Diagnosis Of Stochastic Discrete Event System Based On Dynamic Observations

Discrete event system is a kind of system composed of discrete state and event transition,which focuses on the serialization behavior and logicality of the system.Failure diagnosis has always been a hot research direction of discrete event system.As early as the1990 s,Lin proposed a state based on-line diagnosis and off-line diagnosis for discrete event systems.After Sampath put forward the formal definition of diagnosability,the research on failure diagnosis has attracted the attention of experts and scholars at home and abroad.From the classical discrete event system to fuzzy discrete event system,stochastic discrete event system,distributed discrete event system,timed discrete event system and other systems with different model types.The failure types of diagnosis are also extended from single failure to multiple failure types.In practical application,the discrete event system has been successfully applied in military defense,communication network,industrial manufacturing,automatic control,biomedical and other fields.In the previous framework of discrete event system,the observability of events is assumed to be fixed in the whole operation process of the system.Obviously,the assumption of static observability of this kind of event is idealized,which has great limitations in practical application,because for a practical application system,the sensor devices and agent communication protocols of different components may be different,which leads to some differences in the observability of events when the system runs in different states.With the development of computer,communication and sensor technology,modern systems tend to be large and complex.Dynamic observation also has important research significance and practical application value.Secondly,for the system failure,we can find out the failure through failure diagnosis,but the diagnosis takes a certain period of time,and the system continues to run during this period of time,it may execute some events forbidden by the system,which is extremely dangerous for the system that has failed.Therefore,for practical production applications,safe diagnosis is also a problem we need to pay attention to.In the framework of dynamic observation,this paper studies the diagnosis and safe diagnosis of single failure event and pattern failure event of stochastic discrete event system.In the aspect of failure diagnosis,a kind of dynamic observation is introduced to make the observability of stochastic system event change dynamically with the operation of the system.Secondly,the single failure diagnosability and pattern failure diagnosability of stochastic discrete event system based on dynamic observation are formalized respectively.Thirdly,the necessary and sufficient conditions for single failure diagnosability and pattern failure diagnosability are obtained by constructing the corresponding diagnosers,so as to realize the diagnosis of single failure event and pattern failure event in the stochastic discrete event system under dynamic observation.In the aspect of safe diagnosis,the diagnosis condition is higher than the failure diagnosis condition,so the system needs to diagnose the fault before the forbidden event occurs.Similarly,the formal definitions of single failure safe diagnosability and pattern failure safe diagnosability for stochastic discrete event systems based on dynamic observation are given.The sufficient and necessary conditions for the safe diagnosability of single failure and pattern failure are obtained by constructing the corresponding safe diagnoser,so as to realize the safe diagnosis of single failure and pattern failure events for stochastic discrete event system under dynamic observation.