Use of thermal neutron analysis in conjunction with neural networks for detection of explosives

Full text follows: Thermal neutron activation analysis is based on neutron capture prompt gamma ray analysis and has been used in a wide variety of fields, for example, inspection of checked airline baggage and detection of buried land mines. In all of these applications, the detected gamma ray intensities from the elements present are used to estimate their concentrations. The irradiated samples with thermal neutrons emit characteristic rays which are the fingerprint of each element constituent. This work describes a study of the application the a trained neural network to determine the presence of explosives, using the characteristic spectra (fingerprint) of the substances as patterns which were simulated via a Monte Carlo N-Particle Transport code, version 4B (MCNP4B). After the training of the neural networks, it was possible to determine the presence of several types of explosives, even when they are occult for several materials. The code MCNP can be used for the simulation of the spectra produced by the irradiation of the samples of several types of explosives under different occultation. The neural networks, behaved as a powerful algorithm capable to recognize the patterns of the explosives, independently of their occultation s. Besides that, the network was capable to generalize, identifying the presence of explosives in cases where it had not been training. In that way, it was revealed as an excellent tool to be applied in inspection system in situ. Acknowledgements: The authors would like to thank to Fundacao de Amparo a Pesquisa do Estado do Rio de Janeiro (FAPERJ) and Conselho Nacional de Desenvolvimento Cientifico e Tecnologico (CNPq) for their financial support. (author)