An Object Frame Knowledge Representation Approach for Fault Diagnosis Expert System

Knowledge quality and usability is core to a fault diagnosis system. Frame knowledge technique has been used to represent knowledge in many information and expert systems. To simplify the complexity of knowledge representation in fault diagnosis expert system, a novel object-frame knowledge representation approach based on hierarchical model was proposed. In this approach, domain specific knowledge is expressed by combinations of object-frames. An Object-frame is composed by relevant state-object, test-object and rule-object or repair-object. Production rules are used to connect relevant objects' states. General features of object-frame and inference algorithm are introduced. Object-frame based knowledge items are stored in SQL database, inference engine performs the inference operation of knowledge using forward chaining strategy, implements reasoning, finds the cause of faults and gives repair suggestion driven by test data. Inference interpretation completes the task of explanation, which improved the clarity of reasoning. The advantage of this method is that we do not need knowledge representation language support. Experimental results show that the method proposed is effective, which improved the fault diagnosis and maintenance for a meteorological vehicle system.