一种开环输出反馈预测控制
An open-loop output feedback predictive control
摘要点击 317  全文点击 273  投稿时间:2018-07-13  修订日期:2019-04-03
查看全文  查看/发表评论  下载PDF阅读器
DOI编号  10.7641/CTA.2019.80519
  2020,37(1):31-37
中文关键词  多胞描述模型, 启发式预测控制, 稳态目标计算, 人工干扰
英文关键词  Polytopic description model, Heuristic predictive control, Steady state target calculation, artificial disturbance
基金项目  国家自然科学基金(61573269),国家重点基础研究发展计划(2017YFA0700300)
作者单位E-mail
胡建晨 西安交通大学 371658387@qq.com 
王勇 西安交通大学  
丁宝苍 西安交通大学 baocangding@126.com 
中文摘要
      考虑状态不可测且具有多胞不确定性的系统. 本文通过定义参数依赖的顶点控制作用, 提出一种基于多胞描述模型 的启发式开环输出反馈预测控制方案, 其中顶点控制作用依赖于多胞描述模型参数, 由多胞顶点值构成. 在工业预测控制的 框架下, 方案包含开环预测、稳态目标计算、动态控制三个模块. 上层稳态目标计算模块采用稳态模型优化得到稳态目标值, 下层动态控制跟踪由稳态目标值计算得到的多胞顶点稳态目标值. 考虑到工业预测控制中稳态目标计算模块常采用线性模型, 本文通过在多胞描述模型中加入人工干扰进而改进控制方案, 使得上层稳态目标计算模块可直接采用改进的稳态多胞描述模型, 下层基于改进的动态预测方程跟踪顶点稳态目标值. 通过仿真分别验证了两种方法的可行性和有效性.
英文摘要
      For system with unmeasurable state and polytopic model uncertainty, through defining parameter-dependent vertex control moves in this paper, a heuristic open-loop output feedback predictive control method based on polytopic description model is proposed, among which vertex control moves are depending on polytopic description model parameters, and are constructed by polytopic vertices values. Under the framework of industrial predictive control, it includes open-loop prediction, steady state target calculation and dynamic control. The upper layer steady state target calculation uses steady state model to optimize for the steady state target, the lower layer dynamic control follows the polytopic vertex steady state target obtained from steady state target calculation. Considering the fact that the steady state target calculation module in industrial predictive control always uses linear model, in this paper the artificial disturbance term is added to polytopic description model in order to improve the method, which makes the polytopic description model available for the upper layer steady state target calculation module, the lower layer follows the steady state target based on the improved dynamic predictive equation. The feasibility and effectiveness of the two methods are verified through simulation.