Article(id=1147999688722215632, tenantId=1146029695717560320, journalId=1146123346816638986, issueId=1147999688122430098, articleNumber=1000-8063(2025)02-0029-09, orderNo=null, doi=10.13426/j.cnki.yky.2024.09.09, pmid=null, cstr=null, oa=null, hot=null, price=null, onlineType=0, articleFormat=0, articleType=null, articleTypeStr=null, receivedDate=1727625600000, receivedDateStr=2024-09-30, revisedDate=null, revisedDateStr=null, acceptedDate=null, acceptedDateStr=null, onlineDate=1751634054593, onlineDateStr=2025-07-04, pubDate=1747670400000, pubDateStr=2025-05-20, doiRegisterDate=null, doiRegisterDateStr=null, onlineIssueDate=1751634054593, onlineIssueDateStr=2025-07-04, onlineJustAcceptDate=null, onlineJustAcceptDateStr=null, onlineFirstDate=null, onlineFirstDateStr=null, sourceXml=null, magXml=null, createTime=1751634054593, creator=13701087609, updateTime=1751634054593, updator=13701087609, issue=Issue{id=1147999688122430098, tenantId=1146029695717560320, journalId=1146123346816638986, year='2025', volume='44', issue='2', pageStart='1', pageEnd='160', issueExtLink='null', onlineDate='null', pubDate='null', beforeIssueId=null, nextIssueId=null, price=null, status=1, issueComplete=1, articleOrder=1, issueType=-1, specialIssue=null, createTime=1751634054451, creator=13701087609, updateTime=1759123795578, updator=13701087609, preIssue=null, nextIssue=null, ext={EN=IssueExt(id=1179413939365491632, tenantId=1146029695717560320, journalId=1146123346816638986, issueId=1147999688122430098, language=EN, specialIssueTitle=, coverIllustrator=null, specialIssueEditor=, specialIssueAbout=), CN=IssueExt(id=1179413939365491633, tenantId=1146029695717560320, journalId=1146123346816638986, issueId=1147999688122430098, language=CN, specialIssueTitle=, coverIllustrator=null, specialIssueEditor=, specialIssueAbout=)}, issueFiles=null}, startPage=29, endPage=37, ext={EN=ArticleExt(id=1147999688936125168, articleId=1147999688722215632, tenantId=1146029695717560320, journalId=1146123346816638986, language=EN, title=Application of Physics-Informed Neural Networks in Solving the Advection-Diffusion Equation of Uranium Migration in Groundwater, columnId=1175805041752556213, journalTitle=Uranium Mining and Metallurgy, columnName=MINING AND HYDROMETALLURGY, runingTitle=null, highlight=null, articleAbstract=

In the in-situ leaching of uranium, uranium migration is influenced by both groundwater flow and solute diffusion, and this process can be effectively modeled using the advection-diffusion equation. Accurately modeling the variation of uranium concentration over time and space is crucial for predicting uranium migration in groundwater during in-situ leaching of uranium. Traditional numerical methods, such as the finite difference method, are computationally intensive and prone to errors while dealing with high-dimensional, complex problems. Therefore, this research aims to explore the applicability and accuracy of physics-informed neural networks (PINN) in solving the advection-diffusion equation. Through numerical simulations of the one-dimensional advection-diffusion equation, and by comparing the PINN solutions with numerical and analytical solutions. The results show that PINN provide higher accuracy and better alignment with the analytical solution over long-term simulations compared to numerical methods. Furthermore, PINN exhibit certain extrapolation capabilities. Additionally, the introduction of dropout enhances the generalization ability and convergence speed of the PINN model, confirming the potential of PINN in solving complex physical problems.

, correspAuthors=Longcheng LIU, authorNote=null, correspAuthorsNote=null, copyrightStatement=null, copyrightOwner=null, extLink=null, articleAbsUrl=null, sourceXml=null, magXml=null, pdfUrl=null, pdf=null, pdfFileSize=null, pdfExtLink=null, richHtmlUrl=null, mobilePdfUrl=null, reviewReport=null, pdfFirstPage=null, abstractGraph=null, abstractGraphContent=null, abstractVideo=null, citation=null, cebUrl=null, magXmlContent=null, mapNumber=null, authorCompany=null, fund=null, authors=null, authorsList=Zhe'an ZHANG, Zhiting PENG, Yunlong BAI, Longcheng LIU), CN=ArticleExt(id=1147999691901498338, articleId=1147999688722215632, tenantId=1146029695717560320, journalId=1146123346816638986, language=CN, title=物理信息神经网络在地下水铀迁移对流-弥散方程中的应用, columnId=1175805041991631542, journalTitle=铀矿冶, columnName=开采·选冶, runingTitle=null, highlight=null, articleAbstract=

在原地浸出采铀工艺中,铀的迁移过程受地下水流动和溶质扩散的共同作用,可通过对流-弥散方程进行描述。建立原地浸出采铀工艺中铀浓度随时间和空间变化的模型,对准确预测地下水中铀的迁移过程至关重要。传统数值方法(如有限差分法)在处理高维复杂问题时,计算量大且易产生误差。本研究探索了物理信息神经网络在对流-弥散方程求解中的适用性与精确性,对一维对流-弥散方程进行了数值模拟,并将物理信息神经网络解与数值解、解析解进行了对比。结果表明,在长期模拟中物理信息神经网络解比数值解更贴近解析解,具有较高的精度,并具备一定的外推能力。另外,在引入随机失活处理后,物理信息神经网络模型的泛化能力和收敛速度显著提升,验证了物理信息神经网络在复杂物理问题中的应用潜力。

, correspAuthors=刘龙成, authorNote=null, correspAuthorsNote=
白云龙(1989-),男,山西大同人,博士,高级工程师,主要从事铀矿采冶与环境修复技术研究;
刘龙成(1965-),男,江苏宿迁人,博士,教授,主要从事人工智能在铀矿冶中的应用与核素迁移技术研究。
, copyrightStatement=null, copyrightOwner=null, extLink=null, articleAbsUrl=null, sourceXml=3Sx2NWxhddWRhHdjjudN9w==, magXml=fPnzRnKDHJJ9VSh496K58A==, pdfUrl=null, pdf=hqiUhTDtl/t8ujqVprukag==, pdfFileSize=null, pdfExtLink=null, richHtmlUrl=null, mobilePdfUrl=null, reviewReport=null, pdfFirstPage=null, abstractGraph=null, abstractGraphContent=null, abstractVideo=null, citation=null, cebUrl=null, magXmlContent=56h9HVkoKdIMFnQk6pjlhg==, mapNumber=null, authorCompany=null, fund=null, authors=

张喆安(1998—),男,辽宁大连人,硕士,助理工程师,主要从事人工智能在铀矿冶中的应用与测井技术研究。

, authorsList=张喆安, 彭智婷, 白云龙, 刘龙成)}, authors=[Author(id=1179483340454380324, tenantId=1146029695717560320, journalId=1146123346816638986, articleId=1147999688722215632, orderNo=0, firstName=null, middleName=null, lastName=null, nameCn=null, orcid=null, stid=null, country=null, authorPic=null, dead=0, email=null, emailSecond=null, emailThird=null, correspondingAuthor=0, authorType=1, ext={EN=AuthorExt(id=1179483340538266407, tenantId=1146029695717560320, journalId=1146123346816638986, articleId=1147999688722215632, authorId=1179483340454380324, language=EN, stringName=Zhe'an ZHANG, firstName=Zhe'an, middleName=null, lastName=ZHANG, prefix=null, suffix=null, authorComment=null, nameInitials=null, affiliation=null, department=null, xref=null, address=Beijing Research Institute of Chemical Engineering and Metallurgy, CNNC, Beijing 101149, China, bio=null, bioImg=null, bioContent=null, aboutCorrespAuthor=null), CN=AuthorExt(id=1179483340659901225, tenantId=1146029695717560320, journalId=1146123346816638986, articleId=1147999688722215632, authorId=1179483340454380324, language=CN, stringName=张喆安, firstName=null, middleName=null, lastName=null, prefix=null, suffix=null, authorComment=null, nameInitials=null, affiliation=null, department=null, xref=null, address=核工业北京化工冶金研究院, 北京 101149, bio={"content":"

