Study on Noise Reduction Using a Wavelet Packet for Ultrasonic Flaw Detection Signal of a Small Diameter Steel Pipe with a Thick Wall

Abstract

Ultrasonic detection technology is an important tool that is used to ensure the safe operation of equipment. However, when carrying out ultrasonic testing, especially in the ultrasonic detection of small-diameter steel pipes with thick walls, a mass of structural noise causes the useful information to get lost. This is problematic because the quality of ultrasonic echo signals is the foundation for the pipeline flaw-identification and feature extraction. At present, the wavelet packet thresholding method is the most widely used in signal de-noising processes. In this paper, a de-noising method of a wavelet packet based on the threshold is proposed for the limitation of both soft and hard thresholds. The ultrasonic wave pipeline detection signal that is often contaminated by noise is de-noised by this method. Moreover, the contrastive analysis is performed among the wavelet packet transform based on an adjusted threshold, the original ultrasonic echo signal, and the default threshold. Experiment results show that the noise reduction effect of the wavelet packet based on the adjusted threshold is the most reasonable one, and the signal-to-noise ratio (SNR) of the signals is also improved.