A novel two-level approach to defect detection in braided CFRP using Air-coupled ultrasonic testing

Chenchen Zhang, Yinghong Zhang, Dianzi Liu, Xiaojia Zhang, Xianwei Wu, Zhenghua Qian, Xiangyu Li

Research output: Contribution to journalArticlepeer-review

Abstract

Air-coupled ultrasonic testing and C-scan technique has been increasingly applied to the braided CFRP structures owing to its non-destruction, non-contact and high visualization characteristics. Due to the noise, structural vibration, and airflow in the process of detection, the accuracy of defect identification is easily deteriorated. To address this issue and further determine the relationship between the ultrasonic acoustical pressure attenuation and structural parameters, a novel two-level identification method based on the modified two-dimensional variational mode decomposition (2D-VMD) has been proposed. In the first level, C-scan images have been sparsely decomposed into ensembles of modes by 2D-VMD method. Then, the modes have been screened by mutual information method to realize the reconstruction of new image in the second level. Experimental results have shown that the proposed method has the good ability to identify defects with a minimum detectable diameter of 1~2mm. It has been noted that the ultrasonic acoustical pressure attenuation has become remarkably higher in the twill weave CFRP than the plain weave CFRP and the ratio of pressure attenuation between two weave types of CFRP has decreased with the defect depth increase. Meanwhile, shadows around defects in C-scan images have been suppressed to a great extent. It has been demonstrated that the capability of denoising has enabled the developed method with the accurate detection in terms of the shape, size, depth and weave type. With these advantages, the proposed method has provided valuable insights into the development of an effective method for defect detection of braided CFRP structures.
Original languageEnglish
Article number106884
JournalUltrasonics
Volume128
Early online date8 Nov 2022
DOIs
Publication statusE-pub ahead of print - 8 Nov 2022

Cite this