A 3-D Neutron Distribution Reconstruction Method Based on the Off-Situ Measurement for Reactor

Understanding the core neutron distribution is an indispensable condition to ensure the safe operation and effective burn-up of reactor fuel. It has become one of the new ways for reactor monitoring to use ex-core monitoring signals to calculate the in-core information recently. In this article, a fully convolutional network (FCN)-based architecture was implemented, where the 2-D and 3-D fully convolutional neural networks were used to reconstruct the space- and energy-dependent 3-D neutron distribution of the reactor core through the off-situ measurement directly. The new method has been validated in the H. B. Robinson reactor. The differences between reconstruction results of neutron flux distribution and Monte Carlo (MC) calculation were mostly less than 2%. The average relative deviation of reconstructing the neutron density was kept at 5%. It was demonstrated that the provided method was applicable to monitor the reactor core theoretically and would provide strong support for the subsequent research on the real-time monitoring for an operating reactor, safety, and operational testing for new reactor designs, and evaluation of source term information in a nuclear emergency.