Akademik Veri Yönetim Sistemi
International Conference on Nuclear Security: Shaping the Future, IAEA CN-321/EVT2300037, Vienna, Avusturya, 20 - 24 Mayıs 2024, sa.17
In recent years, in parallel to innovative nuclear security detection technologies with using integrated information systems, active interrogation(x-ray, neutron) and passive measurement techniques are used for a rapid detection of radioactive or nuclear materials during their illicit trafficking for nuclear security purposes. As a primary inspection, many PVT or PS-based radiation portal monitors(RPMs) were deployed at seaports and land borders worldwide , and recently spectroscopic radiation portal monitors(SRPMs) are also used to monitor more accurately the transported goods. However, as a secondary inspection for nuclear security purposes, hand-held, portable radionuclide identification devices(RIDs) based on NaI, LaBr, CeBr scintillators or CdZnTe and HPGe semiconductors are preferably used to facilitate the trade activities in situ applications. Any RID devices needs to use at least one algorithm for automated isotope identification(ID). The ID algorithms used in RIDs can be classified as Library Comparison Method, Region of Interest (ROI) Method, Template Matching Method and Expert Interaction Method. In parallel to advancement radiation detector technologies, especially “medium energy resolution” detectors such as LaBr, CeBr, CdZnTe detectors are more promising for in-situ applications. Therefore there is a need to develop innovative and adaptive or intelligent ID algorithms to extract more accurate ID for the radionuclide, especially from discrimination close lying peaks in the gamma-spectrum due to their limited energy resolutions. In this paper, new emerging ID algorithms was briefly addressed for RIDs used in nuclear security. it was underlined the Artificial Intelligence (AI) tools and Machine Learning (ML), Artificial Neural Networks (ANN) methods which are increasingly used in detection and identification of radionuclides for nuclear security. Specifically, it was mentioned about the implementation of some ID algorithms based on Fuzzy Logic(FL) approach, Bayesian Statistics(BS) peak based approach, Nonlinear Least Squares(NLS) peak deconvolution with Figure-out-Merit(FOM) Criteria for medium energy resolution detectors used in nuclear security