Application Of Synchrosqueezing Wavelet Transform To Crack Detection In Welding AE Signal

AE source signals, generated by welding crack structure changes, has random,transient and mutations characteristics. And during the acoustic emission test, because of the interference of the noise generated by the friction caused by the contact among the testing machine, sample and clamp, the collected AE signal containing a large amount of noise components,which make the AE signals show weak characteristics under the condition of complex noise background.Synchrosqueezing wavelet transform(SST) will restructure the energy of time-scale plane acquired by wavelet transform, which will make it have a high-resolution time-frequency distribution. Simultaneously, SST is not limited by wavelet function selection and has better robustness to noise. Therefore, SST was proposed to analyze the welding crack acoustic emission signals and extract the feature of welding crack acoustic emission signals under the complex noise background, achieving effective detection the welding cracks, which has important significance of academic and application.This paper based on SST theory, applied this algorithm to analyzing a simulated signal for verify characteristic, and analyzed the algorithm, which has better robustness to noise and has adaptability to wavelet function selection. On this basis, a test of welding process crack acoustic emission and a AE test of weldment tensile were designed. Then, the acoustic emission signals of welding crack heating and cooling process after welding were collected, and the frequency characteristics of crack acoustic emission signal were also analyzed, Meanwhile, using SST to extract crack acoustic emission signal feature to evaluation crack detection. The main contents are as follows:(1) Based on SST theory, the performance of SST and continuous wavelet transform(CWT) with a simulation signal were compared. Simultaneously, the characteristics of robustness to white Gaussian noise interference and its adaptability to wavelet function were analyzed. And this characteristics was applied to the wavelet analysis, improving the lack of time-frequency resolution.(2) A crack AE test of welding process was designed, waveform and amplitude-frequency characteristics analysis were used to the acoustic emission signals of welding process heating and cooling process after welding, which collected in test. Meanwhile, the AE signals of welding heating process were effectively todecomposition and reconstruction by SST and can be effective for noise reduction.The SST time-frequency analysis was applied to extract time-frequency characteristics of AE signals of welding process and the AE signals time-frequency distribution of different conditions was analyzed. then the AE signal frequency distribution and energy distribution characteristic were obtained.(3) A AE test of weldment tensile was designed, Combined with the the relationship of time-load and acoustic emission signals collected in stretching process and analysis the characteristics of waveform and frequency distribution of signals in crack formation and expansion. The SST time-frequency analysis was applied to analysis crack acoustic emission signal time-frequency characteristics, which can be obtained the signal characteristics of different crack formation. On this basis, The method was applied to the actual welding process of collecting acoustic emission signals for crack detection, through divided the large sample data was collected,short-divided data can better reflect the different dynamics of the crack state, and using SST time-frequency analysis to the short-divided data,which analysis the the dynamic information contains in acoustic emission signals and analysis the distribution of the frequency and energy.