| Title |
Non-destructive characterization and evaluation of adhesive joints using artificial intelligence tools |
| Authors |
Šeštokė, J ; Butkevičiūtė, E ; Smagulova, D ; Jasiūnienė, E |
| DOI |
10.58286/33197 |
| Full Text |
|
| Is Part of |
e-Journal of nondestructive testing: 14th European conference on non-destructive testing (ECNDT 2026), 15-19 June 2026, Verona, Italy.. Mayen : NDT.net. 2026, vol. 31, spec. iss. 7, p. 1-2.. ISSN 1435-4934 |
| Keywords [eng] |
adhesive joints ; artificial intelligence ; ultrasonic measurement |
| Abstract [eng] |
This investigation examined the potential of artificial intelligence tools for detecting defects in adhesively bonded samples. Adhesive bonds are widely used in various sectors such as aviation, modern transport including buses, trams, pleasure boats, etc. One of the main quality assurance procedures is the non-destructive evaluation of adhesive joints. The aim of the work is to improve non-destructive sample characterization and evaluation of adhesive joints using artificial intelligence tools. C-scan images of aluminum CFRP bonded joints with non-uniform defects ranging from 5 to 15 mm were used during ultrasonic measurement. The study was conducted in two stages, first, 4 different computational model architectures were applied, data were organized, processed and as a result, defects in bonded joints were identified. In the next stage, all data were randomly cropped in order to have more diverse data and trained using the most reliable model architecture to describe and classify different defects in aluminum CFRP plate bonded joints. During model training, defect detection was achieved with an accuracy of almost 95%, and after testing, the defect detection accuracy exceeded 97%. |
| Published |
Mayen : NDT.net |
| Type |
Conference paper |
| Language |
English |
| Publication date |
2026 |
| CC license |
|
