一种面向无模型系统的PID参数整定法  

A PID Parameters Tuning Method for Model-free Systems

作　　者：胡迪锋 周向阳[1] 刘瑛[1] HU Di-feng;ZHOU Xiang-yang;LIU Ying(Department of Special Equipment,Hangzhou Vocational and Technical College,Hangzhou 310018,China)

机构地区：[1]杭州职业技术学院特种设备学院,浙江杭州310018

出　　处：《控制工程》2020年第11期1962-1969,共8页Control Engineering of China

基　　金：浙江省教育厅资助科研项目(Y201738464,Y201840430)。

摘　　要：当前先进PID参数整定方法都是基于系统模型的,然而许多复杂系统无法用某一精确的数学模型进行描述。为此,提出应用响应面法与灰度分析对无模型系统进行PID参数整定。对于无模型系统来说,最好的评价标准是直接的输出响应,例如超调量、上升时间和调整时间等,所以无模型系统的PID参数整定是一个多目标优化问题。因此首先应用灰度分析对多个性能指标进行综合评估;然后应用响应面法建立PID参数与综合性能间的数学关系式,进而预测最优的PID参数组合;最后此方法应用于3种典型的系统以验证该方法的可行性。仿真结果显示使用该方法得到的PID参数组合的控制性能优于Z-N法与遗传算法,与粒子群算法不相上下,但搜索效率高于粒子群算法。Nowadays,most of advanced PID tuning methods are based on system model,but many complicated systems cannot be described by accurate mathematical models.So,a PID parameters tuning method using response surface method(RSM)and grey relational analysis(GRA)is proposed for model-free systems.For model-free systems,the best evaluation criterion is simple output response,such as overshoot,rise time,and settle time,so tuning PID parameters of a model-free system is a multi-criterion problem.Therefore,firstly GRA is used to evaluate controlling performance from the multi-criterion viewpoint.Then,RSM is used to build a surrogate mathematical model between PID parameters and comprehensive performance,and the optimal PID parameters combination can be predicted.In order to verify the effectiveness of this proposed method,three typical kinds of systems are investigated.The simulation results show that this proposed method has better performance than Z-N method and Genetic Algorithm.Besides,compared with PSO,although the controlling performance is about the same,but it has higher searching efficiency.

关 键 词：无模型系统 PID参数整定 响应面法 灰度分析 输入输出响应 

分 类 号：TP13[自动化与计算机技术—控制理论与控制工程]