Study On High-speed Train Speed Information Fusion Technology Based On Weighted Fusion Algorithm

Train speed information is the basis of train control system on command generating anddecision-making, the real-time, accurate train speed information is a prerequisite to ensure thesafety of train operation. In order to improve the accuracy of train-borne control device, themultiple speed information received by speed sensor should be used according to a reasonablesolution to get the speed information with high accuracy.The train-borne device of Chinese train control system at level3(CTCS-3) is studied tosolve the problem of train speed information acquisition, storage and fusion. The workingprinciple and operating characteristics of speed sensor are studied. A method, which is abouthow to determine the running status of wheel and how to adjust the fusion mode of speedinformation by analyzing the speed sensor information, is given in this dissertation.To build a speed information acquisition and fusion simulation platform, the pulsegenerator is used to generate pulse signal as the simulation of speed sensor. Uses PCL-836board for data acquisition to collect pulse signal into configuration software, converts thepulse signal to speed information by writing command language. Design control proceduresin the configuration software for real-time monitoring of speed information acquisition. Thespeed information will be transferred from the OPC server which is built in configurationsoftware to the OPC client which is built in Matlab, and these information can be used in datafusion simulation. A contrast test shows that the collection frequency and the acquisitionaccuracy of this platform has met practical standards.On the basis of analysis of GPS/wheel speed sensor/doppler radar, a fusion method basedon distance threshold detection and idling, coasting detection is given in this dissertation.Remove the observation error of wheel speed sensor by determine the running status of wheelthen calculate the weighting factor by determine the area of the train, and process themulti-speed information with weighted fusion method. Simulation results show that thismethod can promptly and effectively eliminate single sensor observation errors, the fusioninformation is more accurate.