Rapidly Tracking the Arrival of Fusion Energy Using AI and Accessibility

2023-09-04

MIT and other institutions are accelerating the acquisition of fusion energy at the Newton Center for Fusion Energy through AI and accessibility. MIT and other research institutions will receive support from the US Department of Energy (DoE) to improve the acquisition of fusion data and increase diversity in the field of fusion.

As the impact of climate change continues to grow, there is increasing interest in fusion as a clean energy source. Despite ongoing laboratory research on fusion reactions since the 1930s, scientists still need to answer many key questions to make fusion energy a reality, and time is of the essence. As part of the strategy to accelerate fusion energy research and achieve carbon neutrality by 2050, the US Department of Energy announced new funding support for projects by four collaborating institutions, including MIT.

Cristina Rea, a research scientist and group leader at MIT and other research institutions, will serve as the principal investigator for this new three-year collaborative project funded by the new grant. The project will pilot the integration of fusion data into a system that can be read by AI tools. MIT and other research institutions will collaborate with scientists from Williams College, University of Wisconsin-Madison, Auburn University, and the non-profit organization HDF Group to create a comprehensive fusion data platform that provides unprecedented access for researchers, especially for underrepresented groups of students. The project aims to encourage diverse participation in fusion and data science in academia and the workplace through an outreach program led by four out of five female co-investigators.

The funding from the US Department of Energy is intended to support the team's efforts to distribute data generated by fusion devices, such as MIT's toroidal confinement fusion device Alcator C-Mod. Alcator C-Mod operated from 1991 to 2016, and its data is still being studied, in part thanks to MIT's commitment to open knowledge exchange.

Currently, there are nearly 50 public experimental magnetic confinement fusion devices, but accessing the historical and current data from these devices is challenging. Some fusion databases require user agreements, and not all data is cataloged and organized in the same way. Additionally, utilizing machine learning and other AI tools for data analysis and scientific discoveries is difficult due to the time-consuming process of reorganizing the data. As a result, there are fewer scientists involved in fusion research, and the barriers to discoveries are greater, creating a bottleneck in utilizing AI to accelerate progress.

The proposed data platform of this project aims to address these technical barriers by adhering to the "Findable, Accessible, Interoperable, and Reusable" (FAIR) principles and following the recommendations of the UNESCO on Open Science (OS) to improve transparency and inclusivity in science. All deliverables from researchers will comply with the FAIR and OS principles, as required by the US Department of Energy. The database of this platform will be built using MDSplusML, an upgraded version of the MDSplus open-source software developed by MIT researchers in the 1980s for categorizing Alcator C-Mod experimental results. Today, nearly 40 fusion research institutions use MDSplus to store and provide external access to their fusion data. The release of MDSplusML aims to continue this tradition of open collaboration.

The researchers intend to address the barriers to female and underrepresented group participation by improving general access to fusion data and by hosting a three-year summer school at Williams College that focuses on the intersection of fusion and machine learning.

Regarding the importance of their research, Rea said, "This project is in response to the needs of the fusion community to be prepared for success. Accessibility is absolutely crucial to achieving scientific progress in fusion through interdisciplinary collaboration and communication. I think we all understand now that diverse communities have more diverse ideas and can solve problems more quickly."

The work of this collaborative project also aligns with the important research areas identified in the "AI for Fusion" coordinated research project by the International Atomic Energy Agency. Rea has been selected as the technical coordinator for the International Atomic Energy Agency's coordinated research project, with a focus on accelerating fusion research and development through community engagement and knowledge acquisition. After writing a letter of support for the proposed project by this team, the International Atomic Energy Agency stated, "The work that the researchers will carry out will not only benefit our coordinated research project but also the entire international fusion community."

Dennis Whyte, Director of the MIT Plasma Science and Fusion Center and Hitachi America Professor, said, "I am thrilled that MIT and our partners are at the forefront of applying new AI tools and simultaneously encouraging and facilitating the ability to extract critical data from our experiments."

Rea stated, "It is very meaningful for me to have the opportunity to lead such an important project, and I feel a responsibility to demonstrate that women can be leaders in STEM fields. We have an incredible team that is motivated to improve our fusion ecosystem and achieve the goal of fusion energy."

RECENT AI NEWS

RECENT AI TOOLS