Generalizable super-resolution turbulence reconstruction from minimal training data
Journal of Fluid Mechanics · 10.12.2025 · 1024
DOI: 10.1017/jfm.2025.10913 ↗Kurz gefasst
Fachartikel in der Zeitschrift Journal of Fluid Mechanics, online erschienen am 02.12.2025. Verfasst von Haokai Wu und 6 weiteren Autor:innen. Die Arbeit wurde bisher 13-mal zitiert (Crossref) und nennt 61 Literaturangaben.
Abstract (Originalsprache)
Fully resolving turbulent flows remains challenging due to a turbulent systems’ multiscale complexity. Existing data-driven approaches typically demand expensive retraining for each flow scenario and struggle to generalize beyond their training conditions. Leveraging the universality of small-scale turbulent motions (Kolmogorov’s K41 theory), we propose a scale-oriented zonal generative adversarial network (SoZoGAN) framework for high-fidelity, zero-shot turbulence generation across diverse domains. Unlike conventional methods, SoZoGAN is trained exclusively on a single dataset of moderate-Reynolds-number homogeneous isotropic turbulence (HIT). The framework employs a zonal decomposition strategy, partitioning turbulent snapshots into subdomains based on scale-sensitive physical quantities. Within each subdomain, turbulence is synthesized using scale-indexed models pretrained solely on the HIT database. A SoZoGAN demonstrates high accuracy, cross-domain generalizability and robustness in zero-shot super-resolution of unsteady flows, as validated on untrained HIT, turbulent boundary layer and channel flow. Its strong generalization, demonstrated for homogeneous and inhomogeneous turbulence cases, suggests potential applicability to a wider range of industrial and natural turbulent flows. The scale-oriented zonal framework is architecture-agnostic, readily extending beyond generative adversarial networks to other deep learning models.
Zitieren
APA
Wu, H., Cao, Y., Chen, Y., Laima, S., Chen, W., Zhou, D., & Li, H. (2025). Generalizable super-resolution turbulence reconstruction from minimal training data. Journal of Fluid Mechanics, 1024. https://doi.org/10.1017/jfm.2025.10913
Vancouver
Wu H, Cao Y, Chen Y, Laima S, Chen W, Zhou D, et al. Generalizable super-resolution turbulence reconstruction from minimal training data. Journal of Fluid Mechanics. 2025;1024. doi:10.1017/jfm.2025.10913
BibTeX
@article{Wu2025,
author = {Wu, Haokai and Cao, Yong and Chen, Yaoran and Laima, Shujin and Chen, Wen-Li and Zhou, Dai and Li, Hui},
title = {Generalizable super-resolution turbulence reconstruction from minimal training data},
journal = {Journal of Fluid Mechanics},
year = {2025},
volume = {1024},
doi = {10.1017/jfm.2025.10913}
}
EUVIMED