Font Size: a A A

The Research On Faults Diagnosis Method Based On Minimal Observations In Discrete Event Systems

Posted on:2017-02-01Degree:MasterType:Thesis
Country:ChinaCandidate:J N LiFull Text:PDF
GTID:2308330482492247Subject:Computer software and theory
Abstract/Summary:PDF Full Text Request
In recent years, the happen of the industrial accidents, has brought a great damage to people’s property and the produce in a great degree. Meanwhile, in order to improve the efficiency of industrial system fault diagnosis, fault diagnosis and prevention(FDP) becomes a very important research subject by experts and scholars. Diagnosis first appeared on the medical, through the observations of symptoms to determine disease that patient was suffering from. In the real life, most of the fields have a close relationship with diagnosis. Such as large-scale network systems, spacecraft systems engineering, vehicle systems, software testing and so on. The general methods for system fault diagnosis are the fault tree, the fault dictionary, expert systems, model-based diagnosis(MBD) etc., of which the model-based diagnosis method has widely covered a great kind of faults, a strong independence with device, and easy to maintain and update. Therefore, the model-based diagnosis method becomes research direction for many scholars.The method of Model-based diagnosis is mainly to make a actual system modeled as a logical system, and then compare the observed behaviors of the actual system with the expected behaviors of logical system, according to the differences between the two kinds of behaviors to judge and determine system’s failures. A classic method of Model-based diagnosis is diagnoser, through building the global diagnose for the system, according to global diagnoser to determine whether there is composed of loop with fuzzy system states, then to determine faults because the diagnoser is not suitable for the large complex systems, scholars has put forward some improvement methods, such as local diagnosers and etc. Twin plant was another method to judge faults, through the synchronous operation of preliminary diagnoser, which is a method for fault diagnosis method with a polynomial time, effectively makes up the deficiency of diagnoser in complexity of time.In order to improve the efficiency of fault diagnosis, by reducing system’s scale, the global system model is divided into many local models. Then get every fault diagnosis for each local model respectively, and combine the local diagnosis together to get the global result of fault diagnosis in system. According to the different ways of divisions in system, people can divided it into distributed fault diagnosis and the decentralized fault diagnosis. Both of them are to divide the whole system into several small local systems. The difference is between the distributed fault diagnosis and the decentralized fault diagnosis is whether there is communication events between the local small systems. There are some communication events between local systems in a distributed fault diagnosis. In order to further improve the efficiency of fault diagnosis in discrete event systems(DESs), the scholars put forward the hierarchical fault diagnosis and incremental fault diagnosis. Hierarchical diagnosis method let the system model is divided into some different levels, then compare the fault in each levels and get different fault result for different levels. Finally, get the whole diagnosis by combining the different levels result. The incremental diagnosis method is to consider the relationships of the local diagnosis, let the output of fault diagnosis in last local system be the input of the post local system and so on, then gain the fault diagnosis of the whole system.The methods above of fault diagnosis in discrete-event system, are not standing by the angle of minimal observation to study the observation’s sequence that determining the occurrence of system fault. To make the minimization on observable events sequence judging the fault occurrence, the purpose is to achieve minimal event sequence observed to determine fault. This paper proposes a fault diagnosis method based on the observation that the failure of minimal observation sequence model, through building the minimal observation sequence of failure off-line according to system model, and then use it in online diagnosis. Through simulating system’s online diagnosis process, making a comparison between using the minimal observation sequence of faults and not, this paper have recorded the diagnosis times in two situation at the same conditions, to illustrate the effectiveness and advantage of the method based on the observation in fault diagnosis in discrete event systems.
Keywords/Search Tags:Discrete event systems, Model-based diagnosis, Minimal observation sequence model
PDF Full Text Request
Related items