AI controls plasma inside a nuclear fusion reactor

Technology News |
By Rich Pell

While nuclear fusion represents a promising path towards sustainable energy, a core challenge to achieving successful and sustained nuclear fusion is to shape and maintain an inherently unstable high-temperature plasma – which is hotter than the core of the sun – within a vessel. Typically scientists are using magnetic confinement, in particular in the tokamak configuration – a doughnut-shaped vacuum surrounded by magnetic coils – to recreate these extreme conditions.

However, to shape and maintain a high-temperature plasma within the tokamak vessel is a complex challenge, requiring high-dimensional, high-frequency, closed-loop control using magnetic actuator coils, which is further complicated by the diverse requirements across a wide range of plasma configurations. For example, say the researchers, a control system needs to coordinate the tokamak’s many magnetic coils and adjust the voltage on them thousands of times per second to ensure the plasma never touches the walls of the vessel, which would result in heat loss and possibly damage.

To help solve this problem, the researchers collaborated with scientists at the Swiss Plasma Center at EPFL to develop the first deep reinforcement learning (RL) system to autonomously discover how to control these coils and successfully contain the plasma in a tokamak.

“Using a learning architecture that combines deep RL and a simulated environment,” say the researchers, “we produced controllers that can both keep the plasma steady and be used to accurately sculpt it into different shapes. This ‘plasma sculpting’ shows the RL system has successfully controlled the superheated matter and – importantly – allows scientists to investigate how the plasma reacts under different conditions, improving our understanding of fusion reactors.”

The researchers say they successfully produced and controlled a diverse set of plasma configurations on the Tokamak à Configuration Variable – literally “variable configuration tokamak” – including elongated, conventional shapes, as well as advanced configurations, such as negative triangularity and “snowflake” configurations. Their approach, say the researchers, achieves accurate tracking of the location, current and shape for these configurations.

The researchers also demonstrate sustained “droplets” on TCV, in which two separate plasmas are maintained simultaneously within the vessel. This, say the researchers, represents a notable advance for tokamak feedback control, showing the potential of reinforcement learning to accelerate research in the fusion domain, and is one of the most challenging real-world systems to which reinforcement learning has been applied.

“Similar to progress we’ve seen when applying AI to other scientific domains,” say the researchers, “our successful demonstration of tokamak control shows the power of AI to accelerate and assist fusion science, and we expect increasing sophistication in the use of AI going forward. This capability of autonomously creating controllers could be used to design new kinds of tokamaks while simultaneously designing their controllers.”

For more, see “Magnetic control of tokamak plasmas through deep reinforcement learning.”



Linked Articles