Design of a deep learning surrogate model for the prediction of FHR design parameters

A. J. Whyte, Z. Xing, G. T. Parks, E. Shwageraus

Research output: Chapter in Book/Report/Conference proceedingConference contributionpeer-review

1 Scopus citations

Abstract

Following previous work by Xing and Shwageraus, a large corpus of data has been collected for simulated AGR-style fuel assembly design in FHRs. The results exhibit a nonlinear system response, so a ‘deep’ multi-layer perceptron surrogate model is designed and tested for prediction of design parameters. This neuro-surrogate regression model could be useful for the fast optimization of the design parameters, for example in multiobjective optimization problems, due to the extremely fast evaluation time. Source code is made available for the audit and authentication of the scientific method.

Original languageEnglish
Title of host publicationInternational Conference on Mathematics and Computational Methods Applied to Nuclear Science and Engineering, M and C 2019
PublisherAmerican Nuclear Society
Pages298-307
Number of pages10
ISBN (Electronic)9780894487699
StatePublished - 1 Jan 2019
Externally publishedYes
Event2019 International Conference on Mathematics and Computational Methods Applied to Nuclear Science and Engineering, M and C 2019 - Portland, United States
Duration: 25 Aug 201929 Aug 2019

Publication series

NameInternational Conference on Mathematics and Computational Methods Applied to Nuclear Science and Engineering, M and C 2019

Conference

Conference2019 International Conference on Mathematics and Computational Methods Applied to Nuclear Science and Engineering, M and C 2019
Country/TerritoryUnited States
CityPortland
Period25/08/1929/08/19

Keywords

  • AGR
  • Deep learning
  • FHR
  • Fuel design
  • Neural network
  • Surrogate model

ASJC Scopus subject areas

  • Applied Mathematics
  • Nuclear Energy and Engineering

Fingerprint

Dive into the research topics of 'Design of a deep learning surrogate model for the prediction of FHR design parameters'. Together they form a unique fingerprint.

Cite this