Building trust in AI for nuclear power applications

Smart technology is found almost everywhere. Recreational devices, houseware appliances, transportation, healthcare and agriculture; all these have benefited from technology that collects, tracks and analyzes information to react to situations and deliver better results.

One sector that hasn’t used artificial intelligence (AI) to its fullest extent is nuclear power because questions remain about its trustworthiness. But a newly launched IAEA collaborating center at Purdue will unify experts determined to answer those questions and build the necessary trust in using AI for nuclear power applications.

IAEA Collaborating Center at Purdue

The International Atomic Energy Agency (IAEA) has named Purdue’s Center for Science of Information (CSOI) its first IAEA Collaborating Center on using artificial intelligence for nuclear power. The four-year agreement begins immediately.

The genesis of the IAEA was a 1953 address by President Dwight Eisenhower to the United Nations General Assembly. The agency polices almost 200 member states and manages research projects around the world, supported by its collaborating centers, to help achieve the United Nations’ nuclear development goals.

Hany Abdel-Khalik, professor in the School of Nuclear Engineering, will oversee all IAEA Collaborating Center activities at Purdue, including coordinating projects with other member states worldwide. Wojciech Szpankowski, the Saul Rosen Distinguished Professor of Computer Science in the College of Science, will serve as its director. The Collaborating Center at CSOI will advance new frontiers of nuclear technology through modeling and simulations, validating AI concepts for nuclear technologies, and training and education.

Hany Abdel-Khalik

“This Collaborating Center will help build confidence in AI applications for high-consequence systems, such as nuclear reactors,” said Abdel-Khalik. “Without reliable quantification, the nuclear community's ability to realize the potential of AI will be diminished and this will negatively impact its ability to remain competitive in the energy market.”

AI answers the question of automation in nuclear power

Abdel-Khalik said there has been a resurgence of interest to use nuclear power as a safe, reliable energy source for electric and non-electric applications. It is accompanied by many concepts for nuclear energy production and storage, which is expected to heavily rely on digitization for both sensing and predictive analysis.

He said other engineering communities have promoted the continuous fusion of analysis and sensing results by other names like science-based engineering, virtual engineering and digital twinning.

“A key requirement for this vision is the ability to automate these various functions to reduce human errors and to respond to the massive amounts of data from both sensing and analysis,” Abdel-Khalik said. “This is a critical requirement because the state-of-the-art digital platforms have allowed for data availability rates that are unprecedentedly high, making it infeasible to adopt a manual or expert-based approach for automation.”

Abdel-Khalik said AI is a natural answer to automation needs.

“This is because AI has unquestionably demonstrated super-human performance on several, albeit narrow-focused, applications, e.g., image classification, voice, face, and speech recognition, board games, etc., resulting in wider adoption over the past several decades and allowing for improved performance by reducing human errors and cost of human labor,” he said.

“While the nuclear sector is energized to adopt the AI technology, it is approaching this goal in a guarded manner with the stakeholders’ perspective ranging anywhere from allowing AI to be in the driver’s seat, i.e., making autonomous decisions, to being a mere passenger that is aiding human operators.”

A key challenge in this automation is to understand and explore trustfulness which is related to the ability to quantify the value of information from sensing and analysis. This is one of the main topics that has been studied by the NSF-sponsored Center for Science of Information, Szpankowski said.

Purdue expertise benefits IAEA

The IAEA’s goals for the Collaborating Center align with the mission of the CSOI: to advance science and technology through a new quantitative understanding of the representation, communication and processing of information in biological, physical, social and engineered systems.

The CSOI meets its mission through achieving the following goals:

  • Integrative research: Creating a shared intellectual space, integral to the center's activities, providing a collaborative research environment that crosses disciplinary and institutional boundaries.
  • Education, human resources and diversity: Integrating cutting-edge, multidisciplinary research and education efforts across the center to advance the training and diversity of the workforce.
  • Knowledge transfer: Developing effective mechanisms for interactions between the center and external stakeholders to support the exchange of knowledge, data and the application of new technologies.
  • Leadership and management: Accomplishing the center's mission through inspirational leadership, inclusive and transparent decision-making, catalyzing new research opportunities and facilitating collaborative efforts.
  • Ethics: Implementing a multidisciplinary and multi-institutional program to inform and guide all members of the center on the ethical and responsible conduct of scientific research.

Abdel-Khalik said the CSOI, in its role as an IAEA Collaborating Center, will focus on:

  • Supporting advancements and innovations in nuclear power and contributing to the development of technical and educational publications.
  • Jointly working with IAEA on key enabling technologies relevant to advancing nuclear industry, e.g., reactor modelling/simulation, automation, nuclear digital twin technologies, nuclear safety analysis, accident evolution prediction and severe accident management guidelines using AI.
  • Contributing to the IAEA mission of safe, secure and peaceful application of nuclear technology by training next generation of aspirants in nuclear industry.
  • Conducting cutting-edge research in high- and low-fidelity simulation models using AI, as well as in its applications to automation, in-service inspection, evaluation and characterization of structure, system and component defects, real-time risk monitoring, reactor designing and predictive maintenance.
  • Supporting IAEA professionals on implementing and incorporating AI for decision making.
  • Hosting researchers and fellows, IAEA MSCFP interns and IAEA LMP fellows.

How Purdue benefits from the center

Establishing the IAEA collaborating center at Purdue will elevate research visibility, positioning the university as a leader in AI advancements for the nuclear industry. It will facilitate international collaborations, enriching research with diverse perspectives on the value, applications and opportunities for nuclear power worldwide.

It also will help attract funding from government and industry, addressing key goals of the nuclear technology in the 21st Century, focusing on accelerating the deployment of advanced reactors and ensuring optimal extension of the life of existing reactors.

Students will benefit from hands-on experience, preparing them for careers in AI and fostering industry partnerships that offer internships and job opportunities.

Overall, the center promises to enhance academic prominence, set trends in AI nuclear research and contribute significantly to both theoretical advancements and real-world impact in the nuclear sectors.