张喆安(1998—),男,辽宁大连人,硕士,助理工程师,主要从事人工智能在铀矿冶中的应用与测井技术研究。

"}, bioImg=null, bioContent=

张喆安(1998—),男,辽宁大连人,硕士,助理工程师,主要从事人工智能在铀矿冶中的应用与测井技术研究。

, aboutCorrespAuthor=null)}, companyList=[AuthorCompany(id=1179483340341134112, tenantId=1146029695717560320, journalId=1146123346816638986, articleId=1147999688722215632, xref=null, ext=[AuthorCompanyExt(id=1179483340349522721, tenantId=1146029695717560320, journalId=1146123346816638986, articleId=1147999688722215632, companyId=1179483340341134112, language=EN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=Beijing Research Institute of Chemical Engineering and Metallurgy, CNNC, Beijing 101149, China), AuthorCompanyExt(id=1179483340357911330, tenantId=1146029695717560320, journalId=1146123346816638986, articleId=1147999688722215632, companyId=1179483340341134112, language=CN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=核工业北京化工冶金研究院, 北京 101149)])]), Author(id=1179483340781536043, tenantId=1146029695717560320, journalId=1146123346816638986, articleId=1147999688722215632, orderNo=1, firstName=null, middleName=null, lastName=null, nameCn=null, orcid=null, stid=null, country=null, authorPic=null, dead=0, email=null, emailSecond=null, emailThird=null, correspondingAuthor=0, authorType=1, ext={EN=AuthorExt(id=1179483340848644910, tenantId=1146029695717560320, journalId=1146123346816638986, articleId=1147999688722215632, authorId=1179483340781536043, language=EN, stringName=Zhiting PENG, firstName=Zhiting, middleName=null, lastName=PENG, prefix=null, suffix=null, authorComment=null, nameInitials=null, affiliation=null, department=null, xref=null, address=Beijing Research Institute of Chemical Engineering and Metallurgy, CNNC, Beijing 101149, China, bio=null, bioImg=null, bioContent=null, aboutCorrespAuthor=null), CN=AuthorExt(id=1179483340915753776, tenantId=1146029695717560320, journalId=1146123346816638986, articleId=1147999688722215632, authorId=1179483340781536043, language=CN, stringName=彭智婷, firstName=null, middleName=null, lastName=null, prefix=null, suffix=null, authorComment=null, nameInitials=null, affiliation=null, department=null, xref=null, address=核工业北京化工冶金研究院, 北京 101149, bio=null, bioImg=null, bioContent=null, aboutCorrespAuthor=null)}, companyList=[AuthorCompany(id=1179483340341134112, tenantId=1146029695717560320, journalId=1146123346816638986, articleId=1147999688722215632, xref=null, ext=[AuthorCompanyExt(id=1179483340349522721, tenantId=1146029695717560320, journalId=1146123346816638986, articleId=1147999688722215632, companyId=1179483340341134112, language=EN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=Beijing Research Institute of Chemical Engineering and Metallurgy, CNNC, Beijing 101149, China), AuthorCompanyExt(id=1179483340357911330, tenantId=1146029695717560320, journalId=1146123346816638986, articleId=1147999688722215632, companyId=1179483340341134112, language=CN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=核工业北京化工冶金研究院, 北京 101149)])]), Author(id=1179483340966085427, tenantId=1146029695717560320, journalId=1146123346816638986, articleId=1147999688722215632, orderNo=2, firstName=null, middleName=null, lastName=null, nameCn=null, orcid=null, stid=null, country=null, authorPic=null, dead=0, email=null, emailSecond=null, emailThird=null, correspondingAuthor=0, authorType=1, ext={EN=AuthorExt(id=1179483341104497462, tenantId=1146029695717560320, journalId=1146123346816638986, articleId=1147999688722215632, authorId=1179483340966085427, language=EN, stringName=Yunlong BAI, firstName=Yunlong, middleName=null, lastName=BAI, prefix=null, suffix=null, authorComment=null, nameInitials=null, affiliation=null, department=null, xref=null, address=Beijing Research Institute of Chemical Engineering and Metallurgy, CNNC, Beijing 101149, China, bio=null, bioImg=null, bioContent=null, aboutCorrespAuthor=null), CN=AuthorExt(id=1179483341196772152, tenantId=1146029695717560320, journalId=1146123346816638986, articleId=1147999688722215632, authorId=1179483340966085427, language=CN, stringName=白云龙, firstName=null, middleName=null, lastName=null, prefix=null, suffix=null, authorComment=null, nameInitials=null, affiliation=null, department=null, xref=null, address=核工业北京化工冶金研究院, 北京 101149, bio=null, bioImg=null, bioContent=null, aboutCorrespAuthor=null)}, companyList=[AuthorCompany(id=1179483340341134112, tenantId=1146029695717560320, journalId=1146123346816638986, articleId=1147999688722215632, xref=null, ext=[AuthorCompanyExt(id=1179483340349522721, tenantId=1146029695717560320, journalId=1146123346816638986, articleId=1147999688722215632, companyId=1179483340341134112, language=EN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=Beijing Research Institute of Chemical Engineering and Metallurgy, CNNC, Beijing 101149, China), AuthorCompanyExt(id=1179483340357911330, tenantId=1146029695717560320, journalId=1146123346816638986, articleId=1147999688722215632, companyId=1179483340341134112, language=CN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=核工业北京化工冶金研究院, 北京 101149)])]), Author(id=1179483341280658235, tenantId=1146029695717560320, journalId=1146123346816638986, articleId=1147999688722215632, orderNo=3, firstName=null, middleName=null, lastName=null, nameCn=null, orcid=null, stid=null, country=null, authorPic=null, dead=0, email=null, emailSecond=null, emailThird=null, correspondingAuthor=0, authorType=1, ext={EN=AuthorExt(id=1179483341351961406, tenantId=1146029695717560320, journalId=1146123346816638986, articleId=1147999688722215632, authorId=1179483341280658235, language=EN, stringName=Longcheng LIU, firstName=Longcheng, middleName=null, lastName=LIU, prefix=null, suffix=null, authorComment=null, nameInitials=null, affiliation=null, department=null, xref=null, address=Beijing Research Institute of Chemical Engineering and Metallurgy, CNNC, Beijing 101149, China, bio=null, bioImg=null, bioContent=null, aboutCorrespAuthor=null), CN=AuthorExt(id=1179483341431653184, tenantId=1146029695717560320, journalId=1146123346816638986, articleId=1147999688722215632, authorId=1179483341280658235, language=CN, stringName=刘龙成, firstName=null, middleName=null, lastName=null, prefix=null, suffix=null, authorComment=null, nameInitials=null, affiliation=null, department=null, xref=null, address=核工业北京化工冶金研究院, 北京 101149, bio=null, bioImg=null, bioContent=null, aboutCorrespAuthor=null)}, companyList=[AuthorCompany(id=1179483340341134112, tenantId=1146029695717560320, journalId=1146123346816638986, articleId=1147999688722215632, xref=null, ext=[AuthorCompanyExt(id=1179483340349522721, tenantId=1146029695717560320, journalId=1146123346816638986, articleId=1147999688722215632, companyId=1179483340341134112, language=EN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=Beijing Research Institute of Chemical Engineering and Metallurgy, CNNC, Beijing 101149, China), AuthorCompanyExt(id=1179483340357911330, tenantId=1146029695717560320, journalId=1146123346816638986, articleId=1147999688722215632, companyId=1179483340341134112, language=CN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=核工业北京化工冶金研究院, 北京 101149)])])], keywords=[Keyword(id=1179483341628785475, tenantId=1146029695717560320, journalId=1146123346816638986, articleId=1147999688722215632, language=EN, orderNo=1, keyword=in-situ leaching of uranium), Keyword(id=1179483341716865861, tenantId=1146029695717560320, journalId=1146123346816638986, articleId=1147999688722215632, language=EN, orderNo=2, keyword=uranium migration), Keyword(id=1179483341783974727, tenantId=1146029695717560320, journalId=1146123346816638986, articleId=1147999688722215632, language=EN, orderNo=3, keyword=advection-diffusion equation), Keyword(id=1179483341846889289, tenantId=1146029695717560320, journalId=1146123346816638986, articleId=1147999688722215632, language=EN, orderNo=4, keyword=physics-informed neural network), Keyword(id=1179483341897220939, tenantId=1146029695717560320, journalId=1146123346816638986, articleId=1147999688722215632, language=EN, orderNo=5, keyword=finite