Wavelet transform-based denoising method for processing eddy current signals

Abstract

Wavelet transform (WT)-based denoising method is proposed for processing eddy current signals of thin-walled stainless steel fuel tubes with periodic wall thickness variations formed due to fluctuation in tube drawing process parameters. In this method, discrete wavelet transform with level-based threshold has been applied to selectively eliminate the noise due to periodic wall thickness variations towards meeting the quality assurance requirement of detection of holes larger than 0.3 mm diameter and linear defects deeper than 0.075 mm (20% wall thickness). The method has been applied to tubes having machined holes, longitudinal notches, and circumferential notches, and an overall improvement of 20 dB in signal-to-noise ratio has been observed. The method has been able to detect defects present anywhere in the wall thickness variation regions and also in tubes without any wall thickness variations.