Tracking of individual TRISO-fueled pebbles through the application of X-ray imaging with deep metric learning

Abstract

Modern pebble-bed reactor concepts utilizing TRISO-fueled pebbles use on-line continuous refueling, where fuel pebbles are continuously circulated through the reactor core. Presently, no method exists for the tagging, identification, and tracking of individual TRISO-fueled pebbles as they enter and exit the reactor core. This leaves room for improvement in the nuclear material accountability and nuclear safeguards of TRISO-fueled pebbles. This work presents a methodology to identify individual TRISO-fueled pebbles by exploiting the unique distribution of the TRISO-coated particles, which is imprinted during the manufacturing process, within individual TRISO-fueled pebbles. By combining X-ray imaging and deep learning, our method learns a mapping from radiographic images to a compact Euclidean space where distances provide a direct measurement of the similarity of pebble radiographs. A deep convolutional neural network is trained to optimize the image mapping and triplet loss is implemented to enforce a greater distance between mappings that identify different fuel pebbles. A dataset consisting of 1,250 radiographic Monte Carlo N-Particle (MCNP) Transport simulations of unique TRISO-fueled pebbles is generated for training and testing the deep learning algorithm, which achieves an accuracy of 93.49% ± 9.35% and by first transforming the dataset with Gaussian blur transformations it achieves an accuracy of 98.70% ± 2.60%.