Title: Research of probability characteristics in defect detection of composite materials using wavelet transform

Authors: V. Gerasimov, V. Khandetsky, S. Gnoevoy

Addresses: Dnipropetrovsk National University, Radio-physics Department, 9 Nauchnaya St, Dnipropetrovsk 49000, Ukraine. ' Dnipropetrovsk National University, Radio-physics Department, 9 Nauchnaya St, Dnipropetrovsk 49000, Ukraine. ' Dnipropetrovsk National University, Radio-physics Department, 9 Nauchnaya St, Dnipropetrovsk 49000, Ukraine

Abstract: Parameters of identification of surface crack signals K_h and K_w are proposed based on features of the Fourier spectrum and the wavelet spectrogram of modulation impulses defined using an eddy current sensor. Probability performances of defect detection using both spectral and wavelet methods are defined depending on the probability of a false alarm. Simulation data revealed that both methods have a good ability to identify short and shallow surface cracks. The wavelet method has the best probability of defect detection compared with the spectral method in case intensive noise. The spectral method is faster and preferable at small noise levels.

Keywords: defect detection; Fourier transformation; modulation impulse; surface cracks; wavelet transforms; composite materials; eddy current sensors; simulation; non-destructive testing; NDT.

DOI: 10.1504/IJMPT.2006.011271

International Journal of Materials and Product Technology, 2006 Vol.27 No.3/4, pp.210 - 220

Published online: 10 Nov 2006 *

Full-text access for editors Full-text access for subscribers Purchase this article Comment on this article