Research on Adaptive Algorithm for Measuring Nuclear Material Holdup by Gamma Spectroscopy Method

Accurate quantification of nuclear material holdup is essential for nuclear safeguards, radiation protection, and criticality safety. The conventional Generalized Geometry Holdup (GGH) method, while widely used with gamma spectroscopy, requires manual classification of deposits into point, line, or area sources—a subjective and error-prone step that hinders automation. To overcome this, the fundamental innovation of this work is a novel algorithmic architecture that fuses image recognition with the GGH framework. The process begins by applying an improved Canny edge detector to thermal maps of the holdup distribution to extract source geometries. An intelligent system then automatically classifies these geometries and matches them to the appropriate correction model from a pre-defined library, fully automating the workflow and removing human subjectivity. Building upon this core architecture, a secondary yet vital innovation is introduced: an adaptive correction logic for sedimentation thickness. This module addresses the significant bias in self-attenuation factors that arises with thicker deposits. By comparing preliminary results against the theoretical model's asymptotic behavior, the algorithm estimates the effective source thickness and dynamically refines the correction factor. Validation demonstrates that this integrated approach effectively handles diverse source morphologies and thickness variations, consistently achieving a calculation error of less than 10%. The proposed method thus offers a robust, automated solution with strong potential for practical deployment in nuclear facilities.