Skip to main content
U.S. Department of Energy
Office of Scientific and Technical Information

Transfer learning nonlinear plasma dynamic transitions in low dimensional embeddings via deep neural networks

Journal Article · · Machine Learning: Science and Technology
Deep learning algorithms provide a new paradigm to study high-dimensional dynamical behaviors, such as those in fusion plasma systems. Development of novel, data-driven model reduction methods, coupled with detection of abnormal modes with plasma physics, opens a unique opportunity to identify plasma instabilities through automated construction of parsimonious models that can be tuned to balance accuracy and cost. Our fusion transfer learning (FTL) model demonstrates success in rapidly reconstructing nonlinear kink mode structures by learning from a limited amount of nonlinear simulation data. The knowledge transfer process leverages a pre-trained neural encoder–decoder network, initially trained on linear simulations, to effectively capture nonlinear dynamics. The low-dimensional embeddings extract the coherent structures of interest, while preserving the inherent dynamics of the complex system. Experimental results highlight FTL’s capacity to capture transitional behaviors and dynamical features in plasma dynamics—a task often challenging for conventional methods. The model developed in this study is generalizable and can be extended broadly through transfer learning to address various magnetohydrodynamics modes.
Research Organization:
Lawrence Berkeley National Laboratory (LBNL), Berkeley, CA (United States); Lawrence Berkeley National Laboratory (LBNL), Berkeley, CA (United States). National Energy Research Scientific Computing Center (NERSC); Princeton Plasma Physics Laboratory (PPPL), Princeton, NJ (United States)
Sponsoring Organization:
USDOE; USDOE Office of Science (SC), Advanced Scientific Computing Research (ASCR). Scientific Discovery through Advanced Computing (SciDAC); USDOE Office of Science (SC), Basic Energy Sciences (BES). Scientific User Facilities (SUF); USDOE Office of Science (SC), Fusion Energy Sciences (FES)
Grant/Contract Number:
AC02-05CH11231; AC02-09CH11466
OSTI ID:
2557550
Alternate ID(s):
OSTI ID: 2551903
OSTI ID: 2567023
OSTI ID: 2564375
Journal Information:
Machine Learning: Science and Technology, Journal Name: Machine Learning: Science and Technology Journal Issue: 2 Vol. 6; ISSN 2632-2153
Publisher:
IOP PublishingCopyright Statement
Country of Publication:
United States
Language:
English

References (48)

Implementation of AI/DEEP learning disruption predictor into a plasma control system journal June 2023
Physics-informed neural networks (PINNs) for fluid mechanics: a review journal December 2021
Spectral Properties of Dynamical Systems, Model Reduction and Decompositions journal August 2005
Numerical calculations using the full MHD equations in toroidal geometry journal March 1986
Compressible linear and nonlinear resistive MHD calculations in toroidal geometry journal February 1990
An Eulerian gyrokinetic-Maxwell solver journal April 2003
Operator inference for non-intrusive model reduction of systems with non-polynomial nonlinear terms journal December 2020
Model Predictive Control of the Current Density Distribution and Stored Energy in Tokamak Fusion Experiments using Trajectory Linearizations**This work has been carried out within the framework of the EUROfusion Consortium and has received funding from the Euratom research and training programme 2014-2018 under grant agreement No 633053. The views and opinions expressed herein do not necessarily reect those of the European Commission. journal January 2015
Nonlinear magnetohydrodynamics simulation using high-order finite elements journal March 2004
A triangular finite element with first-derivative continuity applied to fusion MHD applications journal October 2004
Electromagnetic gyrokinetic δf particle-in-cell turbulence simulation with realistic equilibrium profiles and geometry journal January 2007
The global version of the gyrokinetic turbulence code GENE journal August 2011
Physics-informed neural networks: A deep learning framework for solving forward and inverse problems involving nonlinear partial differential equations journal February 2019
Gradient-based constrained optimization using a database of linear reduced-order models journal December 2020
Reduced-order modeling: new approaches for computational physics journal February 2004
Spectral analysis of nonlinear flows journal November 2009
Dynamic mode decomposition of numerical and experimental data journal July 2010
Koopman-mode decomposition of the cylinder wake journal June 2013
Predicting disruptive instabilities in controlled fusion plasmas through deep learning journal April 2019
Magnetic control of tokamak plasmas through deep reinforcement learning journal February 2022
Numerical study of neoclassical plasma pedestal in a tokamak geometry journal May 2004
Finite element implementation of Braginskii’s gyroviscous stress with application to the gravitational instability journal September 2006
Island bootstrap current modification of the nonlinear dynamics of the tearing mode journal January 1986
Electron temperature gradient driven turbulence journal May 2000
Excitation of zonal flow by drift waves in toroidal plasmas journal August 2000
Hamiltonian Systems and Transformation in Hilbert Space journal May 1931
Chapter 1: Overview and summary journal December 1999
Chapter 2: Plasma confinement and transport journal December 1999
Full-f gyrokinetic particle simulation of centrally heated global ITG turbulence from magnetic axis to edge pedestal top in a realistic tokamak geometry journal September 2009
NIMROD modeling of quiescent H-mode: reconstruction considerations and saturation mechanism journal September 2016
Neoclassical tearing modes journal December 2000
Zonal flows in plasma—a review journal April 2005
Deep learning based surrogate models for first-principles global simulations of fusion plasmas journal November 2021
Verification and validation of linear gyrokinetic and kinetic-MHD simulations for internal kink instability in DIII-D tokamak journal January 2022
Comparison of machine learning systems trained to detect Alfvén eigenmodes using the CO2 interferometer on DIII-D journal October 2023
Principal component analysis: a review and recent developments
  • Jolliffe, Ian T.; Cadima, Jorge
  • Philosophical Transactions of the Royal Society A: Mathematical, Physical and Engineering Sciences, Vol. 374, Issue 2065 https://doi.org/10.1098/rsta.2015.0202
journal April 2016
Regulation of Alfvén Eigenmodes by Microturbulence in Fusion Plasmas journal May 2022
Saturation of Fishbone Instability by Self-Generated Zonal Flows in Tokamak Plasmas journal February 2024
Studies of Internal Disruptions and m = 1 Oscillations in Tokamak Discharges with Soft—X-Ray Tecniques journal November 1974
Study of High-Beta Magnetohydrodynamic Modes and Fast-Ion Losses in PDX journal March 1983
Turbulent Transport Reduction by Zonal Flows: Massively Parallel Simulations journal September 1998
A Survey of Projection-Based Model Reduction Methods for Parametric Dynamical Systems journal January 2015
Dynamic Mode Decomposition with Control journal January 2016
FTL: Transfer Learning Nonlinear Plasma Dynamic Transitions in Low Dimensional Embeddings (FTL) v1.0 software January 2024
Non-intrusive nonlinear model reduction via machine learning approximations to low-dimensional operators journal December 2021
MHD Instabilities in Tokamaks journal February 2008
Modal Analysis of Fluid Flows: An Overview journal December 2017
Dynamic Mode Decomposition for Compressive System Identification journal February 2020