Parity space-based fault detection for Markovian jump systems
Parity space-based fault detection for Markovian jump systems
复制标题
DOI：
10.1080/00207720802556237
复制
发表时间：
2009-04
影响因子：
4.3
通讯作者：
M. Zhong;Q. Ding;P. Shi
中科院分区：
文献类型：
--
作者：
M. Zhong;Q. Ding;P. Shi
研究了一类离散线性马尔可夫跳变系统基于奇偶空间的故障检测问题。提出了一种新的模式相关冗余关系参数矩阵的计算方法。与线性时不变系统不同，马尔可夫跳变系统的奇偶空间残差发生器不能离线设计，因为它依赖于最后有限步中系统模式的历史。为了克服这个困难，一个有限的奇偶校验矩阵的集合是预先设计的应用统一的方法，线性时不变系统。这样就可以很容易地实现在线残差生成。此外，还考虑了残差估计问题，包括残差估计函数和阈值的确定。最后通过数值算例说明了该方法的有效性。
This article deals with problems of parity space-based fault detection for a class of discrete-time linear Markovian jump systems. A new algorithm is firstly introduced to reduce the computation of mode-dependent redundancy relation parameter matrices. Different from the case of linear time invariant systems, the parity space-based residual generator for a Markovian jump system cannot be designed off-line because it depends on the history of system modes in the last finite steps. In order to overcome this difficulty, a finite set of parity matrices is pre-designed applying a unified approach to linear time invariant systems. Then the on-line residual generation can be easily implemented. Moreover, the problem of residual evaluation is also considered which includes the determination of a residual evaluation function and a threshold. Finally, a numerical example is given to illustrate the effectiveness of the proposed method.