Scientists have used artificial intelligence to overcome the huge challenge of producing nearly limitless clean energy through nuclear fusion.
A team from Princeton University in the US has found a way to use AI models to predict and prevent plasma instability during fusion reactions.
Nuclear fusion has been hailed as the “holy grail” of clean energy because of its potential to generate vast amounts of energy without the need for fossil fuels or leaving behind harmful waste.
This process mimics the same natural reactions that occur within the sun, but harnessing fusion energy has proven extremely difficult.
In 2022, a team at California's Lawrence Livermore National Laboratory achieved the first-ever net energy gain from nuclear fusion. This means that more energy was produced than was put into the reaction.
Although it was only a small amount, enough to bring a kettle to boil, it was a major milestone towards achieving it on a grand scale.
The success is another important milestone: the AI can now recognize plasma instabilities 300 milliseconds before they occur, enough time to make corrections to keep the plasma under control. It means that you have overcome an obstacle.
The new understanding could lead to grid-scale deployment of fusion energy, researchers say.
“By learning from past experiments, rather than incorporating information from physics-based models, AI can develop final control policies that support stable, high-power plasma conditions in real-time in real reactors.” said study leader Egemen Colemen. , he works as a physicist at the Princeton Plasma Physics Laboratory, where groundbreaking discoveries were made.
The latest research has been published in a scientific journal Nature It was published Wednesday in a paper titled “Avoiding fusion plasma tearing instability with deep reinforcement learning.”
“Predicting instability in advance could make it easier to perform these reactions than with current approaches, which are more passive,” said SanKyeun Kim, co-author of the study. .
“There is no longer a need to wait for instability to occur and take quick corrective action before the plasma is disrupted.”