Inversion of nuclear accident source terms combining Bayesian method with machine learning

IF 1.9 3区工程技术 Q1 NUCLEAR SCIENCE & TECHNOLOGY

Annals of Nuclear Energy Pub Date : 2025-03-30 DOI:10.1016/j.anucene.2025.111418

Zhonghao Li , Bo Cao , Dingping Peng , Qingyue You , Xuewei Miao , Zhouliang Chen

{"title":"Inversion of nuclear accident source terms combining Bayesian method with machine learning","authors":"Zhonghao Li , Bo Cao , Dingping Peng , Qingyue You , Xuewei Miao , Zhouliang Chen","doi":"10.1016/j.anucene.2025.111418","DOIUrl":null,"url":null,"abstract":"<div><div>Source term inversion plays a critical role in consequence assessment during nuclear accidents. This study investigates three Bayesian approaches for machine learning-based source term inversion, with datasets generated by the radioactive nuclide atmospheric dispersion program RADC. Firstly, a Bayesian neural network (BNN) was designed using Python to predict the release rate of I-131 during nuclear accidents. The BNN achieved a mean absolute percentage error (MAPE) of 7.92%, showing significantly higher accuracy and robustness compared to a backpropagation neural network (BPNN) with an identical structure and under the same accident scenarios. Secondly, both BNN and BPNN are further enhanced through Bayesian optimization, achieving a significantly improved MAPE values of 5.78% and 3.80% respectively. Furthermore, The Monte Carlo Dropout method was used to generate confidence intervals for the BPNN, offering an uncertainty analysis approach for conventional neural networks in source term inversion. However, its uncertainty analysis showed greater fluctuations compared to BNN.</div></div>","PeriodicalId":8006,"journal":{"name":"Annals of Nuclear Energy","volume":"218 ","pages":"Article 111418"},"PeriodicalIF":1.9000,"publicationDate":"2025-03-30","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":"0","resultStr":null,"platform":"Semanticscholar","paperid":null,"PeriodicalName":"Annals of Nuclear Energy","FirstCategoryId":"5","ListUrlMain":"https://www.sciencedirect.com/science/article/pii/S030645492500235X","RegionNum":3,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"","PubModel":"","JCR":"Q1","JCRName":"NUCLEAR SCIENCE & TECHNOLOGY","Score":null,"Total":0}

引用次数: 0

Abstract

Source term inversion plays a critical role in consequence assessment during nuclear accidents. This study investigates three Bayesian approaches for machine learning-based source term inversion, with datasets generated by the radioactive nuclide atmospheric dispersion program RADC. Firstly, a Bayesian neural network (BNN) was designed using Python to predict the release rate of I-131 during nuclear accidents. The BNN achieved a mean absolute percentage error (MAPE) of 7.92%, showing significantly higher accuracy and robustness compared to a backpropagation neural network (BPNN) with an identical structure and under the same accident scenarios. Secondly, both BNN and BPNN are further enhanced through Bayesian optimization, achieving a significantly improved MAPE values of 5.78% and 3.80% respectively. Furthermore, The Monte Carlo Dropout method was used to generate confidence intervals for the BPNN, offering an uncertainty analysis approach for conventional neural networks in source term inversion. However, its uncertainty analysis showed greater fluctuations compared to BNN.

查看原文本刊更多论文

求助全文

约1分钟内获得全文求助全文

来源期刊

Annals of Nuclear Energy 工程技术-核科学技术

CiteScore

4.30

自引率

21.10%

发文量

632

审稿时长

7.3 months

期刊介绍： Annals of Nuclear Energy provides an international medium for the communication of original research, ideas and developments in all areas of the field of nuclear energy science and technology. Its scope embraces nuclear fuel reserves, fuel cycles and cost, materials, processing, system and component technology (fission only), design and optimization, direct conversion of nuclear energy sources, environmental control, reactor physics, heat transfer and fluid dynamics, structural analysis, fuel management, future developments, nuclear fuel and safety, nuclear aerosol, neutron physics, computer technology (both software and hardware), risk assessment, radioactive waste disposal and reactor thermal hydraulics. Papers submitted to Annals need to demonstrate a clear link to nuclear power generation/nuclear engineering. Papers which deal with pure nuclear physics, pure health physics, imaging, or attenuation and shielding properties of concretes and various geological materials are not within the scope of the journal. Also, papers that deal with policy or economics are not within the scope of the journal.