Wavelet Transform Based De-Noising of ToFD Signals of Austenitic Stainless Steel Welds

Non Destructive Evaluation is generally used for defect detection in welds. ToFD technique is one of the NDE methods, used in weld inspection to identify the weld defects. In ToFD testing, the quality of the signal is an important factor for identifying and classifying the defects. So, signal denoising is a key to successful application of ToFD testing. Many signal processing techniques are followed to improve the quality of the signal. Wavelet Transform is one of the effective signal processing techniques. This paper presents a denoising algorithm ,which is suitable for denoising the ToFD signal of an unknown weld defect.
In this work, five austenitic stainless steel weldments were fabricated with different types of defects. The ToFD experiment is conducted on these welds. According to the characteristics of the resultant ToFD signals, discrete wavelet transform via different thresholdings are employed. Symlet and coiflet are chosen as mother wavelets. Various combination of wavelets, decomposition levels and different thresholdings are applied to find out the optimum denoising method. Finally evaluation of wavelet based denoising is achieved by calculating the SNR.
Results show that the noises can be suppressed well and SNR is improved. Symlet 4 with the 5th decomposition level in association with the hard thresholding is found as the effective signal denoising algorithm for all the 5 different types of defected ToFD signals.