Date/Time
Date(s) - 02/23/2023
1:55 pm - 2:55 pm
Location
Rhines 125
Categories
Daniel Cole, Ph.D.
Associate Professor
University of Pittsburgh
Dr. Dan Cole is an Associate Professor in the Department of Mechanical Engineering and Materials Science in the Swanson School of Engineering at the University of Pittsburgh. He received his B.S. (1991), M.S. (1992), and Ph.D. (1998) in Mechanical Engineering from Virginia Tech.
His research has focused on advanced control‐theoretic techniques to improve operation, maintenance, and control of nuclear power plants. This work has focused on methods of condition monitoring of plant components, and has used AI and machine learning for decision making and asset management of the low‐volume, high‐value systems in power plant. This work on nuclear power has led his research to the security of our critical infrastructure, in particular, our networked industrial control systems.
In these efforts, Dr. Cole is investigating how to choose strategies to protect our infrastructure, determine avenues of attack, detect and mitigate threats, and secure cyber‐physical systems. In this highly interdisciplinary field, he is working to create opportunities for industry‐university collaborations with Pitt and area stakeholders.
Abstract
Competing sources of electricity are able to minimize operation and maintenance (O&M) costs by eliminating scheduled inspections. They are able to do predictive maintenance, enabled by both better data collection and data use.
Predictive maintenance has reduced costs and improved efficiency in other industries, and AI has the potential to do the same for nuclear power industry. We discuss the use of AI to improve O&M in nuclear power plants.
First, we examine the condition monitoring of systems to assess the health of these elements and hazard models that give an estimate of remaining useful life. Second, efficient asset management requires knowing the availability of resources to plan maintenance and reduce downtime. Supply chains can be better monitored to ensure that parts and materials are available when they are needed. Third, AI seems encouraging for decision making for asset management; however, there are practicalities about how the industry functions that make this challenging, and we discuss practical concerns and the complexity of the problem.
Finally, we discuss three practicalities of concern: the regulatory environment and explainable AI used for critical decision making; rare failures leading to imbalanced datasets; and the constraints imposed by how the industry operates.