一种基于概率鲁棒的分散PID控制器设计

Decentralized PID controller design based on probabilistic robustness

基于概率鲁棒方法,针对具有实参数不确定性的多输入多输出被控对象,提出一种分散PID控制器设计方法。根据被控对象模型的参数摄动状态,计算闭环系统满足性能设计要求的概率作为优化算法的目标函数,利用遗传算法对分散PID控制器参数进行优化,用Monte Carlo实验对控制系统进行鲁棒性检验。对5个多变量化工过程进行了仿真试验,并与基于标称参数的设计方法进行比较。仿真结果表明,基于概率鲁棒的分散PID控制器设计方法对模型参数不确定性具有较好的鲁棒性,在被控对象存在一定的不确定性时,系统能以最大的概率满足设计要求。

A tuning method for decentralized PID controllers was developed based on probabilistic robustness for use in multi-input-multi-output plants with varying parameters. The algorithm considers the model uncertainties in calculating the probability that the closed system meets the dynamic performance requirements as the cost function of a genetic algorithm. The algorithm optimizes the decentralized PID controller parameters with the Monte Carlo experiments to ：test the control system robustness. The results for five multivariable chemical processes simulated using the method compare well with results of the design method based on nominal conditions. Thus, the method has better robustness in finding the system parameters having the maximum probability of satisfying the design requirements.