Answers are grounded in this paper and cite their sources.
Study predicts distribution ratios in solvent extraction, suggesting improved models for reprocessing nuclear fuel.
Solvent extraction is the key technique for uranium–plutonium separation in nuclear fuel reprocessing. However, the quantitative partitioning of fission products in multi-component system remains unclear due to limited experimental data. This makes it difficult to develop multi-component extraction models involving key elements U, Pu and interfering fission products. Accordingly, machine learning (ML) has become an effective method to capture multi-element interactions across diverse systems. In this study, experimental data for Re(Ⅶ), Tc(Ⅶ), Th(Ⅳ), and U(Ⅵ) in the nitric acid–TBP system were collected. Three ML models, including Random Forest (RF), Support Vector Regression (SVR), and K-Nearest Neighbors (KNN), were developed to predict distribution ratios under various conditions. To clarify the underlying mechanisms, SHapley Additive exPlanations (SHAP) analysis was integrated to identify key influencing factors and their interaction effects. The optimal coefficient of determination ( R 2 ) values for RF, SVR, and KNN reach 0.9646, 0.9920, and 0.9654, respectively, indicating higher prediction accuracy and robustness than conventional empirical formulas. Furthermore, SHAP analysis reveals competitive extraction behavior between fission elements (Re, Tc) and Th, providing a new perspective for understanding the microscopic mechanisms of the solvent extraction.
No takes yet. Share an insight, caveat, or question.
Wu et al. (2026) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: