The increasingly complex equipments and control systems put forward higher requirements to the fault diagnosis system. On the one hand, the torrent of surveillance information pours into the diagnosis system; on the other hand, with the more and more diversified types and data association, the models of the faults are getting more complex. These development trends pose a significant challenge to the fault diagnosis system on its performance and the capacity of problem solving. In many applications, such as fault diagnosis system for high speed train, knowledge-based fault diagnosis method is widely used. Typically, the expert system is the most popular method for the diagnosis because of its clear descriptive capacity. However, this type of method also suffers some universal problems, including the limited presentation power for fault models, especially for the faults which require complicated temporal constraints; the inefficiency for large scale data and the indifference to the initiative of the diagnosis. Therefor, fault diagnosis approach based on active rules (Event-condition-action rules) reasoning is proposed in this paper. The primary researching work includes:

(1) Based on the research of the basic principle and features of various fault diagnosis methods，the reasoning technology based on active rules reasoning is introduced into the fault diagnosis field for the disadvantages of diagnosis export system;

(2) By formalizing the fault, the ECA–based fault modeling method is presented, which improves the description ability of fault diagnosis system;

(3) On the basis of proposing primitive event model, to reduce the load of diagnosis, the primitive events generating algorithm is presented, which direct the phase of the date filtration, during which the data could be translated into primitive events and primitive event could be circularly triggered by the others;

(4) In view of the algorithm named RTIAE which guides the diagnosis, the strengthened function of conditional processing combines the algorithm with applications; at the same time, the importance of the action is introduced into the heuristic function for supporting the hierarchical faults;

(5) By combining the network and fault diagnosis technology, remote fault diagnosis is supported, which improved the service efficiency and security of the fault diagnosis;

(6) According to the above research, a general framework of fault diagnosis system based on active rules reasoning is designed and a fault diagnosis system for high speed train is developed to validate the feasibility and effectiveness of the framework.