约束不确定系统的误差有界无偏跟踪
DOI:
作者:
作者单位:

武汉大学电子信息学院

作者简介:

通讯作者:

中图分类号:

TP273

基金项目:

国家自然科学基金项目(面上项目,重点项目,重大项目)


Error-bounded and offset-free tracking for constrained uncertain systems
Author:
Affiliation:

Electronic Information School, Wuhan University

Fund Project:

The National Natural Science Foundation of China (General Program, Key Program, Major Research Plan)

  • 摘要
  • |
  • 图/表
  • |
  • 访问统计
  • |
  • 参考文献
  • |
  • 相似文献
  • |
  • 引证文献
  • |
  • 资源附件
  • |
  • 文章评论
    摘要:

    为使精密制造设备中定位机构实现在容许的误差带内对时变轨迹的精准无偏跟踪,针对约束不确定系统提出一种基于鲁棒控制不变(RCI)集的误差有界且无偏模型预测控制(EOMPC)方法. 首先为了消除由扰动引起的稳态误差, 构建包含可测输出、估计状态和估计扰动的增广系统作为EOMPC的预测模型以提高预测精度. 其次基于增广系统和轨迹模型, 使用可以在有限步内终止的迭代方法求解RCI集,并将其作为最优控制问题(OCP)中的状态约束以实现误差有界的跟踪. 然后为了保障OCP的实时性,给出了可在线执行的数值优化方案. 最后,在磁悬浮定位系统上验证了方法的有效性.

    Abstract:

    In order to achieve the precise and offset-free tracking of the positioning mechanism in precision manufacturing for the time-varying trajectories within the allowed error bound, an error-bounded and offset-free model predictive control (EOMPC), based on robust control invariant (RCI) set, is proposed for constrained uncertain systems. Firstly, for the purpose of removing the steady-state errors caused by the disturbances, an augmented system that contains the system output, estimated state, and estimated disturbance, is constructed and used as the prediction model to improve the prediction accuracy. Secondly, on the basis of the augmented system and trajectory model, an iterative procedure that can be terminated in finite steps is presented to obtain the RCI set, which is used as the state constraints in optimal control problem (OCP) to promise the error-bounded tracking. A numerical optimization scheme that can be executed online is then given for the real-time performance of OCP. Finally, the effectiveness of the method is illustrated on the maglev positioning system.

    参考文献
    相似文献
    引证文献
引用本文
分享
文章指标
  • 点击次数:
  • 下载次数:
  • HTML阅读次数:
  • 引用次数:
历史
  • 收稿日期:2021-08-01
  • 最后修改日期:2021-11-04
  • 录用日期:2021-11-10
  • 在线发布日期:
  • 出版日期: