Smith estimator based on neural network. The output was retested with Smith estimator, which made controller operate to eliminate influence of time delay upon the system. The weight regulation of neural network was realized with incomplete derivative adaptive PID controller through an improved Hebb algorithm. And the adaptive control was implemented through regulating weight coefficient online. Simulation results show that the proposed method has higher response speed and better response performance.
出处
《计算机仿真》
CSCD
北大核心
2009年第2期187-189,220,共4页
Computer Simulation