difference numerical simulation), Keyword(id=1179483341960135501, tenantId=1146029695717560320, journalId=1146123346816638986, articleId=1147999688722215632, language=CN, orderNo=1, keyword=原地浸出采铀), Keyword(id=1179483342031438671, tenantId=1146029695717560320, journalId=1146123346816638986, articleId=1147999688722215632, language=CN, orderNo=2, keyword=铀迁移), Keyword(id=1179483342199210834, tenantId=1146029695717560320, journalId=1146123346816638986, articleId=1147999688722215632, language=CN, orderNo=3, keyword=对流-弥散方程), Keyword(id=1179483342299874133, tenantId=1146029695717560320, journalId=1146123346816638986, articleId=1147999688722215632, language=CN, orderNo=4, keyword=物理信息神经网络), Keyword(id=1179483342383760214, tenantId=1146029695717560320, journalId=1146123346816638986, articleId=1147999688722215632, language=CN, orderNo=5, keyword=有限差分数值模拟)], refs=[Reference(id=1179483344363471738, tenantId=1146029695717560320, journalId=1146123346816638986, articleId=1147999688722215632, doi=null, pmid=null, pmcid=null, year=2023, volume=42, issue=1, pageStart=1, pageEnd=9, url=null, language=null, rfNumber=[1], rfOrder=0, authorNames=柳琪, 李春光, 刘龙成, journalName=铀矿冶, refType=null, unstructuredReference=柳琪, 李春光, 刘龙成, 等. 地浸采铀工艺数值模拟进展及电动地浸采铀工艺数值模型建立[J]. 铀矿冶, 2023, 42(1):1-9., articleTitle=地浸采铀工艺数值模拟进展及电动地浸采铀工艺数值模型建立, refAbstract=null), Reference(id=1179483344426386299, tenantId=1146029695717560320, journalId=1146123346816638986, articleId=1147999688722215632, doi=null, pmid=null, pmcid=null, year=2023, volume=42, issue=1, pageStart=1, pageEnd=9, url=null, language=null, rfNumber=[1], rfOrder=1, authorNames=LIU Qi, LI Chunguang, LIU Longcheng, journalName=Uranium Mining and Metallurgy, refType=null, unstructuredReference=LIU Qi, LI Chunguang, LIU Longcheng, et al. Progress of numerical simulation for in-situ leaching uranium and establishment of numerical model for electrokinetic in-situ leaching uranium[J]. Uranium Mining and Metallurgy, 2023, 42(1):1-9 (in Chinese)., articleTitle=Progress of numerical simulation for in-situ leaching uranium and establishment of numerical model for electrokinetic in-situ leaching uranium, refAbstract=null), Reference(id=1179483344485106556, tenantId=1146029695717560320, journalId=1146123346816638986, articleId=1147999688722215632, doi=null, pmid=null, pmcid=null, year=2021, volume=46, issue=7, pageStart=14, pageEnd=18, url=null, language=null, rfNumber=[2], rfOrder=2, authorNames=李洋, 刘腾, 陈佳, journalName=环境科学与管理, refType=null, unstructuredReference=李洋, 刘腾, 陈佳. 原地浸出采铀的辐射环境影响研究[J]. 环境科学与管理, 2021, 46(7):14-18., articleTitle=原地浸出采铀的辐射环境影响研究, refAbstract=null), Reference(id=1179483344573186941, tenantId=1146029695717560320, journalId=1146123346816638986, articleId=1147999688722215632, doi=null, pmid=null, pmcid=null, year=2021, volume=46, issue=7, pageStart=14, pageEnd=18, url=null, language=null, rfNumber=[2], rfOrder=3, authorNames=LI Yang, LIU Teng, CHEN Jia, journalName=Environmental Science and Management, refType=null, unstructuredReference=LI Yang, LIU Teng, CHEN Jia. Analysis of radiation environmental impact of in-situ leaching uranium mining[J]. Environmental Science and Management, 2021, 46(7):14-18 (in Chinese)., articleTitle=Analysis of radiation environmental impact of in-situ leaching uranium mining, refAbstract=null), Reference(id=1179483344732570494, tenantId=1146029695717560320, journalId=1146123346816638986, articleId=1147999688722215632, doi=null, pmid=null, pmcid=null, year=null, volume=null, issue=null, pageStart=null, pageEnd=null, url=null, language=null, rfNumber=[3], rfOrder=4, authorNames=王嗣晨, 杜志明, 刘正邦, journalName=中国核学会.中国核科学技术进展报告(第八卷)中国核学会2023年学术年会论文集, refType=null, unstructuredReference=王嗣晨, 杜志明, 刘正邦, 等. 含矿含水层非均质结构对浸出液运移影响的模拟研究[C]// 中国核学会.中国核科学技术进展报告(第八卷)中国核学会2023年学术年会论文集:第1册. 北京: 核工业北京化工冶金研究院,2023:8., articleTitle=含矿含水层非均质结构对浸出液运移影响的模拟研究, refAbstract=null), Reference(id=1179483344782902143, tenantId=1146029695717560320, journalId=1146123346816638986, articleId=1147999688722215632, doi=null, pmid=null, pmcid=null, year=2023, volume=null, issue=null, pageStart=null, pageEnd=null, url=null, language=null, rfNumber=[4], rfOrder=5, authorNames=袁新, journalName=蒙其古尔铀矿CO2+O2地浸中的溶质运移与孔隙演化特征, refType=null, unstructuredReference=袁新. 蒙其古尔铀矿CO2+O2地浸中的溶质运移与孔隙演化特征[D]. 抚州: 东华理工大学, 2023., articleTitle=null, refAbstract=null), Reference(id=1179483344837428096, tenantId=1146029695717560320, journalId=1146123346816638986, articleId=1147999688722215632, doi=null, pmid=null, pmcid=null, year=2023, volume=null, issue=null, pageStart=null, pageEnd=null, url=null, language=null, rfNumber=[5], rfOrder=6, authorNames=刘春雨, journalName=酸法地浸过程中铀的迁移特征实验研究, refType=null, unstructuredReference=刘春雨. 酸法地浸过程中铀的迁移特征实验研究[D]. 抚州: 东华理工大学, 2023., articleTitle=null, refAbstract=null), Reference(id=1179483344921314177, tenantId=1146029695717560320, journalId=1146123346816638986, articleId=1147999688722215632, doi=null, pmid=null, pmcid=null, year=2022, volume=null, issue=null, pageStart=null, pageEnd=null, url=null, language=null, rfNumber=[6], rfOrder=7, authorNames=王兵, journalName=巴彦乌拉铀矿酸法地浸体系的反应溶质运移模拟, refType=null, unstructuredReference=王兵. 巴彦乌拉铀矿酸法地浸体系的反应溶质运移模拟[D]. 抚州: 东华理工大学, 2022., articleTitle=null, refAbstract=null), Reference(id=1179483344980034434, tenantId=1146029695717560320, journalId=1146123346816638986, articleId=1147999688722215632, doi=null, pmid=null, pmcid=null, year=2014, volume=null, issue=null, pageStart=null, pageEnd=null, url=null, language=null, rfNumber=[7], rfOrder=8, authorNames=杜洋, journalName=721矿尾矿库中核素(铀、钍)迁移特征及其对库区水环境影响, refType=null, unstructuredReference=杜洋. 721矿尾矿库中核素(铀、钍)迁移特征及其对库区水环境影响[D]. 抚州: 东华理工大学, 2014., articleTitle=null, refAbstract=null), Reference(id=1179483345059726211, tenantId=1146029695717560320, journalId=1146123346816638986, articleId=1147999688722215632, doi=null, pmid=null, pmcid=null, year=1995, volume=null, issue=1, pageStart=30, pageEnd=35, url=null, language=null, rfNumber=[8], rfOrder=9, authorNames=田振夫, journalName=宁夏大学学报(自然科学版), refType=null, unstructuredReference=田振夫. 一维对流扩散方程的四阶精度有限差分法[J]. 宁夏大学学报(自然科学版), 1995(1):30-35., articleTitle=一维对流扩散方程的四阶精度有限差分法, refAbstract=null), Reference(id=1179483345122640772, tenantId=1146029695717560320, journalId=1146123346816638986, articleId=1147999688722215632, doi=null, pmid=null, pmcid=null, year=1995, volume=null, issue=1, pageStart=30, pageEnd=35, url=null, language=null, rfNumber=[8], rfOrder=10, authorNames=TIAN Zhenfu, journalName=Journal of Ningxia University(Natural Science Edition), refType=null, unstructuredReference=TIAN Zhenfu. Fourth-order finite difference methods for one-dimensional convection-diffusion equations[J]. Journal of Ningxia University(Natural Science Edition), 1995(1):30-35 (in Chinese)., articleTitle=Fourth-order finite difference methods for one-dimensional convection-diffusion equations, refAbstract=null), Reference(id=1179483345181361029, tenantId=1146029695717560320, journalId=1146123346816638986, articleId=1147999688722215632, doi=null, pmid=null, pmcid=null, year=2023, volume=48, issue=8, pageStart=41, pageEnd=54, url=null, language=null, rfNumber=[9], rfOrder=11, authorNames=王明镜, 田芳, 郭亚妮, journalName=西南师范大学学报(自然科学版), refType=null, unstructuredReference=王明镜, 田芳, 郭亚妮. 求解对流扩散反应方程的高阶指数型组合紧致差分格式[J]. 