Research On Fault Diagnosis For On-board Equipment Of Train Control System Based On Bayesian Network

As one of the core equipment of the train control system, on-board equipment (VOBE) is the key issue to ensure the safety and efficiency of the train operation. However, VOBE sometimes fails, and the method of fault diagnosis and maintenance for VOBE still relies on expert experience in our country. This method can not meet the demands of railway transportation of China, because it is of low degree of automation and the accuracy rate completely depends on expert experience. So it is important to find an efficient and accurate fault diagnosis method for VOBE.Up to now, there has been a lot of fault record information and some diagnosis experience about VOBE. As a complex system, VOBE is full of uncertainty in fault diagnosis. The present fault diagnosis method for VOBE still has some disadvantages in dealing with fault record information, expert experience and uncertainty. So this paper aims to study the fault diagnosis method for VOBE. Fault characteristics of VOBE are analyzed deeply. Bayesian network, which is one of the best ways to solve the problem of uncertainty, is the core algorithm. Combined with the fault record information and expert experience, the fault diagnosis model is established. By rough set theory and Bayesian network node division, the model is optimized. The validity and accuracy of the method are proved by using the operation information.This thesis’s main works are listed as follows:(1) Research on fault diagnosis method for VOBE. Combined with the research status at home and abroad, the existing fault diagnosis methods for VOBE and their shortcomings are analyzed. Through analyzing the characteristics of fault diagnosis for VOBE, fault diagnosis method for VOBE based on Bayesian network is presented. The establishment method of the model is studied deeply. Through analysis of expert experience and the fault record data, the knowledge base of VOBE fault diagnosis is obtained. On this basis, the fault diagnosis model is established by using the Bayesian networks. By analyzing the application of the model in fault diagnosis for VOBE, the effectiveness of the proposed method is proved.(2) Research on Optimization of diagnosis model. Aimed at the problem that there is a lot of redundant information in the fault record of VOBE, a new method of fault diagnosis model reduction based on rough set theory is proposed to compress the fault information. The removal of redundant information can effectively simplify the diagnosis model and improve the efficiency of reasoning. Aimed at the limitation of traditional Bayesian network node division in fault diagnosis, the improved Bayesian network node division is proposed. And the fault diagnosis model for VOBE based on the improved Bayesian network node is established. Through the analysis of the flexibility and accuracy of the fault diagnosis model, the validity of the method is proved.(3)Design and implementation of fault diagnosis system.Using C# and Matlab, the fault diagnosis system for VOBE is implemented.