Based On Neural Netowrk Expert System For Fault Diagnosis Of The Station Signal Control Circuit

As an important component of the railway signal system, railway station signal control system is the key to the safe operation of the train, the current fault diagnosis method for the station signal control circuit is still flawed, and is difficult to meet the requirements for rapid development of the railway, so solving the station signal control circuit fault diagnosis problem by studying a new way is the main contents of this article.In view of limitation of the traditional diagnosis technology,the tradition fault diagnosis expert system exists insufficient which cannot carry on self-study,auto-adapted for unknown fault and abrupt faults cannot diagnosis, cannot satisfy the station signal control circuit fault diagnosis for requirement of real-time. so in this paper, a fault diagnosis for the station signal control circuit is put forward, which combining expert system and neural network.This method includes the design of expert system and neural network structure design.The design of expert system includes knowledge base, inference engine and the design of explaining mechanism. First, the expert experience knowledge is represented through relevant information and on-site data extraction, use the production rules for knowledge representation. The expert system knowledge base is implemented using relevant software. Then the expert system's inference engine is designed and explaining mechanism enable the users to clearly understand the detailed process of the system reasoning by production rules. The design of neural networks includes the construction of network structure and the training and simulation of neural network. First constructed based on BP neural network fault diagnosis system network structure, and then parts of the expertise and fault data are extracted as training samples and test samples for the training and testing of neural network. Finally, the neural network fault diagnosis method is feasible in the system by MATLAB simulation.After the expert system and neural network design is completed, the specific examples verify the correctness of the design, through analysising the expert system and neural network fault diagnosis system diagnostic results, boths of them have advantages, but there also are insufficient, and has a highly complementary, so this article finally adopted based on hybrid neural network expert system fault diagnosis for the station signal control circuit for fault diagnosis.In the implementation of the system, the system design is completed using Visual C++, Microsoft Access database, and MATLAB software. Finally, the feasibility of fault diagnosis system which combining neural network and expert is validated in the fault diagnosis for the station signal control circuit by a fault diagnosis example. The results show that the system can shorten the diagnosis time to a great extent and improve the diagnostic efficiency of the system which using expert system assisted with neural network hybrid expert system for fault diagnosis.