An explicit off-situ inversion method for neutron fields in reactor core based on graph structure

摘要: In the future, advanced reactors will face harsh core environments. Using off-situ detection signals to invert the neutron distribution in the core is an important means to ensure the safe operation of the core. The latest data-driven off-situ inversion algorithms do not fully consider the correlations among various regions or assembly segments of the reactor core, resulting in the problem of limited accuracy reduction when performing off-situ inversion of the high-resolution neutron field in reactor core. For this purpose, in this study an explicit off-situ inversion model of the neutron field in the reactor core based on graph structure is proposed. By structurally modeling the attribute information of each discretized assembly segment within the reactor core and the relationship information between the assembly segments, the interactions and correlations among assembly segments of the reactor core are explored to improve the inversion accuracy of the neutron distribution in the reactor core in high dimensions. The inversion results in the Chinese lead-based research reactor (CLEAR-I) show that, compared to the previous MCRNet and DETNet framework, when a single region of the core variations, the maximum relative deviation decreases from 21.8% to 14.01%. Among them, the number of inversion units with a relative deviation greater than or equal to 10% within 5. In the case of multi-region variations in the core, the maximum relative deviation decreased from 28.08% to about 20%. Among them, the proportion of relative deviations greater than or equal to 10% decreased from 3.14% to 1.77%. It can be seen from this that explicit structured modeling for the reactor core plays a certain role in improving the accuracy of the off-situ inversion calculation of the reactor core.