Distributed contextual diagnosis for very large systems

The occurrence of a fault usually influences immediately a (small) part of a large system, while the behavior in the rest of the system deteriorates only after some delay. Exploiting this, we propose in this paper a distributed diagnosis algorithm using a contextual analysis of Petri Nets. Using a backward reasoning technique we derive the minimal context (what must have happened) that explains the observed events. The method can tackle Petri...