基于正多胞体空间扩展滤波的时变参数系统辨识方法
The orthotopic spatial extension filtering based system identification algorithm for time-varying parameter systems
摘要点击 44  全文点击 42  投稿时间:2019-08-08  修订日期:2019-12-18
查看全文  查看/发表评论  下载PDF阅读器
DOI编号  10.7641/CTA.2019.90661
  2020,37(6):1311-1318
中文关键词  滤波  时变参数  空间扩展  正多胞体  参数辨识
英文关键词  filtering  time-varying parameter  spatial extension  orthotopic  parameter identification
基金项目  国家自然科学基金项目(61802150, 61973138), 江苏省自然科学基金项目(BK20170196), 中国博士后科学基金面上项目(2018M642161), 江南 大学平台基本科研重点项目经费(JUSRP51912B), 江苏省食品先进制造装备技术重点实验室开放课题项目(FM–2019–07)资助.
作者单位E-mail
王子赟 江南大学 wangzy0601@163.com 
张帅 江南大学  
王艳 江南大学  
刘子幸 江南大学  
纪志成 江南大学  
中文摘要
      针对未知但有界噪声时变参数系统, 提出了一种基于正多胞体空间扩展滤波的参数辨识方法. 采用有界误 差方法对测量噪声和参数变化过程进行建模, 通过选取最优扩展系数进而扩大正多胞体大小, 使得正多胞体包含变 化后的参数可行集, 由时不变参数系统约束条件构造扩展系数方程, 通过线性规划方法求解前k步扩展系数值, 选 取最大值作为最终扩展系数. 采用扩展系数更新每一步时变参数正多胞体约束条件, 求解全部参数的上下界得到 包裹参数可行域的最紧致正多胞体. 仿真示例说明该方法辨识时变参数的有效性和准确性.
英文摘要
      A parameter identification method for unknown but bounded noise time-varying parameter systems is proposed, based on the orthotopic spatial extension filtering. The bounded error method is used to model the measurement noise and parameter variation process, and the orthotopic volume is expanded by the optimized expansion coefficient, so that the orthotope contains the changed parameter values, the expansion coefficient equation is constructed by the timeinvariant parameter system constraints, and the first k steps for all the expansion coefficient values can be solved by the linear programming method. Select the maximum value as the final expansion coefficient and use the expansion coefficient to update the variable parameter orthotopic constraints, and solve the minimum and minimum values of each parameter to obtain the most compact orthotope of the feasible domain for wrapping the parameters. The simulations show the effectiveness and accuracy of the presented algorithm when identifying time-varying parameters.