Physics-constrained neural networks for real-time reconstruction of radiated power in DIII-D plasma control system
Review of Scientific Instruments · 01.11.2026 · 97 (11)
DOI: 10.1063/5.0348668 ↗Kurz gefasst
Fachartikel in der Zeitschrift Review of Scientific Instruments, online erschienen am 03.11.2026. Verfasst von T Odstrcil und 6 weiteren Autor:innen. Beteiligte Einrichtung: General Atomics, San Diego, California 92121, USA. Die Arbeit nennt 24 Literaturangaben. Gefördert von: Office of Science.
Abstract (Originalsprache)
A physics-constrained neural network is developed to predict radiated power across multiple plasma regions from foil bolometer measurements using a limited set of real-time equilibrium features. The surrogate model enforces a linear mapping between measured brightness and radiated power, enabling efficient inference with microsecond latency, suitable for real-time deployment. Model training is performed on a combination of large synthetic and real experimental datasets, allowing the neural network to learn the mapping between the measured signal and radiated power across varying plasma shapes while accounting for experimental noise, thereby improving its robustness. The model is optimized to estimate near-axis radiated power, enabling early detection of ongoing impurity accumulation. The approach has been implemented in the DIII-D plasma control system, demonstrating its capability for real-time operation.
Autor:innen und Institutionen
- General Atomics, San Diego, California 92121, USA.
- Lawrence Berkeley National Laboratory, Berkeley, California 94720, USA.
- Princeton Plasma Physics Laboratory, Princeton, New Jersey 08540, USA.
ORCID-Profile: T Odstrcil, Z Bai, H J Farre-Kaga, W Choi, D Eldon, B Sammuli, A Moser
Förderung
- Office of Science
Zitieren
APA
Odstrcil, T., Bai, Z., Farre-Kaga, H. J., Choi, W., Eldon, D., Sammuli, B., & Moser, A. (2026). Physics-constrained neural networks for real-time reconstruction of radiated power in DIII-D plasma control system. Review of Scientific Instruments, 97(11). https://doi.org/10.1063/5.0348668
Vancouver
Odstrcil T, Bai Z, Farre-Kaga HJ, Choi W, Eldon D, Sammuli B, et al. Physics-constrained neural networks for real-time reconstruction of radiated power in DIII-D plasma control system. Review of Scientific Instruments. 2026;97(11). doi:10.1063/5.0348668
BibTeX
@article{Odstrcil2026,
author = {Odstrcil, T and Bai, Z and Farre-Kaga, H J and Choi, W and Eldon, D and Sammuli, B and Moser, A},
title = {Physics-constrained neural networks for real-time reconstruction of radiated power in DIII-D plasma control system},
journal = {Review of Scientific Instruments},
year = {2026},
volume = {97},
number = {11},
doi = {10.1063/5.0348668}
}
EUVIMED