EUVIMEDEUVIMED entdecken
Crossref

Multiple scale method integrated physics-informed neural networks for reconstructing transient natural convection

Nagahiro Ohashi, Nam Phuong Nguyen, Leslie K. Hwang, Beomjin Kwon

Physics of Fluids · 01.12.2024 · 36 (12)

DOI: 10.1063/5.0243548 ↗

Kurz gefasst

Fachartikel in der Zeitschrift Physics of Fluids, online erschienen am 17.12.2024. Verfasst von Nagahiro Ohashi und 3 weiteren Autor:innen. Die Arbeit wurde bisher 13-mal zitiert (Crossref) und nennt 24 Literaturangaben. Gefördert von: National Science Foundation, Ira A. Fulton Schools of Engineering, Arizona State University.

AbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

This study employs physics-informed neural networks (PINNs) to reconstruct multiple flow fields in a transient natural convection system solely based on instantaneous temperature data at an arbitrary moment. Transient convection problems present reconstruction challenges due to the temporal variability of fields across different flow phases. In general, large reconstruction errors are observed during the incipient phase, while the quasi-steady phase exhibits relatively smaller errors, reduced by a factor of 2–4. We hypothesize that reconstruction errors vary across different flow phases due to the changing solution space of a PINN, inferred from the temporal gradients of the fields. Furthermore, we find that reconstruction errors tend to accumulate in regions where the spatial gradients are smaller than the order of 10−6, likely due to the vanishing gradient phenomenon. In convection phenomena, field variations often manifest across multiple scales in space. However, PINN-based reconstruction tends to preserve larger-scale variations, while smaller-scale variations become less pronounced due to the vanishing gradient problem. To mitigate the errors associated with vanishing gradients, we introduce a multi-scale approach that determines scaling constants for the PINN inputs and reformulates inputs across multiple scales. This approach improves the maximum and mean errors by 72.2% and 6.4%, respectively. Our research provides insight into the behavior of PINNs when applied to transient convection problems with large solution space and field variations across multiple scales.

Förderung

  • National Science Foundation
  • Ira A. Fulton Schools of Engineering, Arizona State University

Zitieren

APA

Ohashi, N., Nguyen, N. P., Hwang, L. K., & Kwon, B. (2024). Multiple scale method integrated physics-informed neural networks for reconstructing transient natural convection. Physics of Fluids, 36(12). https://doi.org/10.1063/5.0243548

Vancouver

Ohashi N, Nguyen NP, Hwang LK, Kwon B. Multiple scale method integrated physics-informed neural networks for reconstructing transient natural convection. Physics of Fluids. 2024;36(12). doi:10.1063/5.0243548

BibTeX
@article{Ohashi2024,
  author = {Ohashi, Nagahiro and Nguyen, Nam Phuong and Hwang, Leslie K. and Kwon, Beomjin},
  title = {Multiple scale method integrated physics-informed neural networks for reconstructing transient natural convection},
  journal = {Physics of Fluids},
  year = {2024},
  volume = {36},
  number = {12},
  doi = {10.1063/5.0243548}
}
Kostenlos starten →