In the quest for clean, unlimited energy, scientists have made a monumental leap forward. A collaborative team from Princeton University and the Princeton Plasma Physics Laboratory (PPPL) has successfully harnessed artificial intelligence (AI) to predict and prevent disruptive plasma instabilities in nuclear fusion processes.
The research team contends that their discoveries underscore the potential for utilizing AI-driven dynamic control to address additional plasma instabilities hindering the attainment of sustained magnetic confinement fusion. Detailed in a press release from the Princeton Plasma Physics Laboratory (PPPL) on February 21, their findings were subsequently published in Nature on February 22.
Spearheaded by Egemen Kolemen, an associate professor of mechanical and aerospace engineering at Princeton’s Andlinger Center for Energy and the Environment, and a staff research physicist at PPPL, the study received support from the Department of Energy’s Office of Fusion Energy Sciences, as well as the National Research Foundation of Korea.
“Previous studies have generally focused on either suppressing or mitigating the effects of these tearing instabilities after they occur in the plasma,” remarked Jaemin Seo, the primary author and an assistant professor of physics at Chung-Ang University in South Korea. He conducted a significant portion of the research during his tenure as a postdoctoral researcher in Kolemen’s team. “But our approach allows us to predict and avoid those instabilities before they ever appear,” Seo added.
Tearing mode instabilities represent disruptions in plasma wherein the integrity of magnetic field lines is compromised, potentially facilitating plasma escape. Leveraging data derived from prior experiments conducted at the DIII-D tokamak, the Princeton team developed a “neural network” capable of forecasting the likelihood of future tearing instabilities based on real-time plasma attributes.
Subsequently, they employed this neural network to train a reinforcement learning algorithm, which was tasked with evaluating tactics for managing plasma within a simulated setting. Through iterative testing and refinement, the algorithm gleaned insights into effective strategies for mitigating the instabilities, thus enhancing control over plasma dynamics.
According to PPPL, the team aims to replicate their experimentation with the AI controller at the DIII-D tokamak and subsequently generalize its functionality for other tokamaks. There are aspirations for future research endeavors to extend the algorithm’s capabilities to address diverse plasma control challenges simultaneously.
Jaemin Seo expressed, “We have strong evidence that the controller works quite well at DIII-D, but we need more data to show that it can work in a number of different situations. We want to work toward something more universal.”
Relevant articles:
– AI can predict and prevent fusion plasma instabilities in milliseconds, American Nuclear Society
– Amber Mac on LinkedIn: AI solves nuclear fusion puzzle for near, linkedin.com
– AI solves huge problem holding back fusion power, Freethink