Noise processing of flaw reconstruction by wavelet transform in ultrasonic guided SH waves

The ultrasonic guided wave technique is a potential and useful tool for nondestructive testing. The scattered or reflected wave from a flaw can give its qualitative or even quantitative information such as location, size and severity. However, in experiments and in situ tests, the noises always exist together with signals. In this paper, we propose a wavelet-transform based noise processing approach for reflected wave signals to the quantitative reconstruction of surface flaws on a plate using guided SH waves. We suggest two different denoising methods based on wavelet transform (WT) in time and wavenumber domains respectively. Numerical results show that wavenumber-domain WT operation gives a better denoising effect than direct time-domain WT denoising. Using the former, one can successfully perform the inverse reconstruction of flaw by reflected signals with signal noise ratio as high as -5 dB. This research can act as a theoretical reference for practical applications of ultrasonic guided SH waves in quantitative nondestructive evaluation.