Unfolding Plastic Detector Responses to White and Mono-energetic Neutrons Based on Artificial Neural Networks

Abstract

Neutron spectrum unfolding based on an artificial neural network (ANN) is a highly flexible and robust method that can be applied to any type of radiation detector and wide energy range of incident particles. In this study, we present details of neutron-spectrum deconvolution with well-established multilayer perceptron algorithms implemented in CERN ROOT with the aim of obtaining the incident neutron energy spectra for white neutron and mono-energetic beams with finite energy spreads. The ANN trained with experimental and simulation datasets successfully approximated incident neutron spectra with high accuracy for each case, indicating that a well-trained ANN has high potential for applications in radiation-related fields.