西南师范大学学报(自然科学版), 2023, 48(8):41-54., articleTitle=求解对流扩散反应方程的高阶指数型组合紧致差分格式, refAbstract=null), Reference(id=1179483345256858502, tenantId=1146029695717560320, journalId=1146123346816638986, articleId=1147999688722215632, doi=null, pmid=null, pmcid=null, year=2023, volume=48, issue=8, pageStart=41, pageEnd=54, url=null, language=null, rfNumber=[9], rfOrder=12, authorNames=WANG Mingjing, TIAN Fang, GUO Yani, journalName=Journal of Southwest China Normal University(Natural Science Edition), refType=null, unstructuredReference=WANG Mingjing, TIAN Fang, GUO Yani. A high-order exponential combination compact finite difference scheme for convection-diffusiom-reaction equation[J]. Journal of Southwest China Normal University(Natural Science Edition), 2023, 48(8):41-54 (in Chinese)., articleTitle=A high-order exponential combination compact finite difference scheme for convection-diffusiom-reaction equation, refAbstract=null), Reference(id=1179483345311384455, tenantId=1146029695717560320, journalId=1146123346816638986, articleId=1147999688722215632, doi=null, pmid=null, pmcid=null, year=2022, volume=43, issue=2, pageStart=187, pageEnd=197, url=null, language=null, rfNumber=[10], rfOrder=13, authorNames=魏剑英, 葛永斌, journalName=应用数学和力学, refType=null, unstructuredReference=魏剑英, 葛永斌. 一种求解三维非稳态对流扩散反应方程的高精度有限差分格式[J]. 应用数学和力学, 2022, 43(2):187-197., articleTitle=一种求解三维非稳态对流扩散反应方程的高精度有限差分格式, refAbstract=null), Reference(id=1179483345365910408, tenantId=1146029695717560320, journalId=1146123346816638986, articleId=1147999688722215632, doi=null, pmid=null, pmcid=null, year=2022, volume=43, issue=2, pageStart=187, pageEnd=197, url=null, language=null, rfNumber=[10], rfOrder=14, authorNames=WEI Jianying, GE Yongbin, journalName=Applied Mathematics and Mechanics, refType=null, unstructuredReference=WEI Jianying, GE Yongbin. A high-order finite difference scheme for 3D unsteady convection diffusion reaction equations[J]. Applied Mathematics and Mechanics, 2022, 43(2):187-197 (in Chinese)., articleTitle=A high-order finite difference scheme for 3D unsteady convection diffusion reaction equations, refAbstract=null), Reference(id=1179483345416242057, tenantId=1146029695717560320, journalId=1146123346816638986, articleId=1147999688722215632, doi=null, pmid=null, pmcid=null, year=2018, volume=null, issue=null, pageStart=686, pageEnd=707, url=null, language=null, rfNumber=[11], rfOrder=15, authorNames=RAISSI M, PERDIKARIS P, KARNIADAKIS G E, journalName=Journal of Computational Physics, refType=null, unstructuredReference=RAISSI M, PERDIKARIS P, KARNIADAKIS G E. Physics-informed neural networks:a deep learning framework for solving forward and inverse problems involving nonlinear partial differential equations[J]. Journal of Computational Physics, 2018,378:686-707., articleTitle=Physics-informed neural networks:a deep learning framework for solving forward and inverse problems involving nonlinear partial differential equations, refAbstract=null), Reference(id=1179483345537876874, tenantId=1146029695717560320, journalId=1146123346816638986, articleId=1147999688722215632, doi=null, pmid=null, pmcid=null, year=2024, volume=44, issue=9, pageStart=1459, pageEnd=1465, url=null, language=null, rfNumber=[12], rfOrder=16, authorNames=张振, 苏欣荣, 袁新, journalName=动力工程学报, refType=null, unstructuredReference=张振, 苏欣荣, 袁新. 基于物理信息神经网络的气膜冷却湍流模型反演学习[J]. 动力工程学报, 2024, 44(9):1459-1465., articleTitle=基于物理信息神经网络的气膜冷却湍流模型反演学习, refAbstract=null), Reference(id=1179483345630151563, tenantId=1146029695717560320, journalId=1146123346816638986, articleId=1147999688722215632, doi=null, pmid=null, pmcid=null, year=2024, volume=44, issue=9, pageStart=1459, pageEnd=1465, url=null, language=null, rfNumber=[12], rfOrder=17, authorNames=ZHANG Zhen, SU Xinrong, YUAN Xin, journalName=Journal of Chinese Society of Power Engineering, refType=null, unstructuredReference=ZHANG Zhen, SU Xinrong, YUAN Xin. PINN-based inversion learning of turbulence model for film cooling[J]. Journal of Chinese Society of Power Engineering, 2024, 44(9):1459-1465 (in Chinese)., articleTitle=PINN-based inversion learning of turbulence model for film cooling, refAbstract=null), Reference(id=1179483345735009164, tenantId=1146029695717560320, journalId=1146123346816638986, articleId=1147999688722215632, doi=null, pmid=null, pmcid=null, year=2024, volume=45, issue=9, pageStart=1212, pageEnd=1223, url=null, language=null, rfNumber=[13], rfOrder=18, authorNames=闵建, 傅卓佳, 郭远, journalName=应用数学和力学, refType=null, unstructuredReference=闵建, 傅卓佳, 郭远. 课程-迁移学习物理信息神经网络用于曲面长时间对流扩散行为模拟[J]. 应用数学和力学, 2024, 45(9):1212-1223., articleTitle=课程-迁移学习物理信息神经网络用于曲面长时间对流扩散行为模拟, refAbstract=null), Reference(id=1179483345810506637, tenantId=1146029695717560320, journalId=1146123346816638986, articleId=1147999688722215632, doi=null, pmid=null, pmcid=null, year=2024, volume=45, issue=9, pageStart=1212, pageEnd=1223, url=null, language=null, rfNumber=[13], rfOrder=19, authorNames=MIN Jian, FU Zhuojia, GUO Yuan, journalName=Applied Mathematics and Mechanics, refType=null, unstructuredReference=MIN Jian, FU Zhuojia, GUO Yuan. Curriculum-transfer-learning-based physics-informed neural networks for simulating long-term-evolution convection-diffusion behaviors on curved surfaces[J]. Applied Mathematics and Mechanics, 2024, 45(9):1212-1223 (in Chinese)., articleTitle=Curriculum-transfer-learning-based physics-informed neural networks for simulating long-term-evolution convection-diffusion behaviors on curved surfaces, refAbstract=null), Reference(id=1179483345869226894, tenantId=1146029695717560320, journalId=1146123346816638986, articleId=1147999688722215632, doi=null, pmid=null, pmcid=null, year=2024, volume=47, issue=12, pageStart=124, pageEnd=136, url=null, language=null, rfNumber=[14], rfOrder=20, authorNames=陈豪龙, 唐欣越, 柳兆涛, journalName=重庆大学学报, refType=null, unstructuredReference=陈豪龙, 唐欣越, 柳兆涛, 等. 基于物理信息神经网络的非线性瞬态热传导正/反问题研究[J]. 重庆大学学报, 2024, 47(12):124-136., articleTitle=基于物理信息神经网络的非线性瞬态热传导正/反问题研究, refAbstract=null), Reference(id=1179483345927947151, tenantId=1146029695717560320, journalId=1146123346816638986, articleId=1147999688722215632, doi=null, pmid=null, pmcid=null, year=2024, volume=47, issue=12, pageStart=124, pageEnd=136, url=null, language=null, rfNumber=[14], rfOrder=21, authorNames=CHEN Haolong, TANG Xinyue, LIU Zhaotao, journalName=Journal of Chongqing University, refType=null, unstructuredReference=CHEN Haolong, TANG Xinyue, LIU Zhaotao, et al. Solving nonlinear transient heat conduction forward/inverse problem based on physical-information neural networks[J]. Journal of Chongqing University, 2024, 47(12):124-136 (in Chinese)., articleTitle=Solving nonlinear transient heat conduction forward/inverse problem based on physical-information neural networks, refAbstract=null), Reference(id=1179483345986667408, tenantId=1146029695717560320, journalId=1146123346816638986, articleId=1147999688722215632, doi=null, pmid=null, pmcid=null, year=null, volume=null, issue=null, pageStart=109, pageEnd=null, url=null, language=null, rfNumber=[15], rfOrder=22, authorNames=ZHENG C M, GORDON D B, journalName=Applied contaminant transport modeling, refType=null, unstructuredReference=ZHENG C M, GORDON D B. Applied contaminant transport modeling[M]. 2th ed. New York: John Wiley & Sons,2002:109-111., articleTitle=null, refAbstract=null), Reference(id=1179483346095719313, tenantId=1146029695717560320, journalId=1146123346816638986, articleId=1147999688722215632, doi=null, pmid=null, pmcid=null, year=2009, volume=52, issue=13/14, pageStart=3297, pageEnd=3304, url=null, language=null, rfNumber=[16], rfOrder=23, authorNames=GUERRERO J S P, PIMENTEL L C G, SKAGGS T H, journalName=International Journal of Heat & Mass Transfer, refType=null, unstructuredReference=GUERRERO J S P, PIMENTEL L C G, SKAGGS T H, et al. Analytical solution of the advection-diffusion transport equation using a change-of-variable and integral transform technique[J]. International Journal of Heat & Mass Transfer, 2009, 52(13/14):3297-3304., articleTitle=Analytical solution of the advection-diffusion transport equation using a change-of-variable and integral transform technique, refAbstract=null), Reference(id=1179483346196382610, tenantId=1146029695717560320, journalId=1146123346816638986, articleId=1147999688722215632, doi=null, pmid=null, pmcid=null, year=1992, volume=28, issue=2, pageStart=551, pageEnd=561, url=null, language=null, rfNumber=[17], rfOrder=24, authorNames=NOORISHAD J, TSANG C F, PERROCHET P, journalName=Water Resources Research, refType=null, unstructuredReference=NOORISHAD