Application of Blind Source Separation in Ultrasonic NDE

Blind source separation (BSS) allows the recovery of unknown signals from observed signals mixed by an unknown propagation medium, and is a promising technique for signal processing and data analysis. In ultrasonic nondestructive evaluation (NDE), BSS can serve at least three purposes: defect classification, system modeling and noise reduction. This paper discusses the application of BSS in ultrasonic NDE, and explores the implementation of independent component analysis (ICA) in separation of ultrasonic signals. Simulation results were given, which showed that different source signals can be separated using ICA, and the conclusion is that BSS can play a very important role in ultrasonic NDE.