Research On Fault Diagnosis Theory Based On DEDS

The scale of production systems becomes larger and the structure becomes more complicated with the development of science and technology. And many "manufacture systems" appears. Under this condition, faults occur more continually. Fault diagnosis becomes more important with the improving request of system reliability. But in many systems especially complicated systems a great deal of fault with the character of asynchronies, discrete and event-driven exist. These faults cannot be effectively diagnosed by traditional diagnosis method. The notion of discrete event dynamic systems (DEDS) is more adaptive because of the complexity of systems and event-driven character of fault. Having been developing for a relative long time, theories of DEDS are put forward. In mis way, fault diagnosis methods based on DEDS come into being. A large number of results have been get although these methods are still young and many problems are still in the scale. Furthermore, these methods have been widely applied in many fields, such as networks of computer and communication, power system, 1C manufacture, flexible manufacture, intermittent producing process, working units of multi-robot, intelligent attemperment of freeway, and so on.Methods based on DEDS belong to quantitative method. Researches on DEDS use many methods, which are based on models of different levels. Accordingly, there are numerous fault diagnosis methods based on DEDS.This dissertation concentrates on diagnoser theory initialized by Sampath, and discusses in relatively detail the features, properties, development and applications. The main contributions of the dissertation are as follows:1. First of all. it gives an overview of the emergence and the development process of fault diagnosis theories based on DEDS. Many results of these theories are also introduced. And the advantages and research meaning of fault diagnosis based on DEDS are expatiated. Furthermore, we describe supervisory theory and diagnoser theory in detail because the two theories are the base of this dissertation. Some discuss about applications of fault diagnosis theory based on DEDS is also in the dissertation.2. On the base of those theories and methods introduced in the dissertation, some researches have been carried out and the results gotten are showed as follows:(1) Based on supervisory theory and decentralized supervisory methods, we present a theory of united-normality, which is used for decentralized supervisory aimed at global objective. This theory is helpful to solve some problems about the application of decentralized supervisory and researches of fault diagnosis theory based on DEDS.(2) Active diagnosis method can be used to diagnose systems that are not diagnosable, while the specialty of coordinated decentralized fault diagnosis method is diagnosis of complicated systems with the character of information decentralize. So we present a fault diagnosis method. Decentralized supervisory, active diagnosis and decentralized diagnosis are all integrated into this method. Having those theories and techniques, this method can effectively solve the diagnosis problem of those information-decentralized systems that are not diagnosable.(3) We present a modular approach for diagnosis in large distributed systems. This approach is based on the diagnoser theory. Because of modularity, the whole system need not be thought of when diagnosers are constructed or used and changes of components in the system have a lot smaller effect on existing diagnosers.3. At the end, we summarize the whole dissertation. And some problems in the scale about the realm of fault diagnosis based on DEDS are introduced. Finally, we discuss the researches and applications foreground of this realm.