基于微分器与RBF神经网络补偿的PID控制

An Improved PID Control Algorithm Based on Tracking Differentiator Compensation and Adaptive RBF Neural Network

在实际工业控制环境下,控制系统存在不确定项及非线性问题,传统PID控制方法的环境适应性和抗干扰能力有限,控制精度和快速性等指标难以满足日益增长的控制需求。针对这一问题,提出一种基于跟踪微分器与自适应RBF神经网络补偿的PID控制算法。上述方法采用RBF神经网络逼近系统不确定非线性函数;并通过非线性跟踪微分器实现位置信号滤波与速度信号求解;最后,采用PID控制技术核心架构设计了自适应非线性控制器。为了验证算法的有效性,针对倒立摆系统典型被控对象进行了仿真研究。仿真结果表明,改进方法能有效实现倒立摆系统的快速跟踪,相对于传统控制方法,在精度、快速性及鲁棒性方面都具有一定优势。

Due to the uncertainties and nonlinearities in actual industrial control environment, the traditional PID control method has limit environmental adaptability and anti - interference ability with low control precision and ra- pidity. In order to solve this problem, this paper puts forward an improved PID control algorithm based on tracking differentiator and adaptive RBF neural network. The RBF neural network is used to approximate uncertain nonlinear functions, and the nonlinear tracking differentiator is used to realize position signal filtering and solve speed signal. Based on the PID control technology, an adaptive nonlinear controller is designed. To verify the effectiveness of this method, an inverted pendulum system is taken as a typical simulation object. Simulation results show that the pro- posed method can effectively realize the fast track of inverted pendulum system, and has higher accuracy, speediness and robustness than the traditional control method.