Study on autonomous search for multiple radioactive leakage sources based on updated infotaxis in nuclear emergency rescue

Abstract

Nuclear facilities face leakage risks from natural hazards, human errors, or external attacks, often generating multi-point radioactive leakage sources that produce large-scale dynamic radiation plumes through atmospheric dispersion and multi-source superposition. Unlike orphan source recovery operations (e.g., retrieving displaced or poorly shielded sealed radioactive sources in localized fields), nuclear emergencies require urgent identification of leakage points to enable real-time leakage sources suppression. Based on the Daya Bay nuclear power plant scenario, this study proposes a multi-source radiation leakage inversion model based on an updated infotaxis algorithm, which incorporates the information entropy of superimposed radiation fields from multiple sources. The search path of the mobile detector is optimized by integrating a movement strategy activation function to adjust subsequent positions. Simulation results demonstrate that the hexagonal path unit enhances search efficiency by 21.78% compared to traditional quadrilateral path units. In a scenario involving three radioactive leakage sources, the mobile detector successfully identifies all sources locations through exhaustive grid sampling, achieving an average positioning error of 5.73 m. This approach provides a novel perspective for identifying multiple radioactive leakage sources in nuclear accidents.