J, TSANG C F, PERROCHET P, et al. A perspective on the numerical solution of convection-dominated transport problems:A price to pay for the easy way out[J]. Water Resources Research, 1992, 28(2):551-561., articleTitle=A perspective on the numerical solution of convection-dominated transport problems:A price to pay for the easy way out, refAbstract=null), Reference(id=1179483346297045907, tenantId=1146029695717560320, journalId=1146123346816638986, articleId=1147999688722215632, doi=null, pmid=null, pmcid=null, year=1961, volume=null, issue=37, pageStart=null, pageEnd=null, url=null, language=null, rfNumber=[18], rfOrder=25, authorNames=OGATA A, BANKS R B, journalName=Geological Survey Professional, refType=null, unstructuredReference=OGATA A, BANKS R B. A solution of the differential equation of longitudinal dispersion in porous media[J]. Geological Survey Professional, 1961, 411-A(37)., articleTitle=A solution of the differential equation of longitudinal dispersion in porous media, refAbstract=null), Reference(id=1179483346368349076, tenantId=1146029695717560320, journalId=1146123346816638986, articleId=1147999688722215632, doi=null, pmid=null, pmcid=null, year=2022, volume=92, issue=3, pageStart=88, pageEnd=null, url=null, language=null, rfNumber=[19], rfOrder=26, authorNames=CUOMO S, DI COLA V S, GIAMPAOLO F, journalName=Journal of Scientific Computing, refType=null, unstructuredReference=CUOMO S, DI COLA V S, GIAMPAOLO F, et al. Scientific machine learning through physics-informed neural networks: where we are and what’s next[J]. Journal of Scientific Computing, 2022, 92(3):88., articleTitle=Scientific machine learning through physics-informed neural networks: where we are and what’s next, refAbstract=null), Reference(id=1179483346443846549, tenantId=1146029695717560320, journalId=1146123346816638986, articleId=1147999688722215632, doi=null, pmid=null, pmcid=null, year=2014, volume=21, issue=S1, pageStart=78, pageEnd=84, url=null, language=null, rfNumber=[20], rfOrder=27, authorNames=曲钧天, 丁进良, journalName=控制工程, refType=null, unstructuredReference=曲钧天, 丁进良. 基于多层感知器神经网络的精矿品位预报[J]. 控制工程, 2014, 21(S1):78-84., articleTitle=基于多层感知器神经网络的精矿品位预报, refAbstract=null), Reference(id=1179483346519344022, tenantId=1146029695717560320, journalId=1146123346816638986, articleId=1147999688722215632, doi=null, pmid=null, pmcid=null, year=2014, volume=21, issue=S1, pageStart=78, pageEnd=84, url=null, language=null, rfNumber=[20], rfOrder=28, authorNames=QU Juntian, DING Jinliang, journalName=Control Engineering of China, refType=null, unstructuredReference=QU Juntian, DING Jinliang. Forecasting of concentrate grade based on multilayer perceptron neural network[J]. Control Engineering of China, 2014, 21(S1):78-84 (in Chinese)., articleTitle=Forecasting of concentrate grade based on multilayer perceptron neural network, refAbstract=null)], funds=[Fund(id=1179483344124396407, tenantId=1146029695717560320, journalId=1146123346816638986, articleId=1147999688722215632, awardId=CNNC-JCYJ-202333, language=CN, fundingSource=中核集团基础研究项目(CNNC-JCYJ-202333), fundOrder=null, country=null), Fund(id=1179483344187310968, tenantId=1146029695717560320, journalId=1146123346816638986, articleId=1147999688722215632, awardId=null, language=CN, fundingSource=中核集团基础研究项目(电驱动强化CO2+O2地浸机理研究), fundOrder=null, country=null)], companyList=[AuthorCompany(id=1179483340341134112, tenantId=1146029695717560320, journalId=1146123346816638986, articleId=1147999688722215632, xref=null, ext=[AuthorCompanyExt(id=1179483340349522721, tenantId=1146029695717560320, journalId=1146123346816638986, articleId=1147999688722215632, companyId=1179483340341134112, language=EN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=Beijing Research Institute of Chemical Engineering and Metallurgy, CNNC, Beijing 101149, China), AuthorCompanyExt(id=1179483340357911330, tenantId=1146029695717560320, journalId=1146123346816638986, articleId=1147999688722215632, companyId=1179483340341134112, language=CN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=核工业北京化工冶金研究院, 北京 101149)])], figs=[ArticleFig(id=1179483342656389978, tenantId=1146029695717560320, journalId=1146123346816638986, articleId=1147999688722215632, language=EN, label=Fig. 1, caption=Illustration of grid discretization showing spatial locations and node indices, figureFileSmall=mZB+o1yFXg7ucRqUOyIBWg==, figureFileBig=pe+m2XCgHaAl2o3w0KoslQ==, tableContent=null), ArticleFig(id=1179483342740276060, tenantId=1146029695717560320, journalId=1146123346816638986, articleId=1147999688722215632, language=CN, label=图1, caption=包括空间位置与节点索引的网格剖分示意图, figureFileSmall=mZB+o1yFXg7ucRqUOyIBWg==, figureFileBig=pe+m2XCgHaAl2o3w0KoslQ==, tableContent=null), ArticleFig(id=1179483342866105182, tenantId=1146029695717560320, journalId=1146123346816638986, articleId=1147999688722215632, language=EN, label=Fig. 2, caption=Schematic diagram of PINN architecture, figureFileSmall=fyyV++Dh5jAB9ZZTdneD4w==, figureFileBig=MnJVEkcjw9cXIhzIW72hlw==, tableContent=null), ArticleFig(id=1179483342949991264, tenantId=1146029695717560320, journalId=1146123346816638986, articleId=1147999688722215632, language=CN, label=图2, caption=PINN结构示意图, figureFileSmall=fyyV++Dh5jAB9ZZTdneD4w==, figureFileBig=MnJVEkcjw9cXIhzIW72hlw==, tableContent=null), ArticleFig(id=1179483343033877346, tenantId=1146029695717560320, journalId=1146123346816638986, articleId=1147999688722215632, language=EN, label=Fig. 3, caption=Illustration of random dropout processing, figureFileSmall=xnLHd16P18xE8cdgVPoigQ==, figureFileBig=MkyBnjHuCyhcRbhFUcFoWg==, tableContent=null), ArticleFig(id=1179483343105180516, tenantId=1146029695717560320, journalId=1146123346816638986, articleId=1147999688722215632, language=CN, label=图3, caption=随机失活处理示意图, figureFileSmall=xnLHd16P18xE8cdgVPoigQ==, figureFileBig=MkyBnjHuCyhcRbhFUcFoWg==, tableContent=null), ArticleFig(id=1179483343193260902, tenantId=1146029695717560320, journalId=1146123346816638986, articleId=1147999688722215632, language=EN, label=Fig. 4, caption=Comparison of training performance before and after applying random dropout in the PINN model, figureFileSmall=5/VIAsizpx6Oi96fS2elmA==, figureFileBig=Sj700IYD9MaLTAqI2XqvhA==, tableContent=null), ArticleFig(id=1179483343277146984, tenantId=1146029695717560320, journalId=1146123346816638986, articleId=1147999688722215632, language=CN, label=图4, caption=PINN模型随机失活前后训练效果对比, figureFileSmall=5/VIAsizpx6Oi96fS2elmA==, figureFileBig=Sj700IYD9MaLTAqI2XqvhA==, tableContent=null), ArticleFig(id=1179483343419753322, tenantId=1146029695717560320, journalId=1146123346816638986, articleId=1147999688722215632, language=EN, label=Fig. 5, caption=Comparison of PINN solution, numerical solution and analytical solution within the given interval, figureFileSmall=0B0raCu+PU8pjV+Cy4aFTg==, figureFileBig=SEZFmvpaXw7o3xIf+6/dmw==, tableContent=null), ArticleFig(id=1179483343541388140, tenantId=1146029695717560320, journalId=1146123346816638986, articleId=1147999688722215632, language=CN, label=图5, caption=给定区间内PINN解、数值解与解析解对比, figureFileSmall=0B0raCu+PU8pjV+Cy4aFTg==, figureFileBig=SEZFmvpaXw7o3xIf+6/dmw==, tableContent=null), ArticleFig(id=1179483343675605870, tenantId=1146029695717560320, journalId=1146123346816638986, articleId=1147999688722215632, language=EN, label=Fig. 6, caption=Comparison of PINN predicted solution and analytical solution outside the given interval, figureFileSmall=THnT+izmoOFZuj/0IIeFrA==, figureFileBig=4DvFJuMyktIeQ8Ko7uQMog==, tableContent=null), ArticleFig(id=1179483343788852080, tenantId=1146029695717560320, journalId=1146123346816638986, articleId=1147999688722215632, language=CN, label=图6, caption=给定区间外PINN预测解与解析解对比, figureFileSmall=THnT+izmoOFZuj/0IIeFrA==, figureFileBig=4DvFJuMyktIeQ8Ko7uQMog==, tableContent=null), ArticleFig(id=1179483343864349554, tenantId=1146029695717560320, journalId=1146123346816638986, articleId=1147999688722215632, language=EN, label=Table 1, caption=

