Reliable Decentralized Failure Diagnosis of Discrete Event Systems
Shuhei Nakata, Shigemasa Takai · SICE Journal of Control Measurement and System Integration · 2013
In most existing works on decentralized diagnosis of discrete event systems, it is implicitly assumed that diagnosis decisions of all local diagnosers are available to detect the failure. However, it may be possible that some local diagnosis decisions are not available due to some causes. Letting n be the number of local diagnosers, the notion of (n,k)-reliable codiagnosability guarantees that any occurrence of the failure can be detected by using arbitrary more than or equal to k local diagnosis decisions within a uniformly bounded number of steps. In other words, even if at most n-k local diagnosis decisions are not available, the failure can be detected by using the remaining diagnosis decisions. In this paper, a method for verifying (n,k)-reliable codiagnosability for any k is presented. Then, the delay bound within which any occurrence of the failure can be detected by using arbitrary more than or equal to k local diagnosis decisions is computed.