基于概率鲁棒性的锅炉过热汽温串级PID控制器  

Cascade PID controllers for superheated steam boiler temperatures based on probabilistic robustness

作　　者：王传峰[1] 李东海[1] 姜学智[1] 王维杰[1] 

机构地区：[1]清华大学热能工程系,热科学与动力工程教育部重点实验室,北京100084

出　　处：《清华大学学报（自然科学版）》2009年第2期249-252,共4页Journal of Tsinghua University(Science and Technology)

基　　金：国家“十一五”科技支撑计划资助项目(2006BAA03B04)

摘　　要：基于概率鲁棒性理论,针对过热汽温这类工况参数对模型参数有较大影响的对象,提出一种固定参数的串级鲁棒PID控制器设计方法。根据被控对象模型的参数摄动范围及概率密度分布,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,利用遗传算法对串级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。仿真结果表明,基于概率鲁棒的串级PID控制器对于对象工况参数引起的模型参数不确定性具有较好的鲁棒性,系统可以兼顾多个工况点的性能,以最大概率满足设计要求,最大限度地挖掘了固定参数串级PID控制器对模型参数具有不确定性对象的控制能力。This paper presents a cascade controller design method based on probabilistic robustness for superheated steam boiler temperatures, which vary significantly as the load changes. The probability of the closed system following the load variations was calculated for various perturbations of the parameters and probability density distribution as the cost function of a Genetic Algorithm to optimize the cascade PID controller parameters. Monte-Carlo experiments of the control system robustness show that the controllers have good robustness to plant parameter uncertainties caused by the load changes. For all operating conditions, the system satisfies the design requirements with the a maximum probability. This method makes maximum use of the ability of cascade PID controllers to control whose parameters vary significantly.

关 键 词：过热汽温 串级PID控制器 概率鲁棒性 Monte-Carlo实验 

分 类 号：TP214[自动化与计算机技术—检测技术与自动化装置]