Research On Fault Diagnosis Based On Fuzzy Model And Multi-Agent Systems

In recent years, to meet the requirement of the corporation to the fault diagnosis of real-time, automatization, intelligence, network etc, fault diagnosis system also comes into the age of distributed intelligence. Thus the combination of Agent and fault diagnosis, as well as the forming of diagnostic agent, has been the main trend of intelligent diagnosis technique. It has the important signification to ensure the facility's reliable operation, improve the veracity and diagnostic speed, and reduce the maintenance cost.On the one hand, the application of multi-agent systems in industry can help to decrease the input of mass data, make use of the presence equipments synthetically, increase the flexibility and keep the intrinsic function of the distributed system, enhance the automatization of fault diagnosis system, and solve the conflictbetween generality and adaptability of diagnosis software as better as possible. These days, there are some productions about fault diagnosis based on MAS, but in fact they haven't been applied to practice really.On the other hand, there are already many methods of fault diagnosis, such as the way based on rules reasoning, or the artificial neural network, or cases, which all have the definite limit. Therefore, the way of fault diagnosis based on model has been proposed again, this way does not depend on the experience and case, just dose need .the running model of the object. But in order to get the forecast configuration, it is important to build the object model for fault diagnosis. In fact, it is very difficult or impossible. However, fuzzy model can make up this disadvantage fitly. In the past two years, some foreign learners have begun to pay attention to and research fault diagnosis based on fuzzy model.In this paper, firstly, fuzzy model and Takagi-Sugeno model are discussed. Afterward the fuzzy model of Three-Phase Induction Machines axletree is built, and its excellent approaching capability is validated through computer; Secondly, I present the abstractmodel of MAS in the fault diagnosis systems and the conformations of all agents in this model, and take example for fault diagnosis of electro-machine; Finally, the collaboration of agents has been studied further. I describe how to apply these techniques to the MAS model. Especially, in communication, the combination of blackboard model and message transmission has been put forward, thereinto the sub-area and numbering is brought forward in the design of blackboard model.