Safety Diagnosis of Stochastic Discrete Event Systems Based on Dynamic Observations

Most existing studies of stochastic discrete event systems (SDESs) are based on the assumption that the observability of events is invariable. However, the observability of events in many practical systems is usually related to the states of systems. Therefore, this paper proposes a safety diagnosability method of SDESs under dynamic observations. First, a dynamic observation is introduced, in which each state has its independent observable events. Then, we formalize the notion of the safety diagnosticability of SDESs under dynamic observations. Finally, the sufficient and necessary conditions for the safety diagnosability of SDESs under dynamic observation are deduced based on a safety diagnoser automaton, achieving the safety diagnosis of SDESs under dynamic observations.