EUVIMEDEUVIMED entdecken
Open Access · CC BY 4.0

Efficient two-stage sequential machine learning model for progressive damage analysis in fiber-reinforced composites

Mohammadamin Alibakhshi, Parisa Khodabakhshi

Machine Learning: Engineering · 01.06.2026 · 2 (1) · 015007

DOI: 10.1088/3049-4761/ae565d ↗

Kurz gefasst

Fachartikel in der Zeitschrift Machine Learning: Engineering, online erschienen am 08.04.2026. Verfasst von Mohammadamin Alibakhshi und 1 weiteren Autor:innen. Die Arbeit nennt 54 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0).

AbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

This study presents a two-stage integrated machine learning surrogate model for microscale analysis of the nonlinear behavior of fiber-reinforced composites, with a particular focus on progressive damage analysis. The framework couples a recurrent neural network, used to capture the temporal evolution of damage, with a convolutional neural network that maps the evolving damage pattern to the corresponding damaged homogenized stiffness matrix. The two building blocks of the surrogate model are chosen to align with the underlying micromechanical physics. The surrogate model is trained and tested on micromechanics simulations generated by the high-fidelity generalized method of cells (HFGMCs) for two repeating unit cell (RUC) configurations that differ in their refinement level. Numerical results demonstrate that the surrogate model closely reproduces HFGMC predictions while achieving significant computational savings, with speedups that increase as the RUC becomes more refined, with an order of magnitude speedup observed for the more refined RUC. When embedded into a multiscale analysis framework, the surrogate model gives an accurate yet computationally efficient alternative to the high-fidelity micromechanics model. Moreover, it successfully predicts holistic macroscale data metrics on which neither of its two building blocks was trained.

Zitieren

APA

Alibakhshi, M., & Khodabakhshi, P. (2026). Efficient two-stage sequential machine learning model for progressive damage analysis in fiber-reinforced composites. Machine Learning: Engineering, 2(1), 015007. https://doi.org/10.1088/3049-4761/ae565d

Vancouver

Alibakhshi M, Khodabakhshi P. Efficient two-stage sequential machine learning model for progressive damage analysis in fiber-reinforced composites. Machine Learning: Engineering. 2026;2(1):015007. doi:10.1088/3049-4761/ae565d

BibTeX
@article{Alibakhshi2026,
  author = {Alibakhshi, Mohammadamin and Khodabakhshi, Parisa},
  title = {Efficient two-stage sequential machine learning model for progressive damage analysis in fiber-reinforced composites},
  journal = {Machine Learning: Engineering},
  year = {2026},
  volume = {2},
  number = {1},
  pages = {015007},
  doi = {10.1088/3049-4761/ae565d}
}
Kostenlos starten →