Weld Guided Wave Defect Identification Based On Digital Signal Processing

Welding structure is widely used in national defense,aerospace,nuclear industry,transportation and other important fields,and has high requirements for the welding quality.This paper focuses on the problems related to ultrasonic guided wave defect detection of aluminum alloy welds,and focuses on the relevant processing methods of ultrasonic guided wave detection signals,including noise signal suppression and defect signal recognition.Firstly,the components,attenuation and non-stationary characteristics of ultrasonic guided wave detection signals are analyzed.The results of time domain,frequency domain and time frequency domain analysis of typical defect echo signal show that there are obvious differences in ultrasonic guided wave detection signal characteristics of different defects.Next,the signal processing methods are studied: least square method,fast Fourier transform,Hilbert-yellow transform,mean filtering,correlation.The treatment effect is good.The noise suppression of ultrasonic guided wave detection signal of welding seam is studied,the wavelet decomposition shows that the noise component mainly exists in the high frequency part of the signal.The optimal parameters of wavelet threshold denoising are determined as: dbn wavelet base,vanishing moment order 6,decomposition layer 3.Aiming at the shortcomings of traditional wavelet method in noise suppression and signal recovery,based on correlation analysis,an improved wavelet denoising method based on signal correlation is proposed.The results show that the proposed method effectively compensates for the signal distortion caused by threshold shrinkage.On this basis,an improved wavelet packet denoising method based on signal correlation is proposed.Simulation results show that compared with the traditional wavelet and wavelet packet denoising method,the proposed method has better noise suppression and defect signal recovery effect.The sample of aluminum alloy welding seam with artificial defect is made and the original ultrasonic signal is noise suppressed,experimental results show that the improved wavelet packet noise suppression method based on signal correlation has better denoising effect.Finally,the defect recognition and localization method in ultrasonic guided wave signal is studied.A defect recognition method combining time and frequency domain analysis is proposed,which makes up for the shortcomings of time and frequency methods and can determine the existence of defects more effectively.On this basis,hilbert-yellow transformation(HHT)technology is adopted to extract the characteristic information of ultrasonic guided wave signal with and without defects,and the damage status of the specimen is evaluated through analysis and comparative study.The concept of characteristic index is proposed to define the standard value to judge whether there are defects in the specimen.The experimental results show that the defect size increases monotonously with the characteristic index,that is,the larger the characteristic index is,the more serious the defect damage is.A defect location method based on ultrasonic guided wave signal envelope is proposed.The results show that the proposed method is more accurate than the direct measurement method,although the experiment and calculation process are relatively complex.By using the signal envelope method and the corresponding signal processing method,the measurement error of the position of the length direction of the defect weld can be effectively reduced,and more accurate defect location can be achieved.