AI Model Learns to Prevent Plasma Tears in Nuclear Fusion Reactors
-
Researchers at Princeton developed an AI model that can predict and avoid plasma tearing instabilities in nuclear fusion reactors. This could help achieve stable, high-powered fusion reactions.
-
The AI was trained on data from previous fusion experiments to learn how to maintain high power output while avoiding destabilizing plasma tears.
-
The model successfully predicted tearing instabilities 300 milliseconds in advance during real-world tests, giving it time to adjust reactor parameters to prevent tears.
-
This addresses a major challenge for fusion energy, but other obstacles like containment still need to be overcome before it becomes viable.
-
The AI controller performed well in tests, but more data is needed to confirm it works reliably in different fusion reactor scenarios.