October 22, 2024
Journal Article

Inverse prediction of PuO2 processing conditions using Bayesian seemingly unrelated regression with functional data

Abstract

Over the past decade, various innovative methodologies have been developed to better characterize the relationships between processing conditions and the physical, morphological, and chemical features of special nuclear materials (SNMs). Different processing conditions generate SNM products with unique features known as “signatures,” which indicate the processing conditions used to produce those materials. These signatures can potentially allow forensic analysts to determine the processes used to produce the SNMs and make inferences about where the materials originated. A statistical technique that relates processing conditions to the morphological features of PuO2 particles was investigated in this study. A Bayesian implementation of seemingly unrelated regression (SUR) was developed to inversely predict unknown PuO2 processing conditions from known PuO2 features. Model results from simulated data demonstrated the usefulness of this technique. When applied to empirical data from a bench-scale experiment specifically designed for inverse prediction, the model successfully predicted nitric acid concentration, while the results for Pu concentration and precipitation temperature were equivalent to those of a simple mean model. The proposed technique complements other recent methodologies developed for forensic analyses of nuclear materials and can be generalized across the field of chemometrics for applications to other materials.

Published: October 22, 2024

Citation

McCombs A.L., M.A. Stricklin, K. Goode, G. Huerta, S.W. Kurtis, J.D. Tucker, and A. Zhang, et al. 2024. Inverse prediction of PuO2 processing conditions using Bayesian seemingly unrelated regression with functional data. Frontiers in Nuclear Engineering 3, no. _:Art. No. 1331349. PNNL-SA-198443. doi:10.3389/fnuen.2024.1331349