关于最近邻聚类的RBF网络自整定PID控制算法的研究

针对采用传统PID控制一类非线性滞后系统难以获得满意的控制效果,提出基于RBF神经网络的PID控制参数自整定的方法.利用具有在线能力的最近邻聚类学习算法,训练RBF神经网络.并引入优化策略对聚类半径进行自动调整,以保证聚类的合理性,从而自适应调整系统的控制参数.仿真结果证明了该控制策略不仅能使非线性滞后系统具有良好的动态跟踪性能,而且具有很好的抗干扰能力和鲁棒性.

A study of self-tuning PID control algorithm for a general nonlinear hysteretic system based on the RBF neural network

To solve the problem that a conventional PID controller working with a general nonlinear hysteretic object is unable to insure the accuracy of a control system, the paper puts forward the method of self-tuning PID controller based on the RBF neural network. The nearest neigbor clustering algorithm is used to train the RBF neural network. The optimal strategy that can regulate the cluster radius automati- cally is introduced into the guarantee of the cluster rationality. Simulation results show that the control- strategy can not only have a favorable dynamic tracking performance to the nonlinear hysteretic system, but also possess resistance to disturbance and excellent robustness.