Study on the sequential Bayesian approach for a CsI(Tl)-based radioactive material detection system

Abstract

Traditional algorithms based on energy spectra are widely used to detect radioactive materials in the field of nuclear security. However, it takes a long time to collect photons to obtain accurate energy spectra, which limits the warning time and increases false-alarm probability. The sequential Bayesian algorithm, which takes both photon energies and their interarrival time into account, could help improve the performance of radioactive detection. In this study, we propose a theoretical optimization method of the traditional sequential Bayesian algorithm to suit the parameters of the CsI(Tl) detector. Based on simulation data, dynamic environmental background parameters and repeated tests were added to the algorithm to further optimize its performance. Finally, with the improved sequential Bayesian algorithm, radioactive sources can be quickly detected by the CsI(Tl) detector with low false-alarm probability (less than 1.1 times per day).