Science
Fusion Energy News
Independent intelligence on the global fusion industry
Tuesday, July 21, 2026
Vol. III · Edition · Web
Science · med impact
Princeton PPPL demonstrates ML-based disruption prediction at 95% accuracy
Convolutional network trained on DIII-D and JET data predicts plasma disruptions 50 ms in advance.
Researchers at the Princeton Plasma Physics Laboratory published a paper in <i>Nature Energy</i> demonstrating a convolutional neural network that predicts tokamak plasma disruptions with 95% accuracy 50 milliseconds in advance. The model was trained on a combined DIII-D and JET dataset.
Disruption prediction is critical for reactor-scale tokamaks, where uncontrolled plasma collapse can cause damaging electromagnetic loads on the first wall. Lead author William Tang said the technique is portable across machines — a property earlier ML approaches lacked.
Disruption prediction is critical for reactor-scale tokamaks, where uncontrolled plasma collapse can cause damaging electromagnetic loads on the first wall.
Weekly newsletter
Fusion Energy Weekly
The week in fusion: breakthroughs, companies, and capital — in your inbox. Free, every Monday.
Primary sources
Editorial standards: Fusion Energy News dispatches are compiled from primary filings, peer-reviewed papers, and on-the-record statements. Corrections: corrections@fusionenergynews.com · public log
More on Science
Letters to the editor(0)
Sign in to write a letterNo letters yet. Be the first to write one.