Machine learning for disruption warnings on Alcator C-Mod, DIII-D, and EAST

Abstract

This paper reports on disruption prediction using a shallow machine learning method known as a random forest, trained on large databases containing only plasma parameters that are available in real-time on Alcator C-Mod, DIII-D, and EAST. The database for each tokamak contains parameters sampled ∼10 6 times throughout ∼10 4 discharges (disruptive and nondisruptive) over the last four years of operation. It is found that a number of parameters (e.g. P rad /P input , i , n/n G , B n=1 /B 0 ) exhibit changes in aggregate as a disruption is approached on one or more of these tokamaks. However, for each machine, the most useful parameters, as well as the details of their precursor behaviors, are markedly different. When the prediction problem is framed using a binary classification scheme to discriminate between time slices 'close to disruption' and 'far from disruption', it is found that the prediction algorithms differ substantially in performance among the three machines on a time slice-by-time slice basis, but have similar disruption detection rates (∼80%-90%) on a shot-by-shot basis after appropriate optimisation. This could have important implications for disruption prediction and avoidance on ITER, for which development of a training database of disruptions may be infeasible. The algorithm's output is interpretable using a method that identifies the most strongly contributing input signals, which may have implications for avoiding disruptive scenarios. To further support its real-time capability, successful applications in inter-shot and real-time environments on EAST and DIII-D are also discussed.

MHD stability and disruptions in the SPARC tokamak

“…Most predictive data-driven algorithms focus on the discrimination of stable versus unstable operational spaces, even though identifying the transition time through such boundaries is in itself a challenge (Berkery et al 2017;Alessi et al 2019). Often the classification of unstable phases collapses onto predictions anticipating the CQ phase, and to date the best performing models are capable of true positive rates higher than 90 % with false positive rate below 5 %-10 % (Kates-Harbeck, Svyatkovskiy & Tang 2019; Montes et al 2019). Disruption prediction on Alcator C-Mod has proven challenging (Rea et al 2018;Montes et al 2019) due to the high fraction of disruptions caused by molybdenum flecks; an event with an inherent time scale of the order of milliseconds.…”

“…Often the classification of unstable phases collapses onto predictions anticipating the CQ phase, and to date the best performing models are capable of true positive rates higher than 90 % with false positive rate below 5 %-10 % (Kates-Harbeck, Svyatkovskiy & Tang 2019; Montes et al 2019). Disruption prediction on Alcator C-Mod has proven challenging (Rea et al 2018;Montes et al 2019) due to the high fraction of disruptions caused by molybdenum flecks; an event with an inherent time scale of the order of milliseconds. Continuous monitoring of the plasma-facing components to detect hot spots that might lead to material injection, for example via infrared camera coverage, is envisioned as critical for SPARC operations.…”

Advancing Fusion with Machine Learning Research Needs Workshop Report

“…Predictions from machine learning models trained on large data sets have been employed in fusion energy research since the early-1990s. For example, Wroblewski et al [74] employed a neural network to predict high beta disruptions in real-time from many axisymmetric-only input signals, Windsor et al [72] produced a multi-machine applicable disruption predictor for JET and ASDEX-UG, Rea et al [61] and Montes et al [48] demonstrated use of time series data and explicit look-ahead time windows for disruption predictability in Alcator C-Mod, DIII-D, and EAST (see Fig. 11), and Kates-Harbeck [33] demonstrated use of extensive profile measurements in multi-machine disruption prediction for JET and DIII-D with convolutional and recurrent neural networks.…”

“…11 The left two plots compare the performances of machinespecific disruption predictors on 3 different tokamaks (EAST, DIII-D, C-Mod). The rightmost plot shows the output of a real time predictor installed in the DIII-D plasma control system, demonstrating an effective warning time of several hundred ms before disruption [48] Physics-informed machine learning is a broad area of current research that seeks to incorporate physical principles into machine learning approaches. As an example, these principles could be used to design the structure of a neural network or the covariance function of a Gaussian process.…”

An application of survival analysis to disruption prediction via Random Forests

“…Real-time calculations of this signal have been demonstrated by the DPRF algorithm implemented in the PCS system of the DIII-D tokamak [12]. Efforts are underway to perform full-discharge analyses of disruptivity signals, incorporating time evolution and optimizing both disruptivity thresholds and time-windows to trigger warnings [6]. These are discussed further in section 5.…”