Progress and Challenges of AI Applications in Nuclear Power: A Closer Look at the Future
Abstract: As artificial intelligence and machine learning (AI/ML) become increasingly prevalent and new applications emerge, integrating these technologies into nuclear engineering research and education holds great promise for empowering our students. However, it also presents unique challenges related to explainability, trustworthiness, and deployment of such technologies. In this seminar, Prof. Majdi Radaideh will explore the current state-of-the-art of AI/ML in nuclear engineering, illustrating how these technologies can transform our approach to nuclear computing, modeling, and simulation. Dr. Radaideh's research group at the University of Michigan focuses on the intersection of multiphysics reactor design, advanced computing leveraging AI/ML, and autonomous control. The seminar will be structured into three segments. Initially, Dr. Radaideh will touch upon his previous work in AI/ML and its integration with nuclear reactor design. Subsequently, he will delve deeper into three ongoing projects within his lab: employing reinforcement learning to optimize reactor design, utilizing generative AI to gauge public sentiment on nuclear power via social media analysis, and employing AutoML for establishing digital twins and quantifying their uncertainty. In the final segment, Dr. Radaideh will explore potential research gaps and challenges, including interpretable AI, deployment of AI on engineering systems, and the role of AI in expediting multiphysics simulations. Collaboration on these challenges can be pursued with faculty at the University of Illinois at Urbana-Champaign.
Bio: Prof. Majdi Radaideh (RAD) started as a tenure-track Assistant Professor at the University of Michigan in January 2023. Prof. RAD leads the AIMS lab (Artificial Intelligence and Multiphysics Simulations), which has grown quickly to comprise 2 postdocs, 6 PhD students, and 3 MS students. AIMS focuses on the intersection between nuclear reactor design, multiphysics modeling and simulation, advanced computational methods, and machine learning algorithms to drive advanced reactor research and improve the sustainability of the current reactor fleet. Prof. RAD has extensive skills in the development and usage of nuclear codes, programming experience, uncertainty quantification, software engineering, and machine learning algorithms. He completed his B.Sc. in nuclear engineering from the Jordan University of Science and Technology, M.S. and Ph.D. in nuclear engineering from the University of Illinois at Urbana Champaign with two minors in computational science & engineering and applied statistics. After graduation, he held different R&D appointments at MIT and ORNL, where he completed his postdoctoral studies and collaborated with different national labs and industrial partners. In the past, Prof. RAD has collaborated and worked on 5+ DOE/NEUP/IRP projects, and he is currently the PI and Co-PI on six different funded projects from DOE-NE, U.S. NRC, national labs, and University of Michigan seed grants. In addition, he has received several awards for his work including: 2023 DOE-NE Distinguished Early Career Award, 2021 MIT outstanding postdoctoral service award, 2019 ANS National Mark Mills Award for the best PhD thesis work in nuclear engineering, Best Paper Award in the Best Estimate Plus Uncertainty Conference (BEPU-2018), and several others. To date, Prof. RAD has authored 80+ research publications including 35+ journal papers, where he was the first/major author in 30+ journal papers.
