Princeton researchers have developed a machine learning method to suppress harmful edge instabilities in fusion reactors without sacrificing plasma performance. The approach, which optimizes the system’s suppression response in real-time, demonstrated high fusion performance without edge bursts at two different facilities. Achieving sustained fusion requires maintaining a high-performing plasma that is dense, hot, and confined long enough for fusion to occur.
However, as researchers push the limits, they encounter challenges like bursts of energy escaping from the edge of super-hot plasma. These bursts negatively impact performance and can damage reactor components over time. One traditional fix involves using magnetic coils to apply fields to the plasma edge, breaking up structures that could develop into instabilities.
While this stabilizes the plasma, it typically lowers overall performance. “We have a way to control these instabilities, but in turn, we’ve had to sacrifice performance, which is one of the main motivations for operating in the high-confinement mode in the first place,” said Egemen Kolemen, associate professor at Princeton.
Optimizing fusion reactor performance
The team’s machine learning approach slashes computation time from tens of seconds to the millisecond scale, enabling real-time optimization. The model can monitor plasma status from one millisecond to the next and adjust magnetic perturbations as needed, balancing edge burst suppression and high fusion performance. “With our machine learning surrogate model, we reduced the calculation time of a code that we wanted to use by orders of magnitude,” said co-first author Ricardo Shousha, a postdoctoral researcher.
The researchers demonstrated success at both the KSTAR tokamak in South Korea and the DIII-D tokamak in San Diego, achieving strong confinement and high fusion performance without harmful edge bursts. They are working to refine the model for compatibility with other fusion devices, including future reactors like ITER. Kolemen noted the potential for AI to overcome longstanding bottlenecks in developing fusion power as a clean energy resource.
“These machine learning approaches have unlocked new ways of approaching these well-known fusion challenges,” he said. The findings were reported on May 11 in Nature Communications. The research received support from the U.S. Department of Energy, the National Research Foundation of Korea, and the Korea Institute of Fusion Energy.
