Development of LASSO based optimized scheme for reconstructing radioactive source distributions using monitoring air dose rates

Shi, W.*; Machida, Masahiko  ; Yamada, Susumu  ; Okamoto, Koji*

Clarifying the distribution of radioactive sources within nuclear facilities is crucial for ensuring worker safety during decommissioning and for responding to accidents. However, air dose rate measurements in restricted areas are often limited due to complex structures and high radiation levels in contaminated rooms. To address this, we have proposed a machine learning-based approach, the Least Absolute Shrinkage and Selection Operator (LASSO), to reconstruct radioactive source distributions in simplified room models. LASSO method indicates the good performance of reconstructing radioactive source with high accuracy inside simple room model. However, in more complex environments, obstacles can degrade reconstruction accuracy. To overcome these limitations, we developed an optimized scheme based on the LASSO method to improve inverse estimation in complex rooms. In this scheme, the impact of shielding structures is mitigated by normalizing the radioactive contributions from sources. A series of numerical simulations demonstrate that the optimized approach outperforms the non-optimized version in accurately reconstructing source distributions. Furthermore, experiments in a room with complex structures validate the effectiveness of the optimized method. The inverse estimations performed on experimental data confirm that the use of a normalized contribution matrix significantly improves accuracy by reducing the influence of shielding. Conclusively, this paper optimizes LASSO scheme for reconstructing radioactive source distributions in complex building room using air dose rate measurements. It shows significant improvements over existing scheme and is verified to be successfully applied in complicated situations with high accuracy. We confirm that optimized LASSO scheme holds significant promise for future monitoring and decommissioning projects in both operational and damaged nuclear facilities.