Performance analysis of the centroid method predictor implemented in the JET real time network

Abstract

Abstract Recently, a linear disruption predictor (Vega et al 2020 Nucl. Fusion 60 026001) was installed in the JET real-time network for disruption mitigation purposes. From a mathematical point of view, the predictor is based on computing centroids of disruptive examples and non-disruptive examples in a two-dimensional space. This is the reason for calling it centroid method (CM). It uses a single signal: the mode lock normalized to the plasma current. The predictor is not based on thresholds to trigger alarms but on the differences of amplitudes between consecutive samples. The article analyses its results for the range of discharges 94 152–97 137 (June 2019–March 2020), including discharges of both baseline scenario and hybrid scenario. The article presents a comparison between the CM predictor and several different disruption detection systems operational in the JET real-time event detection platform named PETRA (Plasma Events Triggering for Alarms). The CM predictor outperforms all the other classifiers implemented in PETRA, according to all the main statistical indicators normally used to qualify these tools.