A breakthrough in reactor physics: advanced neural networks reveal new potential in solving K eigenvalue problems
Newswise — One study in reactor physics is published in the journal Nuclear science and technology, researchers from Sichuan University and Shanghai Jiao Tong University introduced two innovative neural networks to address long-standing challenges associated with K eigenvalue problems in neutron scattering theory. These problems, fundamental in the field of nuclear engineering, are essential to the simulation and analysis of nuclear reactors.
This study introduced two pioneering neural networks, the Generalized Inverse Power Method Neural Network (GIPMNN) and its advanced version, the Physically Constrained GIPMNN (PC-GIPMNN), to address the challenges of reactor physics. While GIPMNN uses the inverse power method to iteratively identify the lowest eigenvalue and associated eigenvector, PC-GIPMNN elevates this approach by transparently incorporating conservative interface conditions. This advance proves crucial to meet the interface challenges inherent to reactors comprising varied fuel assemblies. Notably, in a side-by-side evaluation of performance on complex spatial geometries, PC-GIPMNN consistently outperformed its counterpart GIPMNN and others. Distinctively, this study opted for a data-agnostic approach, focusing only on mathematical and numerical solutions, thereby eliminating potential bias.
These discoveries herald a new era in nuclear reactor physics, paving the way for better understanding and more streamlined simulations. The adaptability of the introduced neural networks hints at their potential use in other scientific fields struggling with interface challenges. In essence, the study highlights the revolutionary promise of neural networks in reactor physics. Future efforts will undoubtedly refine these networks and test their effectiveness in increasingly complex scenarios.
###
The references
DO I
Original source URL
https://doi.org/10.1007/s41365-023-01313-0
Funding information
National Natural Science Foundation of China (11971020)
Shanghai Natural Science Foundation (23ZR1429300)
CNNC Innovation Fund (Lingchuang Fund)
About Nuclear science and technology
Nuclear science and technology (NST) reports on scientific discoveries, technical advances and important results in the fields of nuclear science and technology. The objective of this journal is to stimulate the mutual enrichment of knowledge between scientists and engineers working in the fields of nuclear research.
