252Cf source-correlated transmission measurements and genetic programming for nuclear safeguards

One of the main targets of nuclear safeguards is to determine the mass and enrichment of fissile samples enclosed in special, non-accessible containers. In this paper, we present a method to estimate the mass of uranium oxide samples based on 252Cf source-driven noise-analysis measurements. We show that the mass of the samples can be successfully predicted using a genetic programming algorithm. The input presented to the algorithm was in the form of features extracted from the physical properties of the measured correlation functions.