Machine learning enhanced predictions of ICRF heating: Overcoming numerical limitations via data curation

Sanchez-Villar, A.; Bai, Z.; Bertelli, N.; Bethel, E. W.; Hillairet, J.; Perciano, T.; Shiraiwa, S.; Wallace, G. M.; Wright, J. C.

doi:10.1063/5.0268341

Machine learning enhanced predictions of ICRF heating: Overcoming numerical limitations via data curation

Journal Article · Fri Jun 06 00:00:00 EDT 2025 · Physics of Plasmas

DOI:https://doi.org/10.1063/5.0268341· OSTI ID:2586600

^[1]; ^[2]; ^[1]; ^[3]; ^[4]; ^[2]; ^[1]; ^[5]; ^[5]

Princeton Plasma Physics Laboratory (PPPL), Princeton, NJ (United States)
Lawrence Berkeley National Laboratory (LBNL), Berkeley, CA (United States)
San Francisco State University, CA (United States)
CEA-IRFM, Saint-Paul-Lez-Durance (France)
MIT Plasma Science and Fusion Center, Cambridge, MA (United States)

In this work, we present the development of robust surrogate models for Ion Cyclotron Range of Frequencies (ICRF) and High-Harmonic Fast Wave (HHFW) heating predictions in fusion plasmas. Building upon our previous efforts to achieve real-time capable models, we identify the cause of the outliers found using TORIC in certain HHFW heating scenarios. The outliers are observed to be spurious ion Bernstein wave (IBW)-like modes caused by a wavelength control algorithm designed to address challenging scenarios with high perpendicular wavenumbers. The effect arises from the modulation in the perpendicular susceptibility, which can induce sign reversal and IBW-like propagation for scenarios featuring normalized ion Larmor radius λ_i ≫ 1. We use TORIC with this algorithm disabled to generate a novel HHFW-NSTX database that is free of outliers. Surrogate models trained on this database, including Random Forest Regressor (RFR), Multi-Layer Perceptrons, and Gaussian Process Regressors (GPR), demonstrate the ability to accurately predict HHFW heating profiles, with regression scores of R²∈[0.93−0.99]. Additionally we demonstrate that it is possible to generalize predictions beyond training data by the use of both RFR and GPR models, enabling the prediction of scenarios previously limited to the original model. GPR models also provide uncertainty quantification, offering insights into model confidence. This work introduces a comprehensive Verification, Validation, and Uncertainty Quantification methodology for surrogate modeling, applicable not only to ICRF heating but also to other RF heating challenges and fusion physics problems. Beyond accelerated inference, these models show effective extrapolation capabilities, providing an alternative for addressing numerical challenges.

View Accepted Manuscript (DOE)

Research Organization:: Princeton Plasma Physics Laboratory (PPPL), Princeton, NJ (United States)

Sponsoring Organization:: USDOE Office of Science (SC), Basic Energy Sciences (BES). Scientific User Facilities (SUF)

Grant/Contract Number:: AC02-09CH11466; AC02-05CH11231

OSTI ID:: 2586600

Alternate ID(s):: OSTI ID: 2573794

Journal Information:: Physics of Plasmas, Journal Name: Physics of Plasmas Journal Issue: 6 Vol. 32; ISSN 1070-664X; ISSN 1089-7674

Publisher:: American Institute of Physics (AIP)Copyright Statement

Country of Publication:: United States

Language:: English

References (39)

