Fault Diagnosis Method Based On Fuzzy-Colored Petri Net

The system of fault diagnosis technology is an important method to reduce accident risk, improve the reliability of system and ensure normal operation of the system. However, with the scale and complexity of the system become more and more huge and complex, and the fault symptom information also has a lot of uncertainty factors. These problems are urgent to solve by fault diagnosis technology. With the technology proliferating, fault diagnosis technology gradually gets into a new stage, which above signals analysis technology, Analytic model, and artificial intelligence technology. Petri net has been with widely applied in the field of fault diagnosis with its strong modeling ability and analytical skills. Based on Petri net theory, this paper puts forward a new method for fault diagnosis.Aiming at the problems about complex model structure, vast scale and uncertain issue of fault symptom information when Petri Net is modeling about fault diagnosis troubles, we propose a fuzzy-colored Petri nets model. We combine the places and transitions which describe the same class, then the number of places and transitions will lessen. The complexity and scale of model also will reduce. We put credibility as one of the attributions of tokens, set threshold of transitions and functions of arcs as fuzzy information, and solve some uncertain problems about fault symptom information. Then we put forward a fault diagnosis method based on this model. This method can be used for a kind of fault diagnosis problem which reasoning according to fault symptom information, especially for some complex system fault. We put fault symptom information after classification as input, fault status as output. We also set three different trigger rules of transitions describe the evolution of the status marking which about analysis transition, diagnosis transition and concordant transition. Then we get the final result after hierarchical processing according to the steps of fault diagnosis. So this method fold and addition fuzzy factor, the structure of this model become simple, the scale become small and can solve uncertain problems at the same time when we just use Petri Net. We use the fault of railway signal equipment and power grid as the examples. Modeling after analysis their characteristics of signal, the result show that:our method is effective and right. Because the instances in our experiments belong to different fields, and the parameters in the process of fault diagnosis are different from each other, the model of this paper is a kind of nomal model which can solve problems according to fault symptom information reasoning. The experiment results show that this model has a strong semantic describing capacity, simple process of modeling, strong adaptability and has little affluence by the change of system structure. At the same time, the credibility attribution of tokens and the functions of arcs solved the problems about some uncertain information which above information loss and error.