Machine learning for reactor power monitoring with limited labeled data

IF 1.5 3区 物理与天体物理 Q3 INSTRUMENTS & INSTRUMENTATION
C.L. Stewart , B.L. Goldblum , R.G. Abbott , L. Appleby , B.J. Borghetti , V. Hollingshead , J.H. Whetzel
{"title":"Machine learning for reactor power monitoring with limited labeled data","authors":"C.L. Stewart ,&nbsp;B.L. Goldblum ,&nbsp;R.G. Abbott ,&nbsp;L. Appleby ,&nbsp;B.J. Borghetti ,&nbsp;V. Hollingshead ,&nbsp;J.H. Whetzel","doi":"10.1016/j.nima.2025.170285","DOIUrl":null,"url":null,"abstract":"<div><div>Real-time reactor power monitoring is critical for a variety of nuclear applications, spanning safety, security, operations, and maintenance. While machine learning methods have shown promise in monitoring reactor power levels, there is limited research on their efficacy in label-starved environments. The goal of this work is to assess the feasibility of classifying nuclear reactor power level using multisource data in scenarios with limited labels. Data were collected using low-resolution multisensors at four nuclear reactor facilities: two large research reactors and two TRIGA reactors. Within each pair, one reactor dataset served as the source and the other as the target in a transfer learning paradigm. Twenty-three supervised models were trained on labeled sequences of magnetic field and acceleration data from each of the target sites. Self-learning and transfer learning methods were applied to the top performing models to assess their classification performance with increasing amounts of labeled data. While reactor power level classification was achieved with a Matthews Correlation Coefficient of up to 0.739 ± 0.003 and 0.622 ± 0.009 with only 400 sequences per power state for the large research reactor and TRIGA target sites, respectively, self-learning and transfer learning leveraging source site data did not improve target classification performance. These findings suggest that alternative methods, such as higher sensitivity sensors, digital twins, or the use of physics-informed models, are required to enable high-performance classification in machine learning approaches to reactor monitoring with a dearth of target ground truth.</div></div>","PeriodicalId":19359,"journal":{"name":"Nuclear Instruments & Methods in Physics Research Section A-accelerators Spectrometers Detectors and Associated Equipment","volume":"1073 ","pages":"Article 170285"},"PeriodicalIF":1.5000,"publicationDate":"2025-02-10","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":"0","resultStr":null,"platform":"Semanticscholar","paperid":null,"PeriodicalName":"Nuclear Instruments & Methods in Physics Research Section A-accelerators Spectrometers Detectors and Associated Equipment","FirstCategoryId":"101","ListUrlMain":"https://www.sciencedirect.com/science/article/pii/S0168900225000865","RegionNum":3,"RegionCategory":"物理与天体物理","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"","PubModel":"","JCR":"Q3","JCRName":"INSTRUMENTS & INSTRUMENTATION","Score":null,"Total":0}
引用次数: 0

Abstract

Real-time reactor power monitoring is critical for a variety of nuclear applications, spanning safety, security, operations, and maintenance. While machine learning methods have shown promise in monitoring reactor power levels, there is limited research on their efficacy in label-starved environments. The goal of this work is to assess the feasibility of classifying nuclear reactor power level using multisource data in scenarios with limited labels. Data were collected using low-resolution multisensors at four nuclear reactor facilities: two large research reactors and two TRIGA reactors. Within each pair, one reactor dataset served as the source and the other as the target in a transfer learning paradigm. Twenty-three supervised models were trained on labeled sequences of magnetic field and acceleration data from each of the target sites. Self-learning and transfer learning methods were applied to the top performing models to assess their classification performance with increasing amounts of labeled data. While reactor power level classification was achieved with a Matthews Correlation Coefficient of up to 0.739 ± 0.003 and 0.622 ± 0.009 with only 400 sequences per power state for the large research reactor and TRIGA target sites, respectively, self-learning and transfer learning leveraging source site data did not improve target classification performance. These findings suggest that alternative methods, such as higher sensitivity sensors, digital twins, or the use of physics-informed models, are required to enable high-performance classification in machine learning approaches to reactor monitoring with a dearth of target ground truth.
求助全文
约1分钟内获得全文 求助全文
来源期刊
CiteScore
3.20
自引率
21.40%
发文量
787
审稿时长
1 months
期刊介绍: Section A of Nuclear Instruments and Methods in Physics Research publishes papers on design, manufacturing and performance of scientific instruments with an emphasis on large scale facilities. This includes the development of particle accelerators, ion sources, beam transport systems and target arrangements as well as the use of secondary phenomena such as synchrotron radiation and free electron lasers. It also includes all types of instrumentation for the detection and spectrometry of radiations from high energy processes and nuclear decays, as well as instrumentation for experiments at nuclear reactors. Specialized electronics for nuclear and other types of spectrometry as well as computerization of measurements and control systems in this area also find their place in the A section. Theoretical as well as experimental papers are accepted.
×
引用
GB/T 7714-2015
复制
MLA
复制
APA
复制
导出至
BibTeX EndNote RefMan NoteFirst NoteExpress
×
提示
您的信息不完整,为了账户安全,请先补充。
现在去补充
×
提示
您因"违规操作"
具体请查看互助需知
我知道了
×
提示
确定
请完成安全验证×
copy
已复制链接
快去分享给好友吧!
我知道了
右上角分享
点击右上角分享
0
联系我们:info@booksci.cn Book学术提供免费学术资源搜索服务,方便国内外学者检索中英文文献。致力于提供最便捷和优质的服务体验。 Copyright © 2023 布克学术 All rights reserved.
京ICP备2023020795号-1
ghs 京公网安备 11010802042870号
Book学术文献互助
Book学术文献互助群
群 号:481959085
Book学术官方微信