Research On Simplified Diagnoser And Safe Diagnosability

With the science and technology advancing,the society is developing simultaneously.Large-scale complex systems,such as the Internet,power supply systems,vehicle systems and so on,are more closely related to people’s daily life.These systems are equiped with comprehensive functions.Meanwhile,they get more investment for system research and development.These systems are used more frequently.And the system scale is becoming larger and larger.The above phenomena are the cause of the increasing complexity and the coupling degree of the systems.These results make the occurance of faults increasing.Fault is the deviation of the behavior between the actual system and its counterpart,the expected system.No matter how advanced the design is,how careful to improve the hardware and software quality,or how systematically to train the system operators,faults still can not be corrected or eliminated.In fact,considering the complex interaction among subsystems,components and processes of the system,it can even be considered that the occurrence of system faults is inevitable.To solve the problem,fault diagnosis is developed.Model-based diagnosis(MBD)is one of the fault diagnosis methods.MBD has following properties: Firstly,it neither depends on specific expert experience,nor on specific equipment;Secondly,it’s reusable,in other words,it can be transplanted flexibly.As a result,it’s easy to maintain;Lastly,it can diagnose more comprehensive faults.MBD is a hot orientation of research in the field of artificial intelligence.MBD uses finite-state machine(FSM)or other methods to model the actual system as corresponding logical system.Then,the system get the actual behavior according to the data observed by sensors.The actual behavior is compared with the expected behavior to find whether differences occur.If there exists differences,the system determines which kind of faults occurs to cause the differences.The earliest and most basic method of MBD is to construct a global diagnoser for discrete event systems.Whether the diagnoser is diagnosable or not is determined by whether there exists a fault uncertain state loop in the diagnoser.In real life,if a fault event occurs in the system and a sequence of forbidden events corresponding to the fault occurs after the fault.This fault may endanger people’s life and property safety.In order to avoid the occurrence of this kind of events,people hope to get warning of the occurrence of security events in advancce.As a result,safe diagnosability of discrete event systems was investigated in the literature recently by Stephane Lafortune and others.Firstly,in order to improve the safe diagnosis method of DESs,this paper introduces the construction method and related concepts of DESs simplified diagnoser.And then the paper gives an algorithm to obtain the simplified diagnoser to reduce the complexity of constructing the safe diagnoser.Then,the safe diagnoser of DESs is studied based on the simplified diagnose.In this part,the concept and model of safe diagnoser are given.Then the algorithm of constructing the simplified DESs safe diagnoser is given.At the same time,properties,definitions and related proofs of the simplified safe diagnoser are given.The algorithm is tested and compared with the original method.Finally,this paper carry the safe diagnosability of DESs over to distributed DESs.The paper gives the related properties,definitions and proofs.Then,an example is given.It shows that it’s feasible and efficient to use simplified distributed DESs to judge safe diagnosability.