Exploratory Analysis of Fast Rotating Mode Signals as Precursors of Locked Mode-driven Disruptions at DIII-D

This work investigates the information content of fast rotating modes (FRM) as precursors of locked mode-driven disruptions, in DIII-D ITER Baseline Discharges (IBS). Rotating precursors, when existing, are not easily identifiable; their evolution in time can strongly vary between 20 and 200ms. We present an exploratory analysis on signals coming from a specific combination of the Mirnov coils, capable of detecting FRMs while discriminating between even and odd poloidal mode number. The analysis focuses on even m modes, these being likely precursors of m/n=2/1 locked modes at DIII-D. We use amplitude and frequency of even m FRM signals in Machine Learning-based models, exploring their correlation features with B_rⁿ⁼¹, proxy for nonrotating modes. We additionally include 12 plasma parameters, e.g. the normalized internal inductance and the pedestal information, for ∼200 discharges. Different Machine Learning algorithms are implemented, using both regression and classification schemes. While it is possible to see a correlation between the FRM amplitude and B_rⁿ⁼¹ when inspecting the shots, FRM precursors information seem to play a small role in the predictive algorithms.