Fault Diagnosis of Bounded Petri Nets Using Path Marking Graphs

This paper proposes a fault diagnosis method based on the path information of a bounded Petri net model. It relies on the notions of two graphs called path marking graph and modified path marking graph, which characterize different sets of paths and markings consistent with an observed sequence, and a set of unobservable transitions whose firing will enable the sequence. The fault diagnosis method can be applied to a bounded Petri net model with unobservable and acyclic subnets. Moreover, the process of using these two particular graphs to calculate the possibilities of fault behaviors can be taken off-line, which is the most burdensome part for a bounded Petri net model, but also can reduce the memory cost. Finally, this method is verified through a production line example. The main contribution of this paper is to dig out the path information of a bounded Petri net, and to map it into different categories. By analyzing the inherent structure information of Petri net, the presented algorithm can diagnose faults without traversing all the system states.