Abstract
In an effort to expand the capabilities of safeguards actors to verify the integrity of fresh fuel assemblies, Oak Ridge National Laboratory has retrofitting the existing electronics of the JCC-71 uranium neutron collar, which contains 18 3He neutron detectors and an external AmLi neutron source arranged to surround a fuel assembly. The new electronics system allows analysts to record list-mode neutron multiplicity data in addition to the singles and doubles rates that are currently measured. Based on previous proof-of-concept research, analysis of these new data will identify off-normal fuel configurations in an assembly and characterize or localize the specific partial fuel defects.
To analyze the complex list-mode data collected with the modified instrument, multivariate classification algorithms are being developed. A novel classification method, the relevance vector machine, has been developed. This novel approach may be applied to polychotomous problems to produce estimate the probability test data belongs to each possible class of data. In addition, our method identifies the most useful variables/channels for making predictions, which informs the basis for the model’s predictions, and this interpretability is largely unique among data analytics methods. Variable selection occurs during model training an parameter tuning, and does not need any external hyperparameter tuning routines.
Finally, we apply the modified relevance vector machine to a dataset of list-mode neutron collar data simulated with MCNP. The method is able to correctly identify off-normal fuel configurations and categorize the data according the four fuel defect scenarios. Our analysis ranked the channels in the data according to prediction utility; however, the logic applied in calculating the coincidence data in this initial dataset hinders physical interpretation of the signatures identified by our classifier. Future work will amend this shortcoming of our existing data simulation method, as well as expand the range of fuel defect scenarios in our analysis.
