PID控制器参数优化中的仿真研究  

Simulation Research for Optimization PID Parameters

作　　者：高觐悦[1] 

机构地区：[1]苏州市职业大学计算机工程系,江苏苏州215104

出　　处：《计算机仿真》2011年第3期259-262,共4页Computer Simulation

摘　　要：研究PID控制器参数优化选择问题。传统的PID参数优化多采用试验加试凑的方式,费时且难找到最优参数,同时,常规BP神经网络PID参数优化训练收敛速度较慢,控制效果不理想。为了获得最优PID参数,提高系统的控制性能,提出一种遗传算法BP神经网络PID参数优化方法。方法首先采用全局寻优的遗传算法对BP神经网络初始权值进行选择,然后采用参数最优的BP神经网络模型对PID控制参数进行优化,从而实现PID控制参数实时优化。仿真结果表明,相对于传统的PID参数优化方法和常规BP神经网络PID参数优化方法,方法取得了更满意的控制效果,提高了系统的控制精度,验证了算法的有效性。Researching on PID controller parameters optimization problems.The traditional PID parameters optimization mainly uses the methods of tests or trial and error,which takes time and is difficult to find the optimal parameters.Conventional BP neural network for PID parameters optimization has low training convergence speed,so that the optimum parameters and the control effect are not ideal.In order to obtain the optimal PID parameters and improve the system control performance,this paper puts forward a genetic algorithm based BP neural network for PID parameters optimization.This method firstly uses the global optimization function of genetic algorithm to train the weights of BP neural network,then chooses the BP neural network with optimal parameters to optimize the PID control parameters,thus realizes the real-time optimization of PID control.Simulation results show that,compared with the traditional PID parameters optimization method and the BP neural network based PID parameters optimization method,this method has more satisfactory control effect and improves the control precision of the system.The paper verifies the effectiveness of the algorithm.

关 键 词：比例积分微控制器 神经网络 遗传算法 参数优化 

分 类 号：TP273[自动化与计算机技术—检测技术与自动化装置]