Year
2023
File Attachment
finalpaper_177_0427023652.pdf127.06 KB
Abstract
The National Nuclear Security Administration (DOE/NNSA) Office of Defense Nuclear
Nonproliferation Research and Development (DNN R&D) has a vision for developing technology
that will provide unobtrusive surveillance with instantaneous accountability regarding the
monitoring of nuclear material and nuclear material plant operations. Monitoring and
characterization of nuclear facilities is an essential activity to meet the goals of the nuclear
nonproliferation community. New developments in machine learning and cognitive inferencing
have potential to greatly assist in delivering real-time monitoring capabilities and help in
optimizing plant operations. Operating such automated monitoring systems requires good
knowledge management, ability to dynamically update information, quantification of uncertainty
in measurements, and superior quality interpretation of machine learning and artificial intelligence
algorithm outputs. This paper will introduce basic concepts regarding how research teams have
been able to establish analytical techniques that couple observations of diverse physical
phenomena with subject matter experts (SME) knowledge that will inform data collection and how
to draw meaningful conclusions about nuclear activities.