Advancing Fusion with Machine Learning Research Needs Workshop Report Humphreys, David; Kupresanin, A.; Boyer, M. D. Journal of Fusion Energy, Vol. 39, Issue 4 https://doi.org/10.1007/s10894-020-00258-1	journal	August 2020
Extremely randomized trees Geurts, Pierre; Ernst, Damien; Wehenkel, Louis Machine Learning, Vol. 63, Issue 1 https://doi.org/10.1007/s10994-006-6226-1	journal	March 2006
An Empirical Approach to Tokamak Transport Hawryluk, R. J. Physics of Plasmas Close to Thermonuclear Conditions, p. 19-46 https://doi.org/10.1016/B978-1-4832-8385-2.50009-1	book	January 1979
Data-driven surrogate modeling of hPIC ion energy-angle distributions for high-dimensional sensitivity analysis of plasma parameters' uncertainty Seleson, Pablo; Mustafa, Mohammad; Curreli, Davide Computer Physics Communications, Vol. 279 https://doi.org/10.1016/j.cpc.2022.108436	journal	October 2022
Towards fast and accurate predictions of radio frequency power deposition and current profile via data-driven modelling: applications to lower hybrid current drive Wallace, G. M.; Bai, Z.; Sadre, R. Journal of Plasma Physics, Vol. 88, Issue 4 https://doi.org/10.1017/S0022377822000708	journal	August 2022
Data augmentation for disruption prediction via robust surrogate models Rath, Katharina; Rügamer, David; Bischl, Bernd Journal of Plasma Physics, Vol. 88, Issue 5 https://doi.org/10.1017/S0022377822000769	journal	October 2022
Random Forests Breiman, Leo Machine Learning, Vol. 45, Issue 1, p. 5-32 https://doi.org/10.1023/A:1010933404324	journal	January 2001
Learning representations by back-propagating errors Rumelhart, David E.; Hinton, Geoffrey E.; Williams, Ronald J. Nature, Vol. 323, Issue 6088 https://doi.org/10.1038/323533a0	journal	October 1986
Avoiding fusion plasma tearing instability with deep reinforcement learning Seo, Jaemin; Kim, SangKyeun; Jalalvand, Azarakhsh Nature, Vol. 626, Issue 8000 https://doi.org/10.1038/s41586-024-07024-9	journal	February 2024
SciPy 1.0: fundamental algorithms for scientific computing in Python Virtanen, Pauli; Gommers, Ralf; Oliphant, Travis E. Nature Methods https://doi.org/10.1038/s41592-019-0686-2	journal	February 2020
Application of machine learning for detecting and tracking turbulent structures in plasma fusion devices using ultra fast imaging Chouchene, Sarah; Brochard, Frédéric; Desecures, Mikael Scientific Reports, Vol. 14, Issue 1 https://doi.org/10.1038/s41598-024-79251-z	journal	November 2024
RF wave simulation for cold edge plasmas using the MFEM library Shiraiwa, S.; Wright, J. C.; Bonoli, P. T. EPJ Web of Conferences, Vol. 157 https://doi.org/10.1051/epjconf/201715703048	journal	January 2017
All-orders spectral calculation of radio-frequency heating in two-dimensional toroidal plasmas Jaeger, E. F.; Berry, L. A.; D’Azevedo, E. Physics of Plasmas, Vol. 8, Issue 5 https://doi.org/10.1063/1.1359516	journal	May 2001
Simulation of high-power electromagnetic wave heating in the ITER burning plasma Jaeger, E. F.; Berry, L. A.; D’Azevedo, E. F. Physics of Plasmas, Vol. 15, Issue 7 https://doi.org/10.1063/1.2959128	journal	July 2008
High non-inductive fraction H-mode discharges generated by high-harmonic fast wave heating and current drive in the National Spherical Torus Experiment Taylor, G.; Hosea, J. C.; Kessel, C. E. Physics of Plasmas, Vol. 19, Issue 4 https://doi.org/10.1063/1.3699364	journal	April 2012
Progress on ion cyclotron range of frequencies heating physics and technology in support of the International Tokamak Experimental Reactor Wilson, J. R.; Bonoli, P. T. Physics of Plasmas, Vol. 22, Issue 2 https://doi.org/10.1063/1.4901090	journal	November 2014
A generalized plasma dispersion function for electron damping in tokamak plasmas Berry, L. A.; Jaeger, E. F.; Phillips, C. K. Physics of Plasmas, Vol. 23, Issue 10 https://doi.org/10.1063/1.4964766	journal	October 2016
Fast modeling of turbulent transport in fusion plasmas using neural networks van de Plassche, K. L.; Citrin, J.; Bourdelle, C. Physics of Plasmas, Vol. 27, Issue 2 https://doi.org/10.1063/1.5134126	journal	February 2020
High harmonic fast waves in high beta plasmas Ono, Masayuki Physics of Plasmas, Vol. 2, Issue 11 https://doi.org/10.1063/1.871030	journal	November 1995
Mode conversion electron heating in Alcator C-Mod: Theory and experiment Bonoli, P. T.; Brambilla, M.; Nelson-Melby, E. Physics of Plasmas, Vol. 7, Issue 5 https://doi.org/10.1063/1.874012	journal	May 2000
