不完备离散事件系统的可诊断性

在离散事件系统的建模过程中,由于系统行为的复杂,存在物理系统向逻辑系统映射的不完全性,因此产生了不完备模型的概念.提出在模型不完备的前提下,判断模型可诊断性的方法.提出可诊断性的在线判定方法,同时将不完备的行为加入模型,使模型完备.用经典的双树方法判断离线可诊断性,根据观测序列的时序及语言的前缀判断并处理不完备行为.提出判定不完备行为的方法,向模型中添加不完备行为,并根据不完备行为增量地在双树中判定在线可诊断性.通过在线的可诊断性判定,当前系统能够得到在有限观测内唯一判定故障发生与否的结论.该方法适用于具有离散性质的系统.

Diagnosability of Discrete Event Systems with an Incomplete Model

In modeling a discrete event system, the map from physical system to logic system may be not complete due to the complex behaviors of the system. In this paper, the concept of incomplete model is introduced. Next, a method for judging diagnosability in incomplete model is proposed, and the corresponding on-line version of the method is also presented. Incomplete behaviors can be added to the model, and makes the model complete. Offline diagnosability is judged by classical twin-plant method. According to the ordered observations and prefix of language, the incomplete behaviors are judged and disposed. With an additional method that judge incomplete behaviors, the incomplete behaviors are added into model, and the online diagnosability is judged incrementally in twin-plant by incomplete behaviors. By judging diagnosability online, whether a fault can be found exclusively by limit observations is decided. The proposed methods suit for the systems which is discrete.