A solution method for neutron transport in reactors with uniform distribution of materials

Abstract

A method for solving neutron transport for reactors with uniform distribution of materials that combines machine learning and Monte Carlo methods. The first step is to construct a series of reactors with a uniform distribution of materials, each containing a different material composition or density. These reactors were then solved using the Monte Carlo method, and the neutron scattering and fission probabilities and the characteristic distribution of neutrons were calculated after the simulation. Using these probability and feature distribution data, a fully-connected neural network model is trained that takes neutron energy and material composition as input and the feature distribution of scattered and fission neutrons as output. Finally, using the obtained fully connected neural network model, the neutrons and materials produced by fission are used as inputs to obtain the next generation of fission neutrons, and the calculation is iterative until convergence. Compared with the existing Monte Carlo method, the method no longer needs to perform a large number of sampling processes, and the calculation speed is faster, and the neutron transport calculation can be quickly completed for a reactor with uniform distribution of materials.

Description

technical field [0001] The invention relates to the field of nuclear reactor core design and safety, in particular to a neutron transport solution method for a reactor with uniform distribution of materials. Background technique [0002] The numerical simulation of the reactor is closely related to the design, operation and safety of the reactor. The continuous development of the nuclear industry has put forward higher requirements for the accuracy and efficiency of the numerical simulation of the core. The solution of the neutron transport equation is the core of the numerical simulation of the reactor. The current main solution methods are divided into determinism and Monte Carlo method (Mont Carlo method for short). [0003] The deterministic method solves the simplified neutron transport equation after discretizing energy, space and angle. This method is relatively mature, but inevitably introduces computational bias in the discrete process of each variable, and is not ...

AI Technical Summary

Problems solved by technology

Compared with the deterministic method, the advantages of the Monte Carlo method are: the geometric versatility is strong, and various complex geometric structures can be described; the point section of continuous energy can be used to process various complex energy spectra, avoiding tedious combination and resonance Processing process; as long as enough particles and simulation times are guaranteed, good calculation accuracy can be obtained; although the Monte Carlo method has many advantages, the simulation process requires repeated sampling of a large number of particles, which is still quite time-consuming even if parallel computing is used. The sampling process will also generate a large amount of particle data, which occupies a large amount of memory and has a low utilization rate, which also affects the improvement of computing efficiency.

Images