Parametric study on interply tracking in multilayer composites by analytic-signal technology.

The multilayer structure of fiber-reinforced polymers can be extracted by several ultrasonic techniques, each operating in a different frequency range. Generally, it is assumed that ultrasound at higher frequency provides better depth resolution, but at the expense of the dynamic depth range due to excessive attenuation. Hence, a lower frequency is usually necessary to fit the depth setting to the probing requirement. And lower frequency approaches coupled to deconvolution techniques have been developed to improve the depth resolution and signal-to-noise ratio. However, when the frequency becomes as low as the ply-resonance frequency, the conventional metrics such as instantaneous amplitude envelope is completely not valid for distinguishing resin-layer reflections. To add more valid metrics in this situation, recent studies proposed the use of ultrasound at the ply-resonance frequency and subsequent analytic-signal analysis. This study compares the performances of various ultrasonic approaches, in different frequency ranges, for extracting the multilayer structure of composites. The following techniques are studied: (i) 50 MHz ultrasound coupled to instantaneous amplitude envelope, (ii) 15 MHz ultrasound coupled to Wiener deconvolution (with spectral extrapolation), (iii) 5 MHz ultrasound coupled to analytic-signal analysis (with log-Gabor filter). The performances of the various techniques are first investigated and discussed on synthetic data representative for a 24-ply composite. The robustness of the techniques is tested for different signal-to-noise ratios. An experimental study is performed on a 24-ply carbon fiber reinforced polymer to further validate the performances of different techniques. The extracted multilayer structure is presented in the form of both B-scan and C-scan images and the thickness of each ply is estimated. The obtained results indicate the high performance of the 5 MHz frequency ultrasound coupled to analytic-signal analysis combined with a log-Gabor filter.
Copyright © 2020 Elsevier B.V. All rights reserved.