Radiation Field Reconstruction Method Based on Non-uniform Source Activity Distribution Inversion

The accurate reconstruction of the gamma radiation field is the foundation for achieving digitalization of radiation protection and the prerequisite for conducting radiation dose assessment and visual simulation. In order to effectively solve the practical problem of large deviation in three-dimensional gamma radiation field reconstruction caused by high gradient non-uniform radiation source term distribution at nuclear facilities, this paper innovatively proposes a method for selecting non-uniform source activity distribution models based on Bayesian information criteria, and applies the optimal model to the reconstruction of three-dimensional gamma radiation fields. The results show that the accuracy of radiation field reconstruction using this method is significantly better than that of the Kriging interpolation method under high activity gradient and high dose rate gradient conditions; In the case of three-dimensional radiation field reconstruction in actual nuclear facilities, the average relative deviation between the dose rate reconstruction results of the test set and the measured values is only 12.69%, which is lower than the 85.40% obtained by Kriging interpolation. This study can provide advanced technical support for dynamic data-driven digital simulation of radiation protection in nuclear facilities, enhancing its effectiveness in practical applications.

γ辐射场（gamma radiation field）的精准重构是实现辐射防护数字化的基础，亦是开展辐射剂量评估与可视化模拟的必要前提。针对核设施（nuclear facilities）中辐射源项分布呈高梯度非均匀特征，进而引发三维γ辐射场重构偏差过大的实际难题，本文创新性提出一种基于贝叶斯信息准则（Bayesian Information Criteria）的非均匀源活度分布模型选择方法，并将最优模型应用于三维γ辐射场的重构工作。结果表明，在高活度梯度与高剂量率梯度条件下，本方法的辐射场重构精度显著优于克里金插值（Kriging Interpolation）法；在实际核设施的三维辐射场重构场景中，测试集的剂量率重构结果与实测值的平均相对偏差仅为12.69%，远低于克里金插值法得到的85.40%。本研究可为核设施辐射防护的动态数据驱动数字化模拟提供先进技术支撑，提升其实际应用效能。