Fast localization of radiation sources based on Support Vector Machine

BackgroundLost radioactive sources needs to be quickly retrieved, positioning of radioactive source in complex environment is the key to find the lost radioactive source. [Propose] This study aims to develope a novel approach for the rapid positioning of orphan sources using a NaI(Tl) array detection device.MethodFirst of all, by leveraging the shadow effect between array detectors, a response curve between gamma-ray incidence angles and counts was obtained through the use of Monte Carlo simulation software. Then, the support vector machine (SVM) method was employed to establish a predictive mathematical model for the counting rate of array detectors as a function of gamma-ray incidence angle, utilizing. Finally, a radioactive source localization physical experiment platform was constructed, and a series of incidence angle response experiments were conducted for the validation of this approach applied to radioactive source localization under varying conditions.ResultsEexperimental results demonstrate that, through the use of the SVM regression prediction model, the maximum average deviation of the angle is 9.21° whilst the minimum is 1.77° for the angle prediction of an orphan 137Cs point source.ConclusionsThis method can achieve rapid and accurate localization of an orphan radioactive source.