| Destructive Testing and Non-destructive Testing (NDT) are two kinds of traditional detection methods for rail welding quality. The former is one method of sampling inspection. The results of sampling inspection could not represent the welding quality of all the joints, and a large amount of manpower and material resources are wasted. The latter are mainly composed by X-ray and ultrasonic flaw detection, and such NDT methods could not accurately detect flaw such as grey-patches, which have significant effect on the quality of rail welded joint. In order to make up for shortcomings of existing traditional detection methods, a new evaluation approach of rail flash welded joint quality is needed to be found. The flash acoustic signals in rail flash welding was produced by the blasting of contact liquid lintels which were formed in the welding process. And the protective effect of welded joints at high temperature was determined by the stability and intensity of blasting of the lintels. It can be seen that the flash acoustic signals in rail flash welding consequentially carrys important information about the quality of welded joint.In this paper, multi-channel signal synchronous acquisition experiment software and hardware system of rail flash welding was designed based on UN5-150Z mobile rail flash welding machine. Firstly, the designed signal adapter circuit board was used to realize the conditioning and electrical isolation between the original monitoring system on flash welding machine and acquisition experiment system. Secondly, the data acquisition software system which was designed based on virtual instrument development language LabVIEW realized the long time and continuous high speed acquisition of welding voltage, welding current, welding displayment and welding oil pressure by setting up the FIFO on data acquisition card PCI-1710L. Thirdly, the reliability and practicability of multi-channel synchronous acquisition experiment system in the paper was verified by replaying the collected data.The eight-layer sym4wavelet was used to decompose and reconstruct the two-channel low frequency signals of welding current and welding voltage which had been collected with flash acoustic signal synchronously. By comparing the wavelet denoising result with the traditional low-pass filter denoising result of welding current, it confiremed that the traditional filtering methods based on Fourier transform was not suitable for filtering the non-stationary signal, such as welding current, welding voltage,and flash acoustic signal in this paper. In order to avoid missing the useful information in flash acoustic signal, the four-layer db8wavelet packet was used to decomposed the flash acoustic signal which had been under the strong noise environment. High and low frequency noise with high power spectrum in the original flash acoustic signal were removed effectively by using "similarity comparison method". After Fourier transform of reconstructed flash acoustic signal, the frequency range (1050-3850Hz) of real flash acoustic signal was obtained.The programming and simulation of four kinds of time-frequency analysis method (Short-time Fourier transform, Smooth Pseudo Wigner-Ville distribution, Analytic Wavelet transform and Hilbert-Huang transform) were realized based on Lab VIEW. According to Simulation results, the advantages and disadvantages of four kinds of time-frequency analysis method were summarized. The amounts of the data from flash acoustic signal to be analyzed was so large (the length of time was more than130s, the sampling frequency was12KHZ) that the computer couldn’t accomplish the compute tasks when time-frequency analysis of Smooth Pseudo Wigner-Ville distribution was used. On the other hand, the time-frequency resolution couldn’t be guaranteed in high frequency components when Analytic Wavelet transform and Hilbert Huang transform were used to process the signal. Through a comprehensive comparison, the Short-time Fourier transform (STFT) whose algorithm was simple and easy to implement was chosen to do time-frequency analysis for the flash acoustic signal.Two characteristic quantities, namely energy spectrum density (ESD) and weighted average frequency (WAF) were extracted from the two-dimension color STFT spectrogram. Time-ESD tendency chart and time-WAF tendency chart were regarded as the evaluation means of the welded joint quality of rail flash welding, and the quality of collected26welded joints of U75V rail were judged in this way. The results showed that the characteristic quantities (ESD and WAF) extracted from flash acoustic signal could more correctly judge the grey-patches defects in the welded joints. |
PDF Full Text Request