| Bogie air spring is the key component of EMU,which contains many metal rubber bonding components.If there is debonding between metal and rubber,it will directly affect the operation safety of EMU.In order to ensure the safe operation of EMUs,it is of great engineering significance and application value to scientifically and accurately detect the debonding phenomenon in air springs.In this paper,the debonding test of SYS510 E air spring of CRH3 EMU is studied.By consulting a large number of domestic and foreign literatures,the advantages and disadvantages of bonding state detection methods such as percussion method,infrared thermal imaging method,nuclear magnetic resonance imaging method and ultrasonic detection method are analyzed.It is determined that the ultrasonic detection method is simple to operate,environmentally friendly and clean,and has good theoretical and technical basis.The characteristics of different ultrasonic testing methods,such as pulse reflection method,ultrasonic transmission method and ultrasonic guided wave method,are analyzed.Aiming at the structure and maintenance process of the debonding parts of bogie air spring,the ultrasonic pulse reflection method is adopted to detect the bonding state.This paper completes the design of debonding detection system based on FPGA and LabVIEW.The system is divided into two parts: the hardware circuit of ultrasonic detection based on FPGA and the application software of debonding detection based on LabVIEW.The hardware circuit can realize the functions of ultrasonic excitation,ultrasonic echo data acquisition and data transmission for eight ultrasonic probes.The sampled data is transmitted to the industrial computer through the Ethernet interface by the FPGA.The application software on the industrial computer can realize the pretreatment,feature extraction and pattern recognition of the ultrasonic data.In order to facilitate laboratory tests,data analysis and calibration tests,metal and rubber blocks with artificial debonding defects were designed and manufactured using the same material as SYS510 E air spring.On this basis,the ultrasonic signal preprocessing method,feature extraction scheme and pattern recognition method are studied and analyzed.In this paper,the advantages and disadvantages of two methods,singular spectrum analysis and wavelet threshold denoising,are analyzed and compared.According to the noise characteristics in the actual detection process,two methods,singular spectrum analysis and wavelet threshold de-noising,are proposed to pre-process the ultrasonic echo signal.This method can effectively improve the signal-to-noise ratio of the ultrasonic echo signal and reduce the computational complexity.Secondly,the traditional high-order echo acoustic pressure ratio method is studied and analyzed.Because it is difficult to recognize the bonding state effectively by simply depending on the difference of the amplitude of the high-order echo,the EMD decomposition of the ultrasonic echo signal is further studied to find out the correlation coefficients,sample entropy,energy parameters and other new eigenvalues which can characterize the bonding state.Subsequently,GA-BP neural network is used to identify the identified eigenvalues.The results show that GA-BP neural network has higher recognition rate and faster calculation speed than BP neural network.The recognition accuracy of 80 groups of samples is as high as 98.75%,which is enough to meet the requirements of maintenance.Finally,the bonding interface between metal rubber test block and SYS510 E air spring base and auxiliary spring was tested by C-scan.The results show that the performance of the system can meet the expected technical requirements. |
PDF Full Text Request