Avoiding fusion plasma tearing instability with deep reinforcement learning
Avoiding fusion plasma tearing instability with deep reinforcement learning
www.nature.com Avoiding fusion plasma tearing instability with deep reinforcement learning - Nature
Artificial intelligence control is used to avoid the emergence of disruptive tearing instabilities in the magnetically confined fusion plasma in the DIII-D tokamak reactor.
1
comments
As some sources have titled it: "AI fixes fusion!" /s
On a more serious note, I find it interesting that a 25ms control loop is enough, and that:
the tearing prediction model could forecast the instability 300 ms before the disruption
Good model.