Errors between two solution methods and the analytical solution at four time points

, figureFileSmall=null, figureFileBig=null, tableContent=
时间/s MSE{PINN预测解,解析解} MSE{数值解,解析解}
0.25 9.23×10-4 11.12×10-4
0.50 7.29×10-4 45.62×10-4
0.75 6.01×10-4 12.44×10-4
1.00 7.62×10-4 14.31×10-4
), ArticleFig(id=1179483343944041332, tenantId=1146029695717560320, journalId=1146123346816638986, articleId=1147999688722215632, language=CN, label=表1, caption=

4个时间点下2种求解方法与解析解的误差

, figureFileSmall=null, figureFileBig=null, tableContent=
时间/s MSE{PINN预测解,解析解} MSE{数值解,解析解}
0.25 9.23×10-4 11.12×10-4
0.50 7.29×10-4 45.62×10-4
0.75 6.01×10-4 12.44×10-4
1.00 7.62×10-4 14.31×10-4
)], attaches=null, journal=Journal(id=1146119359820312588, delFlag=0, nameCn=铀矿冶, nameEn=Uranium Mining and Metallurgy, nameHistory1=null, nameHistory2=null, issn=1000-8063, eissn=, cn=11-1969/TL, coden=null, periodic=2, language=CN, oaType=0, ccby=null, superviseOffice=null, ownerOffice=null, pubOffice=null, editorOffice=null, officeType=null, aims=null, clcCode=null, officeProv=null, officeCity=null, officeAddr=null, officeZip=null, officeEmail=null, officePhone=null, editDirector=null, officeDirector=null, officeDirectorPhone=null, officeStaffNum=null, officeEmpNum=null, coverPicUrl=KmudR/G+rYvTqd4XcYqogQ==, journalPrice=null, startedYear=null, abbrevIsoEn=Uran Min Metal, journalRemark=null, publicationField=null, createdTime=null, updatedTime=1755587156394, createdBy=null, updatedBy=15831073675, firstLetterCn=U, firstLetterEn=U, subjectCode=Engineering, subjectName=工程, subjectCodeEn=Engineering, subjectNameEn=null, picCn=KmudR/G+rYvTqd4XcYqogQ==, picEn=9rbWk3ds8oZefNFxqqUVow==, jcr=null, cjcr=null, exts=[JournalExt(id=1164580199531238247, language=CN, name=铀矿冶, nameHistory1=null, nameHistory2=null, managedBy=, sponsoredBy=, publishedBy=, editorOffice=, officeProv=null, officeCity=null, officeAddr=, officeZip=, editDirector=null, officeDirector=null, officePhone=null, coverPicUrl=null, journalRemark=, submitArticleUrl=null, websiteUrl=https://ykyi.cbpt.cnki.net/portal, createdTime=1755587156422, updatedTime=1755587156422, createdBy=15831073675, updatedBy=15831073675, submissionGuidelinesUrl=http://ykye.juqk.net/buy/, submissionAuthorUrl=https://ykyi.cbpt.cnki.net/EditorE3N/index.aspx?t=1, submissionEditorUrl=https://ykyi.cbpt.cnki.net/EditorE3N/index.aspx?t=3, submissionReviewUrl=https://ykyi.cbpt.cnki.net/EditorE3N/index.aspx?t=2, submissionCeEditorUrl=, submissionAeEditorUrl=, option={"copyright":""}), JournalExt(id=1164580199573181288, language=EN, name=Uranium Mining and Metallurgy, nameHistory1=null, nameHistory2=null, managedBy=, sponsoredBy=, publishedBy=, editorOffice=, officeProv=null, officeCity=null, officeAddr=, officeZip=, editDirector=null, officeDirector=null, officePhone=null, coverPicUrl=null, journalRemark=, submitArticleUrl=null, websiteUrl=https://ykyie.cbpt.cnki.net/portal, createdTime=1755587156432, updatedTime=1755587156432, createdBy=15831073675, updatedBy=15831073675, submissionGuidelinesUrl=http://ykye.juqk.net/buy/, submissionAuthorUrl=https://ykyi.cbpt.cnki.net/EditorE3N/index.aspx?t=1, submissionEditorUrl=https://ykyi.cbpt.cnki.net/EditorE3N/index.aspx?t=3, submissionReviewUrl=https://ykyi.cbpt.cnki.net/EditorE3N/index.aspx?t=2, submissionCeEditorUrl=, submissionAeEditorUrl=, option={"copyright":""})], databaseList=null, tenantJournalId=1146123346816638986, websiteList=[Website(id=1148007952688087041, webName=null, webTitle=null, webDomain=null, webCopyrigh=null, webIpcNo=null, seoTitle=null, seoKeywords=null, seoDescription=null, tenantJournalId=null, journalId=1146123346816638986, journalNameCn=null, journalNameEn=null, grayFlag=null, tenantId=1146029695717560320, platformId=null, journalGroupId=null, journalGroupNameCn=null, journalGroupNameEn=null, type=1, domain=https://castjournals.cast.org.cn/joweb/yky/EN, language=EN, createTime=1751636024889, createBy=18614031015, updateTime=1753523175948, updateBy=18614031015, name=《铀矿冶》-英文站点, tplId=1146101810881728533, title=Uranium Mining and Metallurgy, delFlag=0, indexPage=/home, props=[WebsiteProps(id=1155923533696983256, tenantId=1146029695717560320, journalId=null, journalGroupId=null, siteId=1148007952688087041, code=articleTextType, value=kx, createTime=1753523246454, updateTime=1753523246454, creator=18614031015, updator=18614031015), WebsiteProps(id=1155923533676011733, tenantId=1146029695717560320, journalId=null, journalGroupId=null, siteId=1148007952688087041, code=banner, value=null, createTime=1753523246449, updateTime=1753523246449, creator=18614031015, updator=18614031015), WebsiteProps(id=1155923533667623124, tenantId=1146029695717560320, journalId=null, journalGroupId=null, siteId=1148007952688087041, code=logo, value=https://castjournals.cast.org.cn/joweb/kjdb/CN/file/pic?fileId=MbrBQF5iwbqJZg/6eQNpYQ==, createTime=1753523246447, updateTime=1753523246447, creator=18614031015, updator=18614031015), WebsiteProps(id=1155923533688594647, tenantId=1146029695717560320, journalId=null, journalGroupId=null, siteId=1148007952688087041, code=picServerUrl, value=https://castjournals.cast.org.cn/joweb/kjdb/CN/file/pic, createTime=1753523246452, updateTime=1753523246452, creator=18614031015, updator=18614031015), WebsiteProps(id=1155923533680206038, tenantId=1146029695717560320, journalId=null, journalGroupId=null, siteId=1148007952688087041, code=staticResourcePath, value=https://castjournals.cast.org.cn/joweb/cast_kjdb_cn_619/, createTime=1753523246450, updateTime=1753523246450, creator=18614031015, updator=18614031015)]), Website(id=1148243567316820393, webName=null, webTitle=null, webDomain=null, webCopyrigh=null, webIpcNo=null, seoTitle=null, seoKeywords=null, seoDescription=null, tenantJournalId=null, journalId=1146123346816638986, journalNameCn=null, journalNameEn=null, grayFlag=null, tenantId=1146029695717560320, platformId=null, journalGroupId=null, journalGroupNameCn=null, journalGroupNameEn=null, type=1, domain=https://castjournals.cast.org.cn/joweb/yky/CN, language=CN, createTime=1751692199786, createBy=18614031015, updateTime=1753523196222, updateBy=18614031015, name=《铀矿冶》-中文站点, tplId=1146099689490845704, title=铀矿冶, delFlag=0, indexPage=/home, props=[WebsiteProps(id=1148623394028941377, tenantId=1146029695717560320, journalId=null, journalGroupId=null, siteId=1148243567316820393, code=articleTextType, value=kx, createTime=1751782757527, updateTime=1751782757527, creator=18614031015, updator=18614031015), WebsiteProps(id=1148623394003775550, tenantId=1146029695717560320, journalId=null, journalGroupId=null, siteId=1148243567316820393, code=banner, value=null, createTime=1751782757521, updateTime=1751782757521, creator=18614031015, updator=18614031015), WebsiteProps(id=1148623393991192637, tenantId=1146029695717560320, journalId=null, journalGroupId=null, siteId=1148243567316820393, code=logo, value=https://castjournals.cast.org.cn/joweb/kjdb/CN/file/pic?fileId=MbrBQF5iwbqJZg/6eQNpYQ==, createTime=1751782757518, updateTime=1751782757518, creator=18614031015, updator=18614031015), WebsiteProps(id=1148623394020552768, tenantId=1146029695717560320, journalId=null, journalGroupId=null, siteId=1148243567316820393, code=picServerUrl, value=https://castjournals.cast.org.cn/joweb/kjdb/CN/file/pic, createTime=1751782757525, updateTime=1751782757525, creator=18614031015, updator=18614031015), WebsiteProps(id=1148623394012164159, tenantId=1146029695717560320, journalId=null, journalGroupId=null, siteId=1148243567316820393, code=staticResourcePath, value=https://castjournals.cast.org.cn/joweb/cast_kjdb_cn_619/, createTime=1751782757523, updateTime=1751782757523, creator=18614031015, updator=18614031015)])], journalTitle=铀矿冶, weixinUrl=null, journalUrl=null, iacademicId=null, status=0, seqNo=null, journalTitleEn=Uranium Mining and Metallurgy, journalPhotoCn=KmudR/G+rYvTqd4XcYqogQ==, journalPhotoEn=9rbWk3ds8oZefNFxqqUVow==, journalFirstLetter=U, journalRecommend=null, journalNew=null, journalCollection=null, jcrJf=null, cjcrJf=null, jcrJfStr=null, cjcrJfStr=null, submissionFirstDecision=null, sciSubjectClassification=null, casSubjectClassification=null, citeScore=null, totalCitationFrequency=null, icpCode=null, psCode=null, advertisingLicenseCode=null, copyrightInformation=null, country=null, option=, provinceCode=null, provinceName=null, collectFlag=false), detailUrlCn=https://castjournals.cast.org.cn/joweb/yky/CN/10.13426/j.cnki.yky.2024.09.09, detailUrlEn=https://castjournals.cast.org.cn/joweb/yky/EN/10.13426/j.cnki.yky.2024.09.09, pdfUrlCn=https://castjournals.cast.org.cn/joweb/yky/CN/PDF/10.13426/j.cnki.yky.2024.09.09, pdfUrlEn=https://castjournals.cast.org.cn/joweb/yky/EN/PDF/10.13426/j.cnki.yky.2024.09.09, aliStartDate=null, aliEndDate=null, collectionFlag=false, citedCount=null, citedUrl=null, reference=null)
收藏切换
物理信息神经网络在地下水铀迁移对流-弥散方程中的应用
收藏切换
PDF下载
张喆安 , 彭智婷 , 白云龙 , 刘龙成
铀矿冶 | 开采·选冶 2025,44(2): 29-37
收起
收藏切换
铀矿冶 | 开采·选冶 2025, 44(2): 29-37
物理信息神经网络在地下水铀迁移对流-弥散方程中的应用
全屏
张喆安, 彭智婷, 白云龙, 刘龙成
作者信息
  • 核工业北京化工冶金研究院, 北京 101149
  • 张喆安(1998—),男,辽宁大连人,硕士,助理工程师,主要从事人工智能在铀矿冶中的应用与测井技术研究。

