基于不确定观测下离散事件系统可诊断性的研究

Research on diagnosability of discrete event systems based on nondeterministic observations

在实际应用系统中,由于传感器故障、传感器限制和网络中的数据包丢失等原因,事件的可观测值变得不确定,使得观测系统行为变得尤为复杂。针对离散事件系统中,同个事件串可能有多个观测值以及不同状态下同个事件观测值也可能不同的问题,提出一种不确定观测下故障诊断验证的方法。首先对不确定观测的离散事件系统的可诊断性进行形式化,然后构建出用于上述故障诊断验证的验证器;基于验证器提出了系统基于不确定观测下可诊断的充要条件及验证算法;最后,实例说明不确定观测下故障诊断验证算法的应用。与现有研究相比,提出的方法对故障事件的观测值没有约束,可以为0个或多个观测值,使此方法应用的场景更为广泛。

In practical systems,due to reasons such as sensor faults,sensor limitations and packet losses in networks,the observation of events becomes nondeterministic,and makes observing system behavior particularly complex.Aiming at the problem that the same event string might have multiple observations in discrete event systems and the same event observations might be different in different states,this paper proposed a method for fault diagnosis and verification under nondeterministic observations.Firstly,it formalized the diagnosability of discrete event systems under nondeterministic observations.Secondly,it constructed a verifier for the above fault diagnosis verification.Based on the verifier,it proposed a necessary and sufficient conditions and the verification algorithm for the system to be diagnosable under nondeterministic observations.Finally,an example illustrates the application of the fault diagnosis verification algorithm under nondeterministic observations.Compared with the existing research,the observation value of the fault event is not constrained in this proposed method,and it can be zero or more observations,which makes the application of this method more extensive.