Research On Health Automatic Detection System Of Locomotive Wheel-set

Wheels is one of key components in high-speed trains. The health status of train wheels is critical to the train operating safety. Explore and find a method to detect the wheels rapidly has become widespread expectation within the industry. The railway administration would detect the train wheel after the trains have travel a certain mileage. Currently, artificial flaw detector still accounts a large proportion in the train wheel NDT. However, this way is so complex and low accuracy. This paper presents an automatic diagnostic system for high-speed train wheels based on ultrasonic testing. And it largely improve the efficiency and accuracy of the train wheels NDT.The system solutions have been designed according to the target of pre-research based on Ultrasonic testing. It uses a scanning technology combination multi-channel and multi-probe. Constructed a dual-core co-processing hardware with FPGA and PowerPC. And the FPGA was used for high-speed data acquisition, PowerPC was used for digital filtering and data compression. The host PC is used to complete data analysis, the results of evaluation, defect location and display.The system software that was based on LabVIEW has been programed in this paper. Because of the LabVIEW graphical programming language, a friendly HMI were designed, it includes real-time graphical display of ultrasonic testing data, positioning of the faults in wheels, real-time display, database management and so on. Optimization for the software system has been presented in this paper such as multi-threading technology based on LabVIEW and reduce image refresh rate. And these methods improve the efficiency of program. It completed a high-speed data transfer between the host PC and collection system, fast processing data collection and the real-time display.This paper researched a method to identify the internal faults in the train wheels. This method decompose the ultrasonic testing signals by wavelet packet and extract the feature vectors, automatic identification of the type of train wheel internal faults by using the improved BP neural network. The simulation results show the feasibility of the method.The hardware and software of the system have been designed in this paper. And the circuit of hardware and software were debugged. The experiment achieved the desired result, and the result shows that system solutions meet the requirements.