Optimizing Neutron Moderation Absorption Using Artificial Neural Network

Abstract Neutron interactions with matter can be either scattering or absorption reactions. Scattering can result in a change in the energy and direction of motion of a neutron but cannot directly cause the disappearance of a free neutron. Absorption leads to the disappearance of free neutrons as a result of a nuclear reaction with fission or the formation of a new nucleus and another particle or particles such as protons alpha particles and gamma photons. Most materials have an absorption cross section that varies inverse with neutron velocity. Artificial neural network technique would be a perfect tool to determine the impact parameters from the experimental observables. One has only to train the network by theoretical simulations and then to feed the trained net work with experimental observables in order to obtain the impact parameter as the output of the network. Feed-forward networks have proven to be valuable tools for data analysis classification of events particle identification function approximation pattern recognition. The advantages of feed-forward nets are the highly parallel algorithm the flexibility because of their trainability the capability to solve high-dimensional problems and the deterministic behavior.