Data-driven profile prediction for DIII-D

A new, fully data-driven algorithm has been developed that uses a neural network to predict plasma profiles on a scale of τ E into the future given an actuator trajectory and the plasma state history. The model was trained and tested on DIII-D data from the 2013–2018 experimental campaigns. The model runs in tens of milliseconds and is very simple to use. This makes it a potentially useful tool for operators and physicists when planning plasma scenarios. It is also fast enough to be used for real-time model-predictive control.

[1]  Carlos Simpson,et al.  Stacks , 2005, New Spaces in Mathematics.

[2]  L. L. Lao,et al.  Self-consistent core-pedestal transport simulations with neural network accelerated models , 2017 .

[3]  Wendell Horton,et al.  Neural network prediction of some classes of tokamak disruptions , 1996 .

[4]  Piergiorgio Sonato,et al.  A prediction tool for real-time application in the disruption protection system at JET , 2007 .

[5]  O. Meneghini,et al.  Real-time pedestal optimization and ELM control with 3D fields and gas flows on DIII-D , 2020, Nuclear Fusion.

[6]  A. Sengupta,et al.  Forecasting disruptions in the ADITYA tokamak using neural networks , 2000 .

[7]  Geoffrey E. Hinton,et al.  ImageNet classification with deep convolutional neural networks , 2012, Commun. ACM.

[8]  L. Lao,et al.  MHD Equilibrium Reconstruction in the DIII-D Tokamak , 2005 .

[9]  Andrea Murari,et al.  An advanced disruption predictor for JET tested in a simulated real-time environment , 2010 .

[10]  G. Staebler,et al.  Progress and challenges in understanding core transport in tokamaks in support to ITER operations , 2019, Plasma Physics and Controlled Fusion.

[11]  J. Contributors,et al.  Fast modelling of turbulent transport in fusion plasmas using neural networks , 2019 .

[12]  Thomas Zehetbauer,et al.  Prediction and mitigation of disruptions in ASDEX Upgrade , 2001 .

[13]  Yoshua Bengio,et al.  Deep Sparse Rectifier Neural Networks , 2011, AISTATS.

[14]  M. Reich,et al.  Real-time beam tracing for control of the deposition location of electron cyclotron waves , 2015 .

[15]  R. Yoshino.  Neural-net disruption predictor in JT-60U , 2003 .

[16]  I. Voitsekhovitch,et al.  Numerical Transport Codes , 2004 .

[17]  J. Stillerman,et al.  MDSplus data acquisition system , 1997 .

[18]  Orso Meneghini,et al.  Orchestrating TRANSP Simulations for Interpretative and Predictive Tokamak Modeling with OMFIT , 2018 .

[19]  E. Kolemen,et al.  CAKE: Consistent Automatic Kinetic Equilibrium reconstruction , 2021 .

[20]  D. McCune,et al.  New techniques for calculating heat and particle source rates due to neutral beam injection in axisymmetric tokamaks , 1981 .

[21]  Sepp Hochreiter,et al.  Fast and Accurate Deep Network Learning by Exponential Linear Units (ELUs) , 2015, ICLR.

[22]  R. J. Hawryluk,et al.  An Empirical Approach to Tokamak Transport , 1981 .

[23]  J. A. Leuer,et al.  Tokamak disruption alarm based on a neural network model of the high- beta limit , 1997 .

[24]  R. Waltz,et al.  A gyro-Landau-fluid transport model , 1997 .

[25]  Yuanzhi Li,et al.  Convergence Analysis of Two-layer Neural Networks with ReLU Activation , 2017, NIPS.

[26]  R. Dux,et al.  Neoclassical impurity transport in the core of an ignited tokamak plasma , 2000 .

[27]  A. D. Turnbull,et al.  Integrated modeling applications for tokamak experiments with OMFIT , 2015 .

[28]  Faa Federico Felici,et al.  RABBIT: Real-time simulation of the NBI fast-ion distribution , 2018, Nuclear Fusion.

[29]  Jürgen Schmidhuber,et al.  Learning to forget: continual prediction with LSTM , 1999 .

[30]  Faa Federico Felici,et al.  Real-time-capable prediction of temperature and density profiles in a tokamak using RAPTOR and a first-principle-based transport model , 2018, Nuclear Fusion.

[31]  P. B. Snyder,et al.  The EPED pedestal model and edge localized mode-suppressed regimes: Studies of quiescent H-mode and development of a model for edge localized mode suppression via resonant magnetic perturbations , 2012 .

[32]  Mark D. Boyer,et al.  Machine learning control for disruption and tearing mode avoidance , 2020 .

[33]  G. Pautasso,et al.  Prediction of disruptions on ASDEX Upgrade using discriminant analysis , 2011 .

[34]  C. D. Challis,et al.  Non-inductively driven currents in JET , 1989 .

[35]  Michael I. Jordan,et al.  Advances in Neural Information Processing Systems 30 , 1995 .

[36]  Thomas Brox,et al.  U-Net: Convolutional Networks for Biomedical Image Segmentation , 2015, MICCAI.

[37]  G. Staebler,et al.  Neural-network accelerated coupled core-pedestal simulations with self-consistent transport of impurities and compatible with ITER IMAS , 2020, Nuclear Fusion.

[38]  Geoffrey E. Hinton,et al.  Rectified Linear Units Improve Restricted Boltzmann Machines , 2010, ICML.

[39]  JET Contributors,et al.  Tractable flux-driven temperature, density, and rotation profile evolution with the quasilinear gyrokinetic transport model QuaLiKiz , 2017, 1708.01224.

[40]  Gyrokinetic simulation of global and local Alfvén eigenmodes driven by energetic particles in a DIII-D discharge , 2013 .

[41]  A. Murari,et al.  Unbiased and non-supervised learning methods for disruption prediction at JET , 2009 .

[42]  Julian Kates-Harbeck,et al.  Predicting disruptive instabilities in controlled fusion plasmas through deep learning , 2019, Nature.

[43]  J. Citrin,et al.  First principle integrated modeling of multi-channel transport including Tungsten in JET , 2018, Nuclear Fusion.

[44]  A. V. Olgac,et al.  Performance Analysis of Various Activation Functions in Generalized MLP Architectures of Neural Networks , 2011 .