San Diego, CA (February 21) - In a significant stride towards sustainable energy, a team led by Princeton University has harnessed the power of artificial intelligence to control plasma instabilities in fusion reactors. Researchers at the DIII-D National Fusion Facility demonstrated that AI can predict and prevent plasma disruptions, a major hurdle in fusion energy development.
Fusion energy, which powers the sun, holds the promise of a clean and virtually limitless energy source. However, replicating this process on Earth requires containing superheated plasma within powerful magnetic fields. These plasmas are prone to instabilities that can halt the fusion reaction.
The AI model, trained on past experimental data, can forecast potential plasma instabilities known as tearing mode instabilities up to 300 milliseconds in advance. This allows the AI controller to adjust operational parameters, preventing disruptions to the plasma's magnetic field lines and maintaining a stable reaction.
"By learning from past experiments, rather than incorporating information from physics-based models, the AI was able to develop a final control policy that sustained a stable and powerful plasma regime in real-time, in a real reactor," said research leader Egemen Kolemen.
This breakthrough paves the way for more dynamic control of fusion reactions and the use of AI to address a range of plasma instabilities. The team's findings, published in Nature, represent a significant step towards achieving sustained fusion reactions and realizing the potential of fusion energy.