Material Identification of Bodies Stored in Nuclear Waste Drums using Muon Scattering Tomography and Machine Learning - 21102

Muon Scattering Tomography (MST) is a non-destructive assay technique for the characterization of sealed heterogeneous nuclear waste packages. Using MST in combination with machine learning techniques allows for a greater understanding of a waste drum's contents. Here we describe a method that uses multivariate analysis classifiers in combination with MST data to identify objects stored in a waste drum and determine their most likely material composition. We test our method through simulation studies using a generic MST detector system and establish that a wide range of objects can be correctly identified after a 10-day exposure. We also determine the system's efficiency at detecting small uranium objects as 0.90_-0.12^+0.07. (authors)