通讯作者:

白云龙(1989-),男,山西大同人,博士,高级工程师,主要从事铀矿采冶与环境修复技术研究;
刘龙成(1965-),男,江苏宿迁人,博士,教授,主要从事人工智能在铀矿冶中的应用与核素迁移技术研究。
Application of Physics-Informed Neural Networks in Solving the Advection-Diffusion Equation of Uranium Migration in Groundwater
Zhe'an ZHANG, Zhiting PENG, Yunlong BAI, Longcheng LIU
Affiliations
  • Beijing Research Institute of Chemical Engineering and Metallurgy, CNNC, Beijing 101149, China
出版时间: 2025-05-20 doi: 10.13426/j.cnki.yky.2024.09.09
文章导航
收藏切换

在原地浸出采铀工艺中,铀的迁移过程受地下水流动和溶质扩散的共同作用,可通过对流-弥散方程进行描述。建立原地浸出采铀工艺中铀浓度随时间和空间变化的模型,对准确预测地下水中铀的迁移过程至关重要。传统数值方法(如有限差分法)在处理高维复杂问题时,计算量大且易产生误差。本研究探索了物理信息神经网络在对流-弥散方程求解中的适用性与精确性,对一维对流-弥散方程进行了数值模拟,并将物理信息神经网络解与数值解、解析解进行了对比。结果表明,在长期模拟中物理信息神经网络解比数值解更贴近解析解,具有较高的精度,并具备一定的外推能力。另外,在引入随机失活处理后,物理信息神经网络模型的泛化能力和收敛速度显著提升,验证了物理信息神经网络在复杂物理问题中的应用潜力。

原地浸出采铀  /  铀迁移  /  对流-弥散方程  /  物理信息神经网络  /  有限差分数值模拟

In the in-situ leaching of uranium, uranium migration is influenced by both groundwater flow and solute diffusion, and this process can be effectively modeled using the advection-diffusion equation. Accurately modeling the variation of uranium concentration over time and space is crucial for predicting uranium migration in groundwater during in-situ leaching of uranium. Traditional numerical methods, such as the finite difference method, are computationally intensive and prone to errors while dealing with high-dimensional, complex problems. Therefore, this research aims to explore the applicability and accuracy of physics-informed neural networks (PINN) in solving the advection-diffusion equation. Through numerical simulations of the one-dimensional advection-diffusion equation, and by comparing the PINN solutions with numerical and analytical solutions. The results show that PINN provide higher accuracy and better alignment with the analytical solution over long-term simulations compared to numerical methods. Furthermore, PINN exhibit certain extrapolation capabilities. Additionally, the introduction of dropout enhances the generalization ability and convergence speed of the PINN model, confirming the potential of PINN in solving complex physical problems.

in-situ leaching of uranium  /  uranium migration  /  advection-diffusion equation  /  physics-informed neural network  /  finite difference numerical simulation
张喆安, 彭智婷, 白云龙, 刘龙成. 物理信息神经网络在地下水铀迁移对流-弥散方程中的应用. 铀矿冶, 2025 , 44 (2) : 29 -37 . DOI: 10.13426/j.cnki.yky.2024.09.09
Zhe'an ZHANG, Zhiting PENG, Yunlong BAI, Longcheng LIU. Application of Physics-Informed Neural Networks in Solving the Advection-Diffusion Equation of Uranium Migration in Groundwater[J]. Uranium Mining and Metallurgy, 2025 , 44 (2) : 29 -37 . DOI: 10.13426/j.cnki.yky.2024.09.09
原地浸出采铀技术通过向矿体注入浸出剂,使浸出剂与铀在地下水的流动和弥散作用下发生化学反应而溶解矿石中的铀,再通过抽液井将含铀溶液提取到地表进行处理。该技术环境破坏小、成本较低,在铀矿资源开发领域被广泛应用[1]。然而,随着开采活动的进行,如何准确描述和预测浸出液在地下水中的迁移过程成为亟待解决的问题[2]。现有研究主要关注非均质含水层结构对浸出液迁移的影响、不同岩性层的渗透特性,以及复杂地质条件下的溶质运移规律[3-4]。利用渗透系数随机场、蒙特卡洛模拟等方法,进行的泥岩夹层、砂岩互层结构、化学环境对浸出液运移的影响模拟,在一定程度上揭示了泥岩连续性、局部岩性变化等因素对浸出路径和迁移速率的影响[5-7]。然而,这些研究在模型维度、求解方法及计算效率上仍存在局限,尤其在高维复杂结构中计算量较大,在实际应用中存在一定限制。此外,部分研究的模拟模型在外推预测能力上也存在不足,难以在边界条件复杂的多维场景中准确预测溶质迁移行为。
对流-弥散方程(Advection-Diffusion Equation, ADE)是经典的数学模型之一,广泛应用于溶质在地下水运移中的研究,可为浸出液在水文地质环境中的弥散与富集提供有效的理论工具[8]。ADE可描述溶质随地下水流动的对流作用,刻画因介质孔隙度差异而引发的弥散过程,同时考虑了浸出液中铀的化学反应、吸附及沉淀等效应。在一维情况下,ADE通过解析解即可获得浓度的时空变化规律;但对于含有非线性项、复杂边界条件及非均匀物理参数的对流-弥散方程,需借助数值解法(如高精度紧致差分法)实现求解。这些数值解法在一维至三维的对流-弥散反应方程中表现出良好的精度和稳定性,但在高维问题中面临计算成本增加和误差控制的挑战[9-10]。物理信息神经网络(Physics-Informed Neural Networks, PINN)是一种新兴的偏微分方程求解方法,通过将物理方程直接融入神经网络的训练过程,避免了传统数值方法中的网格划分问题,并在处理高维问题、复杂边界条件及非线性方程上表现出优势[11]。此外,PINN能够结合观测数据,实现对未知物理参数的反演和不确定性分析,近年来已广泛应用于流体力学、传热等复杂物理问题的求解,并在湍流建模、长时间模拟以及热传导等方面展现出较高的灵活性和计算效率[12-14]
本文基于物理信息神经网络(PINN),以对流-弥散方程为基础,在一维条件下建立简化模型,以评估浸出液在均质环境中的迁移行为。这种基于PINN的求解方式,不仅避免了传统数值方法中对网格剖分的依赖,还能通过物理约束提升模型的外推能力,一定程度上克服了传统模拟方法在数据获取困难、计算复杂度高等问题上的不足。本研究旨在为多孔介质中浸出液迁移的理解提供初步依据,并为后续二维及三维模型的推广与精度提升奠定理论基础。
原地浸出采铀工艺通过向铀矿体注入浸出剂,使铀溶解并随着地下水流动和弥散,进而通过井泵抽取含铀溶液。ADE被广泛应用于模拟浸出液的迁移过程、地下水污染情况及环境修复效果等方面,ADE还可以结合浸出液中铀的化学吸附与沉淀机制,帮助研究地浸采铀矿床的成矿机理及其对环境的长期影响。ADE的通用方程形式为[15]
$\begin{array}{c}\frac{\partial C}{\partial t}=-v·\Delta C+\Delta ·\left(D\Delta C\right)+R,\end{array}$
式中:C为铀在轴向距离x(m)和时间t(s)时的浓度,mg/L;v为地下水流的速度场(对流项),m/s,代表溶质随水流的运动;D为弥散系数张量(弥散项),m2/s,用于描述溶质因分子弥散和水流速度不均匀性而产生的弥散现象;R为溶质反应或衰变项,mg/(L·s),如铀的吸附、沉淀和放射性衰变;Δ为梯度算子,描述浓度变化的方向。
地浸采铀过程涉及对铀浓度在时间和空间上的变化进行建模,可以采用对流-弥散方程来预测铀的迁移速率及其分布。某铀矿区的地下水流动是单向的,且主要由对流和弥散决定,在不涉及化学反应的前提下,其铀浓度C(x,t)满足一维对流-弥散方程[16]:
$\begin{array}{c}\frac{\partial C}{\partial t}=D\frac{{\partial }^{2}C}{\partial {x}^{2}}-v\frac{\partial C}{\partial x},\end{array}$
Ogata和Banks[17]给出了方程(2)的解析解:
$\begin{array}{c}\begin{array}{c}\frac{C}{{C}_{0}}=\frac{1}{2}\left[\mathrm{e}\mathrm{r}\mathrm{f}\mathrm{c}\left(\frac{x-vt}{\sqrt[ ]{4Dt}}\right)+\mathrm{e}\mathrm{x}\mathrm{p}\left(\frac{xv}{D}\right)\mathrm{e}\mathrm{r}\mathrm{f}\mathrm{c}\left(\frac{x+vt}{\sqrt[ ]{4Dt}}\right)\right]。\end{array}\end{array}$
在初始时刻,C(x,t)=0,表示初始时刻地下水中没有浸出液存在;在注入点x=0处,浸出液的初始浓度为常数C0,mg/L,即C(0,t)= C0, t>0;在远端x=L处,浸出液的浓度梯度趋于0,即$\frac{\partial C(L,t)}{\partial x}$=0。本研究取C0=5 mg/L,D=1×10-3 m2/s,v=0.1 mg/L,时间范围取0~1 s,距离范围为0~1 m的一维区间。这种小尺度的时间和空间范围选择,是基于对铀迁移行为的高分辨率动态观察需求,有助于精确捕捉铀浓度在采铀初始阶段的细微变化。
由于铀的迁移速度较慢,初始阶段的弥散过程往往对后续铀分布有显著影响,短时间和小范围的研究能够更好地分析初始的弥散与对流作用。在实际的地浸采铀过程中,通过模拟浓度的微小变化可以更深入理解浸出剂在铀矿体中的扩散与对流行为,为后续优化浸出剂的注入速率、浓度和抽取位置提供依据,从而提升铀的提取效率。同时,小尺度范围的模型研究便于分析边界条件及局部参数对铀浓度分布的影响,为现场实际应用提供精准参数支撑,并支持更合理的采矿工艺设计。
方程(2)作为一个连续偏微分方程,可通过有限差分数值模拟方法进行系统近似。首先,需要将其在时间域和空间域进行有限差分网格剖分,即将连续的区域离散为一系列单元和节点。对于1.1所述问题,进行等间距网格剖分,空间步长Δx=0.01 m,时间步长Δt=0.01 s,同时以每个单元的中心为网格节点,计算节点上的浓度。图1为网格剖分示意图,为了便于说明,图中仅展示了3个节点。在图示中,x代表网格中的空间位置,表示不同节点的空间坐标;j是网格中的节点索引,指代网格中的特定节点。
方程(2)的二阶导数弥散项在节点j处可近似为

