AI Model Lays Groundwork for More Reliable Fusion Energy by Predicting and Preventing Plasma Disruptions
-
Princeton researchers developed an AI model to predict and prevent plasma instabilities in fusion reactors, enabling real-time control for more reliable fusion energy.
-
The AI model, trained on past experimental data, predicted tearing mode instabilities up to 300ms in advance, allowing time to change parameters to avoid plasma disruptions.
-
The approach opens the door for more dynamic control of fusion reactions than current passive approaches and provides a foundation for using AI to solve plasma instabilities.
-
Researchers used simulated data to train a reinforcement learning algorithm on how to maintain high power while avoiding instabilities before testing in a real tokamak.
-
The proof-of-concept demonstrates AI's potential to effectively control fusion reactions, with plans to expand the controller's capabilities and apply it to more fusion reactors.