Sign in Subscribe
Periodical publishing

Machine Learning Algorithms Improve Nuclear Material Identification

Strategic Insights

— 1 min read

Researchers from the University of Florida have developed a machine learning framework to classify and identify special nuclear materials using scintillation detectors. This study highlights the challenges in discriminating between different levels of enrichments, weak radiation signals, and complex self-shielding effects. The researchers used supervised learning algorithms, including Random Forest, XGBoost, and a feedforward Deep Neural Network, to achieve an accuracy of over 95% in identifying special nuclear materials. The study demonstrates the potential of machine learning in nuclear safety and security applications.

Key Takeaways:

  • The researchers used three scintillation detectors, an EJ-309 liquid scintillator, a CLYC crystal scintillator, and an EJ-276 plastic scintillator, to measure gamma-ray and neutron data from special nuclear material at the National Criticality Experiments Research Center (NCERC) at the National Nuclear Security Site (NNSS), at Nevada, USA.
  • The radiation detector pulse data was extracted from the collected digitized data and applied to three separate supervised learning models, achieving an accuracy of over 95% in identifying special nuclear materials.
  • The Countrate parameter feature, which is the overall gamma-ray and neutron incidents for each detector, was found to be the most influential parameter and essential to include for improved classification.
  • The initial model versions not including the Countrate parameter feature failed to classify special nuclear materials.
  • The machine learning development framework developed in this study will be beneficial for future applications in discriminating between different fuel enrichments and additives such as burnable poisons.

Statistics:

  • Accuracy of over 95% achieved in identifying special nuclear materials using the machine learning framework.
  • Three scintillation detectors used to measure gamma-ray and neutron data from special nuclear material.
  • Countrate parameter feature accounted for over 90% of the classification accuracy.

Sources:

  • VerticalNews journalists, "Studies from University of Florida in the Area of Machine Learning Described (Applying Machine Learning Algorithms to Classify Digitized Special Nuclear Material Obtained from Scintillation Detectors)", Journal of Engineering, 2025, p 4726.
  • Applying Machine Learning Algorithms to Classify Digitized Special Nuclear Material Obtained from Scintillation Detectors. Journal of Nuclear Engineering, 2025,6(3):31.
  • MDPI AG, Publisher of Journal of Nuclear Engineering.

Read more