Ultrasonic flaw detection based on temporal and subband signals applied to neural network

Abstract

Ultrasonic Non-Destructive Evaluation (NDE) uses high frequency acoustic waves to evaluate materials, and often signal processing is required to detect echoes from defects in the presence of micro-structure scattering noise. Scattering noise is known as the clutter. The clutter interferes with the flaw signal and cannot be completely separated from it by using conventional signal processing methods such as subband filtering. In this paper, a Neural Network (NN) is used for ultrasonic flaw echo detection using two different methods to design and train the networks. The first method is based on pattern recognition in time domain. The second algorithm is a combination of the Split Spectrum Processing (SSP) and NN. Eight frequency components are extracted from the original signal to train the NN. Both algorithms have reliable accuracy for target echo detection. The feasibility of using machine learning algorithms for material characterization and signal analysis is also discussed.