Rotation-invariant rapid TRISO-fueled pebble identification based on feature matching and point cloud registration

Abstract

The Pebble-bed reactor (PBR) utilizing TRISO-fueled pebbles is a promising Gen-IV reactor design due to its inherent safety and thermal efficiency. Identification of pebbles exiting the core may provide IAEA with additional information enhancing PBR safeguards help validate software predicting pebble flow behavior and burnup, and optimize fuel usage strategies should granular fuel management be needed. We propose a rapid, accurate pebble identification method applicable to commercial PBRs. Our approach relies on the unique spatial distribution of TRISO particles associated with each pebble, which can be extracted through X-ray CT. We developed an algorithm to identify a single pebble from 100,000 pebbles and achieved 100% identification accuracy in 90,000 trials, with arbitrary rotations and high noise levels. Remarkably, the outermost TRISO particle positions within a 0.6-mm shell suffice for unique identification. With an average identification time of under 50s per pebble, our method is compatible with PBR operational demands.