Year
2000
Abstract
Nuclear materials safeguard necessitates the use of non-destructive methods to determine the attributes of fissile samples enclosed in special, non-accessible containers. To this end, a large variety of methods has been developed at the Oak Ridge National Laboratory (ORNL) and elsewhere.1,2 Active non-destructive assay evolved based on the use of an interrogation source emitting neutrons and gamma rays, whose scope is that of inducing fission in the fissile material within the sample. Usually, a given set of statistics of the stochastic neutron-photon coupled field, such as source-detector, detector-detector cross correlation functions, and multiplicities are measured over a range of known samples to develop calibration algorithms. In this manner, the attributes of unknown samples can be inferred by the use of the calibration results. The goal of this paper is to develop an artificial intelligence approach to this problem whereby neural networks (NNs) and genetic programming (GP) algorithms are used for sample identification purposes. To this end, a number of Monte Carlo simulations were performed to obtain source-detector time-dependent cross correlation functions for a set of uranium metal samples of different shapes, masses, and enrichments. A number of features were identified and extracted from the cross correlation functions and effectively related to the sample’s mass and enrichment.