A fault diagnosis method for systems based on labeled time Petri nets with tables

Abstract Existing fault diagnosis problems in discrete event systems are typically modeled using (variants of) Petri nets, where faults are detected based on modified state class graph (MSCG). However, these methods overlook data flow in systems, leading to models that cannot accurately describe system operational behavior. Additionally, existing fault detection methods often fail to identify faults related to data flow due to the lack of data flow descriptions in MSCG, resulting in misjudgments. To address these issues, data flow needs to be considered when modeling the system. Specifically, we extended the labeled time Petri net by adding corresponding data element operations and database table operations on transitions. This extension enables the simulation of data flow and reflects interactions between data elements, resulting in a new model called the labeled time Petri net with table. Additionally, we refined the MSCG by incorporating data element information and the values of data items in the database table, generating a new graph called MSCG$^{*}$. MSCG$^{*}$ effectively shows the data flow changes in the system. Based on the MSCG$^{*}$, we identify transition sequences that are logically consistent with the observable label sequences. Linear inequalities are then constructed from these transition sequences to find effective transition sequences that satisfy the time constraints. The faults in the system caused by data flow require further detection and verification. We define the system’s data constraints as follows: if the states in MSCG$^{*}$ satisfy the data constraints, the system has no faults; otherwise, the system has faults. This diagnostic method detects faults in the system from the perspectives of time and data constraints, thus accurately judging the system’s fault states. Finally, we verified the feasibility and effectiveness of the proposed method through case analysis.