Prediction of Structural Integrity Assessment Parameters from Ultrasound Images via Machine Learning
e-Journal of Nondestructive Testing · 2026-07 · 31 (7)
DOI: 10.58286/32905 ↗Kurz gefasst
Fachartikel in der Zeitschrift e-Journal of Nondestructive Testing, 2026-07. Verfasst von Junlei An und 3 weiteren Autor:innen.
Abstract (Originalsprache)
Structural health monitoring (SHM) is essential for the safety and durability of engineering structures, where accurate fracture parameter measurement is critical. This study presents a machine learning framework for predicting crack length, crack mouth opening displacement, and stress fields around crack tips directly from ultrasonic data acquired with a permanently installed, low-cost, low-profile array. The array’s lower signal-to-noise ratio than conventional systems was addressed by training convolutional neural networks and U-Net variants on total focusing method images paired with stress and strain fields from digital image correlation. This pairing supplied explicit crack-related information, while parameter tuning provided insight into model configurations and trade-offs. Experimental validation confirmed reliable prediction of fracture metrics despite noisy input. The results highlight the potential of combining affordable ultrasonic arrays with machine learning to deliver scalable, cost-effective, and efficient SHM solutions for infrastructure management.
Zitieren
APA
An, J., Liu, Q., Larrosa, N., & Zhang, J. (2026). Prediction of Structural Integrity Assessment Parameters from Ultrasound Images via Machine Learning. e-Journal of Nondestructive Testing, 31(7). https://doi.org/10.58286/32905
Vancouver
An J, Liu Q, Larrosa N, Zhang J. Prediction of Structural Integrity Assessment Parameters from Ultrasound Images via Machine Learning. e-Journal of Nondestructive Testing. 2026;31(7). doi:10.58286/32905
BibTeX
@article{An2026,
author = {An, Junlei and Liu, Qiang and Larrosa, Nicolas and Zhang, Jie},
title = {Prediction of Structural Integrity Assessment Parameters from Ultrasound Images via Machine Learning},
journal = {e-Journal of Nondestructive Testing},
year = {2026},
volume = {31},
number = {7},
doi = {10.58286/32905}
}
EUVIMED