Fast-wave heating of a two-component plasma Stix, T. H. Nuclear Fusion, Vol. 15, Issue 5 https://doi.org/10.1088/0029-5515/15/5/003	journal	October 1975
Heating tokamaks via the ion-cyclotron and ion-ion hybrid resonances Perkins, F. W. Nuclear Fusion, Vol. 17, Issue 6 https://doi.org/10.1088/0029-5515/17/6/008	journal	December 1977
Simulation of ion cyclotron heating of tokamak plasmas using coupled Maxwell and quasilinear-Fokker–Planck solvers Brambilla, M.; Bilato, R. Nuclear Fusion, Vol. 46, Issue 7 https://doi.org/10.1088/0029-5515/46/7/S01	journal	June 2006
Variational approach to radiofrequency waves in magnetic fusion devices Dumont, R. J. Nuclear Fusion, Vol. 49, Issue 7 https://doi.org/10.1088/0029-5515/49/7/075033	journal	July 2009
Finite Larmor radius wave equations in Tokamak plasmas in the ion cyclotron frequency range Brambilla, M. Plasma Physics and Controlled Fusion, Vol. 31, Issue 5 https://doi.org/10.1088/0741-3335/31/5/004	journal	May 1989
On the radiofrequency response of tokamak plasmas Lamalle, P. U. Plasma Physics and Controlled Fusion, Vol. 39, Issue 9 https://doi.org/10.1088/0741-3335/39/9/011	journal	September 1997
Numerical simulation of ion cyclotron waves in tokamak plasmas Brambilla, M. Plasma Physics and Controlled Fusion, Vol. 41, Issue 1 https://doi.org/10.1088/0741-3335/41/1/002	journal	January 1999
`Quasi-local' wave equations in toroidal geometry with applications to fast wave propagation and absorption at high harmonics of the ion cyclotron frequency Brambilla, Marco Plasma Physics and Controlled Fusion, Vol. 44, Issue 11 https://doi.org/10.1088/0741-3335/44/11/307	journal	October 2002
Incorporating large larmor radius effects in the full wave code TORIC Brambilla, Marco; Bilato, Roberto Plasma Physics and Controlled Fusion, Vol. 66, Issue 7 https://doi.org/10.1088/1361-6587/ad4741	journal	May 2024
Overview of NSTX Upgrade initial results and modelling highlights Menard, J. E.; Allain, J. P.; Battaglia, D. J. Nuclear Fusion, Vol. 57, Issue 10, Article No. 102006 https://doi.org/10.1088/1741-4326/aa600a	journal	June 2017
Self-consistent core-pedestal transport simulations with neural network accelerated models Meneghini, O.; Smith, S. P.; Snyder, P. B. Nuclear Fusion, Vol. 57, Issue 8 https://doi.org/10.1088/1741-4326/aa7776	journal	July 2017
Real-time capable modeling of neutral beam injection on NSTX-U using neural networks Boyer, M. D.; Kaye, S.; Erickson, K. Nuclear Fusion, Vol. 59, Issue 5 https://doi.org/10.1088/1741-4326/ab0762	journal	March 2019
A real-time machine learning-based disruption predictor in DIII-D Rea, C.; Montes, K. J.; Erickson, K. G. Nuclear Fusion, Vol. 59, Issue 9 https://doi.org/10.1088/1741-4326/ab28bf	journal	July 2019
Neural-network accelerated coupled core-pedestal simulations with self-consistent transport of impurities and compatible with ITER IMAS Meneghini, O.; Snoep, G.; Lyons, B. C. Nuclear Fusion, Vol. 61, Issue 2 https://doi.org/10.1088/1741-4326/abb918	journal	December 2020
Operating a full tungsten actively cooled tokamak: overview of WEST first phase of operation Bucalossi, J.; Achard, J.; Agullo, O. Nuclear Fusion, Vol. 62, Issue 4 https://doi.org/10.1088/1741-4326/ac2525	journal	February 2022
3D full wave fast wave modeling with realistic HHFW antenna geometry and SOL plasma in NSTX-U Bertelli, N.; Shiraiwa, S.; Ono, M. Nuclear Fusion, Vol. 62, Issue 12 https://doi.org/10.1088/1741-4326/ac9690	journal	October 2022
Machine learning-based real-time kinetic profile reconstruction in DIII-D Shousha, Ricardo; Seo, Jaemin; Erickson, Keith Nuclear Fusion, Vol. 64, Issue 2 https://doi.org/10.1088/1741-4326/ad142f	journal	December 2023
Automated experimental design of safe rampdowns via probabilistic machine learning Mehta, Viraj; Barr, Jayson; Abbate, Joseph Nuclear Fusion, Vol. 64, Issue 4 https://doi.org/10.1088/1741-4326/ad22f5	journal	February 2024
ICRF Heating Theory Perkins, F. W. IEEE Transactions on Plasma Science, Vol. 12, Issue 2 https://doi.org/10.1109/TPS.1984.4316299	journal	January 1984

Similar Records

Related Subjects

Artificial neural networks
Dielectric properties
Gaussian processes
Machine learning
Plasma heating
Plasma waves

Machine learning enhanced predictions of ICRF heating: Overcoming numerical limitations via data curation

Citation Formats

References (39)

Similar Records

Related Subjects