引用本文:周祖鹏,谭永红.基于鲁棒观测器的带间隙三明治系统故障预报[J].控制理论与应用,2015,32(6):753~761.[点击复制]
ZHOU Zu-peng,TAN Yong-hong.Fault prediction for sandwich systems with backlash based on robust observer[J].Control Theory and Technology,2015,32(6):753~761.[点击复制]
基于鲁棒观测器的带间隙三明治系统故障预报
Fault prediction for sandwich systems with backlash based on robust observer
摘要点击 2147  全文点击 952  投稿时间:2014-06-13  修订日期:2015-02-09
查看全文  查看/发表评论  下载PDF阅读器
DOI编号  10.7641/CTA.2015.40545
  2015,32(6):753-761
中文关键词  鲁棒观测器  间隙  三明治系统  故障预报
英文关键词  robust observer  backlash  sandwich system  fault prediction
基金项目  国家自然科学基金项目(61263016, 60971004, 61171088, 61203108), 广西重点实验室主任基金项目(13--071--11--62--006, 12--071--11--62--003), 上海市教育委员会科研创新项目(13YZ056), 上海市科委国际科技合作项目(14140711200), 上海市自然科学基金项目(09ZR1423400,10ZR1422400)资助.
作者单位E-mail
周祖鹏 桂林电子科技大学 机电工程学院 zhouzupeng@guet.edu.cn 
谭永红* 上海师范大学 信息与机电工程学院 tany@shnu.edu.cn 
中文摘要
      工程实践中常见的带间隙的三明治系统的准确故障预报具有重要的现实意义, 为此, 本文构建了一种新的 动态鲁棒观测器对其进行故障预报. 首先, 通过将非光滑项转化为干扰项的方法, 将间隙非光滑 三明治系统转化为可用动态鲁棒观测器设计方法设计的系统. 其次, 采用零点配置和最小化基准 区间观测器的范数(H_{\infty, F}/H_{ - , F})指标的方法确定动态鲁棒观测器的增益矩阵. 最后, 通过仿真, 分别比较了基于非光滑鲁棒观测 器和基于传统观测器的故障预报效果, 比较结果表明:鲁棒观测器能够及时地准确预报传统观测 器无法预报的故障, 且有效减少了故障的漏报和错报现象.
英文摘要
      In the practical application, realizing the accurate fault prediction of non-smooth sandwich system with backlash which widely exists in the real engineering systems is of great importance. Therefore, a novel dynamic robust observer is proposed in this paper. First, the non-smooth sandwich system with backlash is transformed into a system in which the conventional dynamic robust observer design method can be applied by converting the non-smooth item into a general disturbance. Then, by zeros assignment and minimizing the (H_{\infty, F}/H_{ - , F}) indicator on the basic operation zone, the feedback matrices of this observer can be obtained. Finally, the simulation results are presented. The comparison between the proposed robust non-smooth scheme and the conventional method demonstrates that the proposed robust fault detection approach can detect the actuator and sensor faults accurately and rapidly while the conventional observer cannot detect the fault at all. Moreover, the missing and false alarms are effectively reduced by the robust observer.