Decentralized Fault Diagnosis of Labeled Petri Nets

This paper presents a fault diagnosis approach for labeled Petri nets under a decentralized architecture consisting of multiple local sites and a coordinator. The key contribution is a communication protocol that defines the data exchange between each local site and the coordinator. A local site performs fault diagnosis by solving an integer linear programming (ILP) problem, which is formulated based on its local observation and the data received from the coordinator. Since the ILP problem inspects a smaller set of transition sequences consistent with a local observation, the proposed approach achieves more accurate diagnosis decisions. Specifically, some faults that could not be detected by the documented studies can be successfully isolated. An example is provided to demonstrate the advantages of the proposed approach.