网络环境下不敏感系统的分布式区域预测控制方法
Distributed predictive control method with zone control for insensitive system under network environment
摘要点击 1043  全文点击 630  投稿时间:2014-06-05  修订日期:2014-12-24
查看全文  查看/发表评论  下载PDF阅读器
DOI编号  10.7641/CTA.2015.40512
  2015,32(4):504-512
中文关键词  分布式控制系统  可控性  区域控制  条件数  不敏感系统
英文关键词  distributed control systems  controllability  zone control  conditional number  insensitive systems
基金项目  国家自然科学基金项目(61374112), 国家高技术研究发展计划(“863”计划)项目(2014AA041802), 中国科学院重点部署项目(KGZD--EW--302), 中国博士后科学基金项目(2013M530953)资助.
学科分类代码  
作者单位E-mail
庞强 中国科学院沈阳自动化研究所 信息服务与智能控制技术研究室 pangqiang@sia.cn 
邹涛 中国科学院沈阳自动化研究所 信息服务与智能控制技术研究室  
丛秋梅 沈阳中科博微自动化技术有限公司  
中文摘要
      针对二分图网络结构下的不敏感系统的可控性比较差的问题, 本文提出了带有区域控制的分布式模型预测控 制方法. 该方法首先利用分支定界法对分布式控制系统结构进行最优设计; 然后, 结合区域控制提高控制系统 的动态性能指标; 最后, 依据回路之间的关联性的强弱选择每个回路的控制方式(精确控制或区域控制). 仿真结果 显示带有区域控制的分布式预测控制系统的调节时间比没有带区域控制的分布式控制系统明显缩短; 同时, 通过对回路 控制方式的选择增加了精确控制的智能体数量. 仿真结果证明了利用分布式区域预测控制方法可以提高系统的容错性, 通过对 系统的可控性和关联性进行分析, 可以在精确控制和区域控制之间寻找到最优组合, 从而达到快速精确的控制效果.
英文摘要
      Because the controllability of insensitive system with bipartite graph network structure is very poor, we propose a distributed model predictive control (DMPC) method with zone control. Firstly, the optimal controllability structure of DMPC is designed using a branch and bound algorithm. Then, the DMPC is combined with zone control to improve the dynamic performance index of control system. Finally, the control mode (precise control or zone control) of each loop is selected according to the correlation degree between the loops. The simulation results show that the settling time of distributed model predictive controller with zone control is significantly shorter than that of distributed model predictive controller without zone control. The number of intelligent agents being precisely controlled becomes greater than that before selecting the control mode of the loop. It is demonstrated that the distributed predictive control method with zone control improves the fault tolerance of control system, and the optimal combination between precise control and zone control can be determined through the analysis of the controllability and relevance. The control performance indices can be achieved rapidly and precisely.