Intelligent physics-informed framework for crack detection and propagation modeling in rock engineering
Canadian Geotechnical Journal · 01.01.2026 · 63 · 1-22
DOI: 10.1139/cgj-2026-0178 ↗Kurz gefasst
Fachartikel in der Zeitschrift Canadian Geotechnical Journal, erschienen am 01.01.2026. Verfasst von Qipeng Cai und 3 weiteren Autor:innen. Die Arbeit nennt 56 Literaturangaben. Gefördert von: National Natural Science Foundation of China, Huaqiao University Research Fund.
Abstract (Originalsprache)
Accurate prediction of crack propagation in rock masses is necessary for rock engineering applications; yet predicting complex fracture evolution remains a significant challenge. Existing studies typically focus on simplified fracture geometries, struggling to simulate multi-crack branching while neglecting principles of fracture mechanics. This study proposes an intelligent physics-informed framework, which incorporates the deep Ritz method with phase-field fracture mechanics, enabling forward forecasting of crack evolution and inverse estimation of material properties. The framework leverages an encoder–decoder U-Net with Monte Carlo dropout for quantifying prediction uncertainty, offering confidence intervals for the predicted fracture patterns. The framework is comprehensively validated using laboratory experiments and field-scale LiDAR data from the Xiamen project. The results reveal that the proposed method achieved the highest predictive accuracy, with an R 2 of 0.946 and the lowest root mean square error of 0.065 in predicting rock fracture propagation, indicating strong performance compared to existing deep learning models. The proposed method provides a foundation for future applications to reduce the manual efforts in fracture analysis while improving the efficiency of crack propagation prediction in underground projects.
Förderung
- National Natural Science Foundation of China
- Huaqiao University Research Fund
Zitieren
APA
Cai, Q., Elbaz, K., Huang, C., & Wei, Y. (2026). Intelligent physics-informed framework for crack detection and propagation modeling in rock engineering. Canadian Geotechnical Journal, 63, 1-22. https://doi.org/10.1139/cgj-2026-0178
Vancouver
Cai Q, Elbaz K, Huang C, Wei Y. Intelligent physics-informed framework for crack detection and propagation modeling in rock engineering. Canadian Geotechnical Journal. 2026;63:1-22. doi:10.1139/cgj-2026-0178
BibTeX
@article{Cai2026,
author = {Cai, Qipeng and Elbaz, Khalid and Huang, Chengyue and Wei, YingJie},
title = {Intelligent physics-informed framework for crack detection and propagation modeling in rock engineering},
journal = {Canadian Geotechnical Journal},
year = {2026},
volume = {63},
pages = {1--22},
doi = {10.1139/cgj-2026-0178}
}
EUVIMED