The Research And Realization Of Fault Remote Diagnosis Method Of Cab Signal System Based On Neural Network And B/S Mode

The cab signal system, which is the important part of train operation control system, includes on-board equipment and track circuits on ground. If the cab signal is not in normal state, the train could not be safely operated.As the improvement of train speed, the fault diagnosis of cab signal has higher requirements. The traditional fault diagnosis method of cab signal system can not meet the requirements of railway development. On the one hand, the detect methods are obsolete, the field technicians are uneven in quality and the betimes of detection are not satisfied. On the other hand, the fault factors, especially the faults caused by abnormal signal of track circuits have the characteristics of diversity, compositionality, contingency and randomness. These factors bring many problems to fault location and diagnosis. Therefore, improve the traditional fault diagnosis methods, adopt advanced fault diagnosis technologies and develop a fault diagnosis system have an important significance.The paper is based on wave data from railway field, takes frequency-shift track circuit as research object, summarizes the typical cab signal fault, takes neural network as algorithm and develops a cab signal automatic fault diagnosis system based on B/S mode. The main structure is as follows: firstly, based on statistics and analysis of the fault cases and track circuit signals from railway administrations, summarize the typical faults. Secondly, in algorithm design aspects, construct eigenvectors according to frequency spectrum characteristics of track circuit signals. Realize exact identification of fault type by training BP neural network. Thirdly, in system design and realization aspects, the paper proposes requirement analysis according to design goal. The system is divided by three subsystems such as user management subsystem, document management subsystem and faults diagnosis subsystem. The paper provides the detailed designs of system modules and database. Based on comparison of structure and implementation technology, the paper uses ASP.NET technology, Visual Studio 2005 tool, C# language, MATLAB neural network toolbox and SQL Server 2000 database to develop the system software. Finally, based on the data and cases from railway field, the paper simulates and verifies the network aiming at amplitude asymmetry, carrier frequency migration, low frequency migration and spectrum disturbance and their combination cases of frequency-shift track circuit mode. The test results show that the system combine the advantages of neural network and B/S mode, has the characteristics of high recognition rate, good generality, convenient and fast and convenient for managing, and could satisfy the requirement of the current cab signal system operation.