Optimization and miniaturization of a thermal neutron radiography system using neutron tube technology and machine learning algorithms

This study focuses on the miniaturization technology of thermal neutron imaging systems based on D-T neutron tube sources. Through a controlled variable approach, the selection of moderating materials and the optimization design of moderating structures were systematically investigated, successfully achieving neutron beam performance metrics that meet the requirements for thermal neutron radiography. Building on this, an innovative imaging strategy was proposed, effectively integrating the high penetration characteristics of fast neutrons for thick samples with the high sensitivity of thermal neutrons for defect detection, thereby realizing the synergistic utilization of both neutron properties. Furthermore, this study introduces machine learning algorithms into the field, offering novel solutions for the intelligent development of non-destructive testing technologies. The research findings hold significant theoretical and practical value for the miniaturization and intelligent application of thermal neutron imaging technology.