Circular statistics vector for improving phase shift migration imaging of wrinkles in composites using ultrasonic array

Abstract

Multilayer composites are essential for aerospace, automotive industry and medical devices, but subtle wrinkles inside will significantly impair its mechanical performance. The phase shift migration (PSM) algorithm is well-suited for imaging defects in multilayer composites due to its layer-by-layer wavefield extrapolation technique. However, it performs poorly for ultrasonic detection of wrinkles in composites because weak wrinkle echo signals are mixed and superimposed with the interlayer reflections, mechanical noise and multiple reflection waves. Based on the motivation, an imaging method of circular statistics vector factor weighted PSM (CSVF-PSM) is proposed. Firstly, the instantaneous phase of the received signal is extracted using the Hilbert transform. The cosine and sine components of the phase are abstracted into circular statistical vector (CSV). Secondly, the CSV factor (CSVF) in the frequency domain is constructed to characterize the phase distribution consistency at any pixel, which is multiplied by the original frequency domain matrix of PSM in the initial wavefield. Finally, by wavefield extrapolation, the wavefield at each layer is corrected with CSVF to obtain more accurate wavefield at different depths. The experimental results indicate that CSVF-PSM significantly improves computation efficiency compared to the conventional time-domain total focusing method (TFM) algorithm, with an improvement of at least 8.67 times. Besides, compared to PSM and sign coherence factor weighted phase shift migration (SCF-PSM), the CSVF-PSM improves the imaging clarity and texture contrast for subtle wrinkles, with the error less than 10% for the wrinkle angle detection.