存在未知时延和Markov丢包的网络控制系统故障检测与优化  

Fault detection and optimization for networked control systems with unknown delay and Markov packet dropouts

作　　者：王昭磊[1] 王青[1] 董朝阳[2] 牛尔卓[3] 

机构地区：[1]北京航空航天大学自动化科学与电气工程学院,北京100191 [2]北京航空航天大学航空科学与工程学院,北京100191 [3]中航工业西安飞行自动控制研究所,西安710065

出　　处：《控制与决策》2014年第9期1537-1544,共8页Control and Decision

基　　金：国家自然科学基金项目(61074027)

摘　　要：将网络控制系统(NCSs)的未知短时延处理成范数有界不确定性,结合Markov丢包影响将NCSs建模为不确定Markov跳变系统,设计模态依赖的鲁棒故障检测滤波器.为了提高检测系统性能,采用后置滤波器对残差信号进行时域优化,并以Moore-Penrose逆形式给出其最优解.同时,设计自适应检测阈值,并给出时变参数阵的迭代方法,降低了计算量.数值仿真表明,所提出的方法能够有效地抑制时延和丢包影响,提高故障检测系统的检测能力和检测速度.Considering the influence of the Markovian packet dropouts and the unknown short delay which can be treated as norm-bounded uncertainties, the networked control systems（NCSs） are modeled as the uncertain Markov jump linear systems（UMJLS）. Then the mode dependent robust fault detection filters are designed based on the UMJLS model. Furthermore, a time domain optimization approach using the post-filter is proposed to improve the fault detection performance, and the optimal solution of the post-filter is derived in the form of Moore-Penrose inverse of matrix. An adaptive threshold is also designed, and a iterative algorithm is introduced to reduce the computation. Finally, numerical examples show that the proposed method can restrain superiorly the impact of the delays and packet dropouts, and improve the detection capability and detection speed of the fault detection system.

关 键 词：网络控制系统 故障检测 时延及丢包 时域优化 

分 类 号：TP273[自动化与计算机技术—检测技术与自动化装置]