模型精度对自控性能的影响及谨慎控制策略  

Impacts of model fitness on automatic process control performance and cautious control strategy

作　　者：叶亮[1] 潘尔顺[1] 奚立峰[1] 

机构地区：[1]上海交通大学机械与动力工程学院,上海200030

出　　处：《计算机集成制造系统》2007年第10期1952-1957,共6页Computer Integrated Manufacturing Systems

基　　金：国家自然科学基金资助项目(50405016)。~~

摘　　要：为消除过程模型精度对自动过程控制性能的影响,根据仿真分析了试验设计与回归模型精度之间的关系。针对较低精度模型会恶化自动控制性能的问题,将过程模型估计误差引入到控制模型的目标函数中,提出了谨慎控制策略,改善了过程控制性能对模型精度的依赖程度。使用谨慎控制律,能够使过程质量在使用模型精度较高和较低的情况下都能保持在较好的水平,大大提高了自动过程控制对模型精度的鲁棒性,也降低了试验成本。To eliminate impact of process model fitness on automatic process control performance, relationship between design of experiment approach and regression model fitness based on simulation was analyzed. To improve the deteriorated control performance by the poorer regression model, a cautious control strategy was proposed by introducing estimated errors of process model into objective function of control model, which eased the dependency of control performance on model precision. Using cautious control law, the process quality could keep better performance no matter with good regression model or a poor one. Meanwhile, the proposed cautious control law improved robustness of control performance to the model uncertainty and decreased experiment cost.

关 键 词：试验设计 自动过程控制 鲁棒参数设计 谨慎控制 

分 类 号：O231[理学—运筹学与控制论]