On Probabilistic Diagnosis for Free-Choice Petri Nets

We describe an on-line probabilistic diagnosis algorithm for free choice Petri nets. The plant observation is given by a subset of transitions whose occurrence is reported while the rest of transitions including faults are unobservable silent. We propose two notions for the probabilistic diagnosis, the likelihood that a fault must have happened prior to the last observed event, and the likelihood that a fault may have happened given the available observations.