Study On Time-frequency Analysis Method Of Laser Ultrasonic Defect Signal

Because of its good directivity and coherence,laser ultrasound has been widely applied in the field of defect detection.This paper focuses on time-frequency analysis of laser ultrasonic flaw signals,and extracts feature quantities to achieve quantitative characterization of defect depth.When EMD is used to denoise the laser ultrasonic signal,it is found that the serious modal aliasing occurs when the reflected wave signal is processed,which leads to the distortion of the signal after the denoising and the cause of the phenomenon of modal aliasing.By consulting the literature and combining the special point of the laser ultrasonic signal,the EMD denoising is improved.Combining with the wavelet denoising,the improved EMD denoising process can preserve the characteristics and details of the original signal better,and overcome the distortion caused by the modal aliasing.Then,the time-frequency analysis of the denoised signal is processed.(1)The Fourier transform is used to process the transmission wave signal,and the peak value in the frequency domain is selected.With the increase of the depth of the defect,the peak value of the selected region is approximately linearly reduced,and the result of the simulation data is consistent.The expression about the above region is obtained by fitting and the recognition of the unknown signal is obtained.It shows that this method is feasible.(2)The wavelet packet is used to analyze the frequency band energy of the reflected echo and the transmitted wave signal,and the DB8 wavelet base is selected to decompose the 8 layers of the laser ultrasonic defect signal,and the relationship between the frequency band energy distribution and the defect depth is obtained.On this basis,the wavelet packet energy change rate index is proposed,when the defect depth is in the 0.6mm,In the following,the energy change rate of the wavelet packet in the S80 and S81 bands increases with the increase of the defect.When the defect depth is below 0.7mm,the index of the wavelet packet energy change rate of the reflected echo signal in the S83 band increases with the increase of the defect depth;and the above three frequency bands are obtained by passing through fitting.Expressions,and through the analysis of the same set of experimental signals,the results of the three fitting expressions are similar,indicating that this method is feasible.(3)In view of the relationship between the formation mechanism of the reflection echo and the depth of the defect,the reflection echo signal is extracted by EMD decomposition,and the intrinsic modal function is obtained.The key analysis of the IMF1 which has the maximum correlation with the original signal is selected.The results show that the corresponding IMF1 oscillation time is also increasing when the defect depth is increased.Then,the fitting relationship between the two is obtained;then,the Wigner-Ville analysis of the selected IMF1 with different depths is carried out to get the corresponding time frequency amplitude three-dimensional graph,which can be obtained by analysis.When the depth of the defect is increased,the time of peak appearance in the three-dimensional graph is also increasing,and the two are obtained between the two.The fitting relationship is verified by experimental data,and its error is within acceptable range.