Machine learning control for disruption and tearing mode avoidance

Real-time feedback control based on machine learning algorithms (MLA) was successfully developed and tested on DIII-D plasmas to avoid tearing modes and disruptions while maximizing the plasma performance, which is measured by normalized plasma beta. The control uses MLAs that were trained with ensemble learning methods using only the data available to the real-time Plasma Control System (PCS) from several thousand DIII-D discharges. A “tearability” metric that quantifies the likelihood of the onset of 2/1 tearing modes in a given time window, and a “disruptivity” metric that quantifies the likelihood of the onset of plasma disruptions were first tested off-line and then implemented on the PCS. A real-time control system based on these MLAs was successfully tested on DIII-D discharges, using feedback algorithms to maximize βN while avoiding tearing modes and to dynamically adjust ramp down to avoid high-current disruptions in ramp down.

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

[2]  N. W. Eidietis,et al.  Disruptions in ITER and strategies for their control and mitigation , 2015 .

[3]  J. Vega,et al.  Adaptive high learning rate probabilistic disruption predictors from scratch for the next generation of tokamaks , 2014 .

[4]  N. W. Eidietis,et al.  Implementing a finite-state off-normal and fault response system for disruption avoidance in tokamaks , 2018 .

[5]  R. E. Bell,et al.  A reduced resistive wall mode kinetic stability model for disruption forecasting , 2017 .

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

[7]  V. Udintsev,et al.  Heterodyne ECE diagnostic in the mode detection and disruption avoidance at TEXTOR , 2003 .

[8]  A. Donné,et al.  Effect of heating on the suppression of tearing modes in tokamaks. , 2007, Physical review letters.

[9]  J. Contributors,et al.  Survey of disruption causes at JET , 2011 .

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

[11]  David A. Landgrebe,et al.  A survey of decision tree classifier methodology , 1991, IEEE Trans. Syst. Man Cybern..

[12]  Leo Breiman,et al.  Random Forests , 2001, Machine Learning.

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

[14]  Gilles Louppe,et al.  Understanding variable importances in forests of randomized trees , 2013, NIPS.

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

[16]  Mori,et al.  Avoidance of qa=3 disruption by electron cyclotron heating in the JFT-2M tokamak. , 1992, Physical review letters.

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

[18]  M. L. Walker,et al.  A flexible software architecture for tokamak discharge control systems , 1995, Proceedings of 16th International Symposium on Fusion Engineering.

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

[20]  R. Polikar,et al.  Ensemble based systems in decision making , 2006, IEEE Circuits and Systems Magazine.

[21]  Jet Efda Contributors,et al.  A cross-tokamak neural network disruption predictor for the JET and ASDEX Upgrade tokamaks , 2005 .

[22]  Leo Breiman,et al.  Bagging Predictors , 1996, Machine Learning.

[23]  G. Pautasso,et al.  Requirements for Triggering the ITER Disruption Mitigation System , 2016 .

[24]  Tom Fawcett,et al.  An introduction to ROC analysis , 2006, Pattern Recognit. Lett..

[25]  Gaël Varoquaux,et al.  Scikit-learn: Machine Learning in Python , 2011, J. Mach. Learn. Res..

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

[27]  Yoav Freund,et al.  A Short Introduction to Boosting , 1999 .

[28]  L. L. Lao,et al.  High beta tokamak operation in DIII-D limited at low density/collisionality by resistive tearing modes , 1997 .

[29]  D. A. Humphreys,et al.  Experimental simulation of ITER rampdown in DIII-D , 2010 .

[30]  D. Boos.  Introduction to the Bootstrap World , 2003 .

[31]  Alessandra Fanni,et al.  Disruption forecasting at JET using neural networks , 2002 .

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

[33]  L. L. Lao,et al.  Real time equilibrium reconstruction for tokamak discharge control , 1998 .

[34]  J. Schweinzer,et al.  Optimized tokamak power exhaust with double radiative feedback in ASDEX Upgrade , 2012 .

[35]  V. Igochine,et al.  Active Control of Magneto-hydrodynamic Instabilities in Hot Plasmas , 2015 .

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

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

[38]  N. W. Eidietis,et al.  Disruption prediction investigations using Machine Learning tools on DIII-D and Alcator C-Mod , 2018, Plasma Physics and Controlled Fusion.

[39]  Olivier Sauter,et al.  On the form of NTM onset scalings , 2004 .

[40]  Pierre Geurts,et al.  Extremely randomized trees , 2006, Machine Learning.

[41]  D. Ruppert.  The Elements of Statistical Learning: Data Mining, Inference, and Prediction , 2004 .

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

[43]  Allen H. Boozer,et al.  Theory of tokamak disruptionsa) , 2012 .

[44]  S. A. Sabbagh,et al.  Detection of disruptions in the high-β spherical torus NSTX , 2013 .

[45]  A. Murari,et al.  Real-Time Implementation in JET of the SPAD Disruption Predictor Using MARTe , 2018, IEEE Transactions on Nuclear Science.

[46]  R. R. Khayrutdinov,et al.  Multi-machine analysis of termination scenarios with comparison to simulations of controlled shutdown of ITER discharges , 2018 .

[47]  J. R. Martin-Solis,et al.  Disruption avoidance in the Frascati Tokamak Upgrade by means of magnetohydrodynamic mode stabilization using electron-cyclotron-resonance heating. , 2008, Physical review letters.

[48]  L. L. Lao,et al.  LONG-PULSE, HIGH-PERFORMANCE DISCHARGES IN THE DIII-D TOKAMAK , 2000 .

[49]  J. Vega,et al.  Results of the JET real-time disruption predictor in the ITER-like wall campaigns , 2012 .