引用本文:王素珍,辛诚,孙国法.连续搅拌反应釜的自适应输出反馈控制及参数整定[J].控制理论与应用,2021,38(10):1587~1596.[点击复制]
WANG Su-zhen,XIN Cheng,SUN Guo-fa.Adaptive output feedback control and parameter tuning for continuous stirred tank reactor[J].Control Theory and Technology,2021,38(10):1587~1596.[点击复制]
连续搅拌反应釜的自适应输出反馈控制及参数整定
Adaptive output feedback control and parameter tuning for continuous stirred tank reactor
摘要点击 1501  全文点击 438  投稿时间：2020-10-25  修订日期：2021-08-22
查看全文  查看/发表评论  下载PDF阅读器
DOI编号  10.7641/CTA.2021.00742
  2021,38(10):1587-1596
中文关键词  连续搅拌反应釜  扩张状态观测器  反步法  自适应控制  参数整定  非线性系统
英文关键词  continuous stirred tank reactor  extended state observer  backstepping  adaptive control  parameter tuning  nonlinear system
基金项目  国家自然科学基金项目(61640302, 61703224), 国家级大学生创新项目(GC202010429061)资助.
作者单位邮编
王素珍 青岛理工大学 266520
辛诚* 青岛理工大学 266520
孙国法 青岛理工大学 
中文摘要
      针对一类连续搅拌反应釜(CSTR), 提出了基于扩张状态观测器和反步法的自适应控制方法, 并结合连续动 作强化学习器(CARLA)进行控制器参数整定. 将CSTR视为包含不确定函数的非严格反馈非线性系统, 利用扩张状 态观测器对系统中的状态变量实时估计, 并对不确定函数在线逼近, 将系统补偿为线性二阶积分串联系统, 为其设 计反步法控制器. 通过李雅普诺夫稳定性定理对系统稳定性进行分析, 证明了闭环系统中所有信号均有界. 最后, 针 对大量控制器参数难以人工整定的问题, 设计CARLA算法快速搜索控制器参数最优值, 提升了控制品质. 仿真实验 进一步验证了该方法的有效性.
英文摘要
      For a class of continuous stirred tank reactor (CSTR), an adaptive control method based on extended state observer (ESO) and backstepping method is proposed. Meanwhile, continuous action reinforcement learning automata (CARLA) is used to tune the controller parameters. The CSTR is regarded as a non-strict-feedback nonlinear system with uncertain functions. The extended state observer is used to estimate the state variables of the system in real time, and the uncertain function of the system is approximated online by that. In this way, the system is compensated as a linear secondorder integral series system, and the backstepping controller is designed for it. The stability of the system is analyzed by Lyapunov stability theorem, which is proved that all signals in the closed-loop system are bounded. Furthermore, CARLA algorithm is designed to search the optimal value of controller parameters rapidly, which solves the problem that a large number of controller parameters are difficult to be tuned manually, and the control quality is improved. Finally, the simulation results verify the effectiveness of this method, and the superiority of that is further proved by comparing with other methods.