Characterization of cyber events in a nuclear system using AI/ML
Abstract: There is increased interest from the nuclear industry and nuclear engineering community to explore the applicability of AI/ML in the nuclear domain. A successfully implementation of AI/ML in the nuclear realm could provide tangible benefits to stakeholders. For example, AI/ML could enhance the detection of operational anomalies, predict system failures before they occur, and optimize maintenance schedules, thereby reducing the risk of accidents and ensuring adherence to strict safety standards. In this presentation, we will discuss and summarize the methodology, implementation, performance evaluation, and lessons learned of an experimental and computational effort to assess the feasibility of AI/ML technologies to characterize cyber events in a nuclear system and to test the main hypothesis of a recent research project: “AI/ML can be feasibly and usefully applied to characterize system states resulting from cyber events.”
Bio: Stylianos Chatzidakis is an Assistant Professor in the School of Nuclear Engineering at Purdue University. He also serves as Associate Reactor Director for PUR-1 and Director of Radiation Laboratories. Prior to joining Purdue, he was R&D Staff and Weinberg Distinguished Fellow at ORNL. Dr. Chatzidakis’ research focuses in the fields of digital twins, AI/ML, quantum-based cybersecurity for advanced reactors, and anomaly detection using intelligent algorithms. He holds a Senior Reactor Operator License from the NRC. He is one of the developers of the PUR-1 digital twin, the Tube Acoustic and Pressure measurement System (TAPS) and the Mobile Examination and Remediation Facility (MERF). Dr. Chatzidakis is a recipient of several research and teaching awards including the Weinberg Distinguished Fellowship at ORNL, the Roy G. Post Foundation Scholarship, the Bilsland Dissertation Fellowship, and Purdue’s Outstanding Engineering Teacher Award.
