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

In this paper, we report on disruption prediction using a shallow machinelearning method known as Random Forests, trained on large databases containingonly plasma parameters that are available in real-time on Alcator C-Mod, DIII-D, and EAST. The database for each tokamak contains parameters sampled ~10⁶ times throughout ~10⁴ discharges (disruptive and non-disruptive) over the last 4years of operation. It is found that a number of parameters (e.g.P_rad/P_input,l_i,n/n_G,B_n=1/B₀) exhibit changes in aggregate as a disruption is approached on one ormore 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 ona 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 developmentof a training database of disruptions may be infeasible. The algorithm's output is interpretable using a method that identifies the most strongly contributing inputsignals, which may have implications for avoiding disruptive scenarios. Lastly, to further support its real-time capability, successful applications in inter-shot and real-time environments on EAST and DIII-D are also discussed.