Research On Diagnosability Of Discrete Event System

In the reality,the system often fails.No matter the scale of the system is large or small,a slight fault may cause incalculable consequences,which causing irreparable damage and disaster.Therefore,fault diagnosis has received considerable attention to guarantee the performance of a reliable system.After fault diagnosis proposed by the researchers,a lot of achievements on fault diagnosis have been presented.Many diagnosis approaches have been investigated.The most widely used one is model-based diagnosis(MBD).Based on the structure and behavior of the system,the final diagnostic results can be deduced through comparing the actual behavior and expected behavior.Discrete event system(DES)is a discrete-state,event-driven system,the states evolution of which depends entirely on the occurrence of asynchronous discrete events over time.In the real world,many systems can be presented by DES,such as queueing systems,communication systems,manufacturing systems and database systems.To deal with diagnosis problem of DESs precisely,stochastic discrete event system(SDES)was proposed by Lunze and Schroder.In recent years,using MBD to solve diagnosis problems have been received a lot of attention.In this paper,we mainly focused on verifying diagnosability of DES and investigating the diagnosis problems in SDES.1)Diagnosability is an important property in fault diagnosis.Diagnosability property refers to the ability to detect the occurrence of failure events on the basis of observations and using model-based inferencing.In an ideal situation,researchers usually assume that a system to be diagnosed is complete;however,this assumption is rather restrictive.An absolute complete model,including all states and behaviors of the system,does not exist.Most of the models only include part of the states and behaviors of the systems.Recently,researchers have applied rough set theory to solve fault diagnosis problems to different kinds of systems,such as power transformer system and diesel engines.MBD is a diagnosis method that can be applied to many kinds of sytems because systems can be translated into the same formalism.In order to diagnose more kinds of systems with rough set theory,an approach employing MBD with rough set theory is proposed.The information system and decision system are used to describe the incomplete model and the observations monitored by the sensors.Then,based on these observations,we propose an algorithm to obtain a repaired model.Furthermore,a necessary and sufficient condition for a system to be diagnosable is given according to the decision table.In ensuring the diagnosability of a system,we also propose an algorithm to minimize the observable events and reduce the cost of sensor selection.2)SDES is a more precise formulation of DES.SDESs extend DESs by probabilistic of the transitions.Considering that DES cannot distinguish highly probable and less probable strings or states,A-diagnosability and AA-diagnosability of SDES were proposed.Testing diagnosability is a path-finding problem.An effcient approach to solve a path-finding problem is to reduce them to propositional logic problems.This approach is similar to diagnosability testing,which is what we are pursuing in this paper.Therefore,we use logical expressions to present SDES and propose a logical diagnoser to test A-diagnosability and AA-diagnosability of the SDESs.Our algorithm can achieve diagnosability directly and avoid any synchronization operations.Experimental results demonstrate that our algorithm improves the accuracy and efficiency of verifying diagnosability of SDES.3)In large scale discrete event systems,the complexity of the fault diagnosis is very large,no matter from space complexity or from the time complexity.Therefore,discrete event systems with decentralized information have received a lot of attention.The diagnosability can be obtained through verifying diagnosability of each local model.According to the decentralized information,the global system can be partitioned into a set of local models.DESs with decentralized information can be classified into distributed DESs and decentralized DESs.In distributed DESs,the local models communicate with each other by the communication events between them;however,communication events do not exist between the local models in decentralized DESs,a coordinator is constructed to exchange the local diagnosis information.Inspiring by the DESs with decentralized information,SDESs with decentralized information are partitioned into decentralized SDESs and distributed SDESs.A-diagnosability is an important property in failure diagnosis of SDES.In this paper,we investigate A-diagnosability in distributed SDESs.We construct a synchronized stochastic diagnoser and propose a necessary and sufficient condition for a distributed SDES to be A-diagnosable.Some examples are described to illustrate our algorithms.