Energy

DeepMind's AI helps confine plasma for nuclear fusion research

DeepMind's AI helps confine pl...
Plasma inside the tokamak at the SPC
Plasma inside the tokamak at the SPC
View 3 Images
Plasma inside the tokamak at the SPC
1/3
Plasma inside the tokamak at the SPC
A 3D model of the tokamak reactor at the SPC
2/3
A 3D model of the tokamak reactor at the SPC
DeepMind's AI was used to produce advanced plasma configurations
3/3
DeepMind's AI was used to produce advanced plasma configurations
View gallery - 3 images

Across decades of development, scientists have been pursuing nuclear fusion technology through endless experiments, calculations and simulations, trying to find the optimal mix of conditions for atoms to fuse and continually release huge amounts of energy. The Alphabet-owned company DeepMind has now lent its considerable artificial intelligence know-how to the effort through a new partnership with Ecole Polytechnique Federale de Lausanne's (EPFL's) Swiss Plasma Center (SPC), where it has already proved its worth.

DeepMind has been making some impressive moves in the world of artificial intelligence over the past few years, beating the world's best players at Go, predicting rainfall with a high-degree of accuracy and even solving a 50-year scientific problem by predicting the 3D structures of unique proteins.

In harnessing the technology for nuclear fusion research, scientists hope to come up with ways to more successfully sustain streams of plasma, enabling more opportunities for critical fusion reactions to occur. The type of device used for these experiments at the SPC is known as a tokamak, which is a donut-shaped chamber that uses a powerful magnetic field to contain streams of super-hot plasma, in which hydrogen atoms fuse into one helium atom and release energy.

A 3D model of the tokamak reactor at the SPC
A 3D model of the tokamak reactor at the SPC

The SPC's tokamak is known as a variable-condition tokamak (TCV), in that it allows for experiments using plasma in different kinds of configurations. Researchers here are continually experimenting with new ways to control the plasma, so that it doesn't crash into the vessel walls and collapse.

"Our simulator is based on more than 20 years of research and is updated continuously," said Federico Felici, an SPC scientist. "But even so, lengthy calculations are still needed to determine the right value for each variable in the control system. That's where our joint research project with DeepMind comes in."

DeepMind developed a new AI algorithm that was trained on the SPC's simulator by having it attempt many different control strategies. In time, as it gained more experience through the simulations, the algorithm was able to calculate control strategies for producing requested plasma configurations. The team then tasked the algorithm to work in reverse, identifying the correct settings to generate a specific plasma configuration.

After training, the algorithm was tested on the real-world tokamak, where it was able to create and control a wide range of plasma shapes, including elongated and advanced shapes such as "negative triangularity" and "snowflake" configurations. One experiment involved maintaining two separate plasmas simultaneously.

DeepMind's AI was used to produce advanced plasma configurations
DeepMind's AI was used to produce advanced plasma configurations

"Our team’s mission is to research a new generation of AI systems – closed-loop controllers – that can learn in complex dynamic environments completely from scratch," said Martin Riedmiller, control team lead at DeepMind. "Controlling a fusion plasma in the real world offers fantastic, albeit extremely challenging and complex, opportunities.”

The research was published in the journal Nature.

Source: EPFL

View gallery - 3 images
9 comments
9 comments
vince
What we need is a tool to curb the unrealistic promises made by the fusion groups around the world over 75 years ago. It's a pie-in-the-eye Ponzi scheme to get research funds for personal gain. There will never be commercial fusion power plants at least not in the 21st century or even the 22nd century.
Douglas Rogers
Instead of solving a puzzle for currency, maybe we could make ASICs to solve important problems like this for currency!
akarp
Using quantum computers to replace simple calculations seems strange...but I wonder if controlling a fusion reactor is like a plausible use case for quantum computers.
Phaedrus
Fusion has proved to be a tough nut to crack, at least sustainable forms that might generate surplus power. But the potential is so incredible that it make sense to keep at it even if success is ultimately centuries away.
MarkGovers
I wonder if the Plasma various blowouts could also be studied by the AI and potentially reversed mid-stream. In other words, could the plasma be saved and reset with the correct on-the-fly setting or settings? Think of the wind and a candle, with the quick move of our hands, we can block the wind and save the light, although many times more difficult with Plasma and it would be beyond human possibility, but ought to be in range for AI and Quantum Computers.
michael_dowling
Gotta agree with vince. I was a fan of fusion generators at first,but then read an article written by a physicist who spent his career in this field. He has a negative view of hot fusion,to put it mildly.Even if plasma stability can be tamed,hot fusion generates fast neutrons,which would make the reactor vessel intensely radioactive,and makes the steel of the reactor brittle: https://thebulletin.org/2017/04/fusion-reactors-not-what-theyre-cracked-up-to-be/
Jeek
Fusion was 50 years away 50 years ago and is still 50 years away from whenever *now* is.
christopher
Shudder: AI is the tech you use when you are too lazy to work out the math yourself... and it never understands any edge cases, usually catastrophically, and it produces "incomprehensible" results (nothing exists to test that AI "answers" are correct and working form the correct data. For example: self-driving AI that can be fooled by simple stickers on signs, etc).

They need to stop working with stuff they do not understand (AI) and do better to properly understand what they're actually supposed to be working on (the reactions themselves).
Rustgecko
I find it interesting all the commentators who are saying that Fusion research is a waste of money. They sound suspiciously like the people that said that heavier than air machines would never fly one hundred years ago, or that human transplants would never work, or that rockets to the moon would never work, or that mobile phones would never work. Unlike some of those things we know that fusion is possible because it happens all the time elsewhere.