Research On Model And Method Of Complicated Malfunctions Diagnosis Based On Learning Algorithm

Because of signal system of the railway being a complicated control system made up of many kinds of electromechanical equipment, the variety of control among the signal devices in the station makes the reasons that cause the faults obviously uncertain fuzzy, and randomness, the fault diagnosis of signal devices in the railway station needs to be finished by expert system. How to select a suitable model of complicated fault diagnosis and develop a realistic algorithm is a key problem of building the actual expert system based on the principle of malfunction of signal devices in the railway station.This paper studied the model and method of intelligent fault diagnosis based on decision tree algorithms by some actual work of the project Intelligent System of Fault Diagnosis Expert System of the Railway Signal Devices, which was led by the tutor.Firstly, the paper studied systemically the process, content and method of railway signals fault diagnosis according to the characters of the signal devices in the railway station. Secondly, it aimed at the shortage of the knowledge representation and acquisition methods,and discussed the knowledge representation and acquisition method based on decision tree, especially studied the model of building in application algorithm C4.5 to the fault diagnose expert system of the railway station signal device. It put forward the applied model of acquisition and representation of knowledge and the complicated fault diagnosis algorithm based on decision tree. The proposed method used the advantage that the decision tree possesses the knowledge representation and acquisition, and carried out the knowledge representation and knowledge acquisition simultaneously. Finally,the proposed method was applied to the research of the actual fault diagnosis expert system of the railway signal devices, which included the development and the case analysis based on decision tree, the building of knowledge of the expert system and the estimate of Algorithm C4.5. The result showed that not only it can implement the automatic acquisition and representation of knowledge,but also the acquired knowledge based on decision tree possessed in greatly high inference efficiency.