ANN-MoC Method for Solving Unidimensional Neutral Particle Transport Problems

TL;DR

This paper introduces the ANN-MoC method, combining artificial neural networks with the Method of Characteristics, to efficiently solve unidimensional neutral particle transport problems, demonstrating promising results in test cases.

Contribution

The paper presents a novel ANN-MoC approach that integrates neural networks with classical transport methods, offering a new way to estimate particle fluxes after convergence.

Findings

Successful application to test cases

Potential for improved accuracy and efficiency

Demonstrates feasibility of neural network integration

Abstract

Neutral particle transport problems are fundamental in the modeling of energy transfer by radiation (photons) and by neutrons with many important applications. In this work, the novel ANN-MoC method for solving unidimensional neutral particle transport problems is presented. Following the Method of Discrete Ordinates (DOM) and decoupling with a Source Iteration (SI) scheme, the proposed method applies Artificial Neural Networks (ANNs) together with the Method of Characteristics (MoC) to solve the transport problem. Once the SI scheme converges, the method gives an ANN that estimates the average flux of particles at any points in the computational domain. Details of the proposed method are given and results for two test cases are discussed. The achieve results show the potential of this novel approach for solving neutral particle transport problems.