High-speed Train Security State Assessment Simulation Platform-Generation Of Simulation Data And Fault Identification

China’s railway has conducted six large-scale speed-up from1997to2007. reduce the conflicts of the transportation load and the transportation capacity, receiving a very significant social and economic benefits. Into the21st century, along with the running of high speed train, the development of high-speed railway entered a stage of vigorous development. However, the safety of railway is not optimistic. The train safety issues emerge frequently, because of ’air spring loss of gas’.’yaw damper fault’.’Lateral damper fault’ and so on. These faults not only lead to safety problems, but also bring substantial damage to property. The effective monitoring and evaluation of high-speed train security state is an important means to ensure the safety warning and health maintenance. Running gear is a major component of the rolling stock, but also the most critical part of the core in safe operation of train. Its faults will seriously affect the safe operation of the train. The sensors on bogie collect the vibration monitoring signal to accurately reflect the current state of train operation. But the limit of human and conditions in the field monitoring conditions or shaking table tests. It takes too many resources to collect a variety of data and the limited experimental means, it is very difficult to fully reflect the true state of train operation via the acquisition of motion parameters.Therefore, it is necessary to use simulation experiments to simulate the motion situations of the train. Through the Multi-body dynamics software SIMPACK to model a type of vehicle. Simulating the operation conditions under the different excitation of track spectrum. At the same time, building the co-simulation platform by SIMPACK and MATLAB.MATLAB/SIMULINK as external vehicle dynamic excitation output to produce real time external random excitation. SIMULINK transfer the external excitation to the SIMPACK through SIMAT.SIMPACK calculation the current acceleration and displacement of each component of the motion parameters. Then the motion parameters return to the MATLAB through SIMAT module to the next step feature extraction.Simulation data generated by co-simulation platform is considered the locomotive of random excitation. Simulation data can react more real of train’s operation conditions. To overcome the one-sidedness and limitations of the single feature in the classification performance. Using five kinds of wavelet entropy extract feature to four kinds of single condition signal. Designing SVM classifier, choosing a higher recognition rate of features to feature fusion. Through the category decisions resulting of a number of features to Merge decision.It is found after a decision judgment of features fusion, correcting the sample misjudgment of individual characteristics.After the decision fusion the recognition rate is higher than any one obtained in a separate recognition, reducing the probability of misjudgment to faults by Single feature. Therefore, this method can improve the recognition rate. It is an effective method of identification.