D$\frac{{\partial }^{2}C}{\partial {x}^{2}}$D$\frac{{\left(\frac{\partial C}{\partial x}\right)}_{j+\frac{1}{2}}-{\left(\frac{\partial C}{\partial x}\right)}_{j-\frac{1}{2}}}{\Delta x}$

=D$\frac{{C}_{j+1}-2{C}_{j}+{C}_{j-1}}{{\left(\mathrm{\Delta }x\right)}^{2}}$,
其中(Cj+1-2Cj+Cj-1)近似表示单元j左右两个面的浓度梯度。
方程(2)的一阶导数对流项在节点j处可以近似表示为
$\begin{array}{c}v\frac{\partial C}{\partial x}\approx v\frac{{C}_{j+\frac{1}{2}}-{C}_{j-\frac{1}{2}}}{\mathrm{\Delta }x},\end{array}$
其中,${C}_{j+\frac{1}{2}}$${C}_{j-\frac{1}{2}}$分别表示单元j左、右面的浓度。对流项界面浓度的处理方法会显著影响对流-弥散方程的数值解。界面浓度的一般表达式为[18]
$\begin{array}{c}{C}_{j+\frac{1}{2}}=(1-\alpha ){C}_{j}+\alpha {C}_{j+1},\end{array}$
式中:α为空间加权因子,通常取α=0.5。该公式称为中心或空间中心加权方案。根据中心加权方案,对流项的有限差分近似式在空间剖分上精确到二阶,即其截断误差与网格间距Δx2成正比。弥散项也采用同样的处理方法来精确到二阶。
方程(2)的时间导数项可以近似为$\frac{{C}_{j}^{n+1}-{C}_{j}^{n}}{\mathrm{\Delta }t}$。其中,n为上一计算时间点,n+1为当前计算时间点,即用上一时间节点的浓度${C}_{j}^{n}$近似表示迁移方程中的弥散项和对流项。因此,可以得到对流-弥散方程的有限差分离散形式:
$\begin{array}{l}{C}_{j}^{n+1}={C}_{j}^{n}+\frac{D\mathrm{\Delta }t}{{\left(\mathrm{\Delta }x\right)}^{2}}({C}_{j+1}^{n}-2{C}_{j}^{n}+{C}_{j-1}^{n})-\\ \frac{v\Delta t}{\Delta x}\left[\right(1-\alpha ){C}_{j}^{n}+\alpha {C}_{j+1}^{n}-(1-\alpha ){C}_{j-1}^{n}-\alpha {C}_{j}^{n}]。\end{array}$
PINN通常采用全连接前馈神经网络,其输入为时间、空间坐标等独立变量,输出为温度、压力或速度等物理量。通过在损失函数中引入物理方程残差,PINN可以避免完全依赖数据的训练。损失函数一般由数据损失和物理损失两部分组成,数据损失为基于少量观测数据计算网络预测值与真实数据之间的误差;物理损失为基于物理方程的残差,即计算预测值与物理规律之间的偏差。其优化目标是同时最小化由数据损失和物理损失导致的误差,从而在数据较少的情况下,依然能够逼近物理方程的精确解。
与传统数值方法相比,PINN通过引入物理约束减少了对大规模观测数据的需求,并且更加适用于高维问题[19],此外,PINN在预测过程中可以始终保持物理上的一致性,确保预测结果符合物理定律,避免了纯数据驱动方法可能产生的不合理现象。这使得PINN在物理建模与科学计算中具有广阔的应用前景。
本研究建立了一个具有1个输入层、7个全连接隐藏层和1个输出层的PINN神经网络,每个隐藏层中包含20个神经元,其结构见图2,这一结构是基于对网络表达能力与计算效率的平衡而确定的。7个隐藏层可以有效增强模型的表达能力,确保其能够拟合复杂的函数关系,同时又避免因层数过多导致计算量过大或出现过拟合问题。每层20个神经元在提供足够表达能力的同时也避免了参数过多带来的训练困难,使得模型具有较强的拟合能力和适当的计算效率。蓝色圆圈为输入层神经元,橙色圆圈为隐藏层神经元,紫色圆圈为输出层神经元,绿色圆圈为物理损失项,蓝色线框为数据损失项。
PINN表达式如下:
$\begin{array}{c}\begin{array}{c}\left\{\begin{array}{l}{h}^{\left(1\right)}=\sigma ({W}^{\left(1\right)}x+{b}^{\left(1\right)})\\ {h}^{\left(\mathrm{i}\right)}=\sigma ({W}^{\left(\mathrm{i}\right)}{h}^{(\mathrm{i}-1)}+{b}^{\left(\mathrm{i}\right)}),i=2,\dots,8,\\ y={W}^{\left(9\right)}{h}^{\left(8\right)}+{b}^{\left(9\right)}\end{array}\right.\end{array}\end{array}$
式中:x为输入向量且x=(x,t),W(i)为维度为20×20的权重矩阵,b(i)为维度为20的偏置向量,h(i)为第i个隐藏层的输出,σ为每个隐藏层的激活函数。输入层将空间和时间变量xt传递至第1隐藏层,第1层通过表达式h(1)=σ(W(1)x+b(1))进行非线性转换。随后,每个隐藏层输出均以h(i)=σ(W(i)h(i-1)+b(i))递归计算,其中i=2,…,8,每一层通过激活函数σ对上一层的输出进行非线性变换,从而增强网络对复杂关系的拟合能力。最终在第9层,即输出层,直接得到预测值y=W(9)h(8)+b(9),输出模型对目标函数的预测。
由于全连接神经网络具有较强的拟合能力,为防止其在进行多次训练后出现“过拟合”现象,笔者对传统PINN结构进行了改进,在每个隐藏层中加入了随机失活(Dropout)处理来提高模型的泛化能力。Dropout是一种用于防止神经网络过拟合的正则化方法,其通过在训练过程中随机丢弃部分神经元,以提高模型的泛化能力。在每一轮训练中,Dropout以一定的概率,随机将网络中的神经元置为无效,使得这些神经元在前向传播和反向传播中不参与计算。这一机制迫使网络在不同训练批次中利用不同的子网络进行学习,从而减少对某一特定神经元或路径的依赖,类似于集成了多个子网络的效果。通过这种方式,Dropout能有效提升模型的鲁棒性,并减少“过拟合”[20]。随机失活处理效果见图3,⊗为被随机失活的神经元。
模型通过确保神经网络的输出C(x,t)满足一维对流-弥散方程及其边界和初始条件来进行训练。损失函数的设计包含两个部分,分别是在计算域内部的对流-弥散方程残差,以及预测值与边界和初始条件之间的误差。均方误差(Mean Squared Error,MSE)是最常用的回归损失函数,其核心是求取模型的预测结果与实际目标值之间的差异,通过调整模型参数最小化MSE,使其预测值尽可能接近实际目标值。MSE值越小,意味着模型的预测结果与实际目标值之间的差异越小,模型的性能越好。根据本研究建立的物理信息神经网络的结构特性,自行构建模型MSE,目的是确保模型输出同时满足对流-弥散方程及其初始和边界条件。本研究所采用的模型MSE定义如下:
$\begin{array}{c}\begin{array}{c}\left\{\begin{array}{l}\mathrm{M}\mathrm{S}\mathrm{E}=\mathrm{M}\mathrm{S}{\mathrm{E}}_{\{u,BC,IC\}}+\mathrm{M}\mathrm{S}{\mathrm{E}}_{R}\\ \mathrm{M}\mathrm{S}{\mathrm{E}}_{\{u,BC,IC\}}=\frac{1}{{N}_{u}}\stackrel{{N}_{u}}{\sum _{i=1}}{\left|u({x}_{i}^{u},{t}_{i}^{u})-{u}_{i}\right|}^{2}+\\ \frac{1}{{N}_{g}}\stackrel{{N}_{g}}{\sum _{i=1}}{\left|g({x}_{i}^{g},{t}_{i}^{g})-{g}_{i}\right|}^{2}\\ \mathrm{M}\mathrm{S}{\mathrm{E}}_{R}=\frac{1}{{N}_{f}}\stackrel{{N}_{f}}{\sum _{i=1}}{\left|f({x}_{i}^{f},{t}_{i}^{f})\right|}^{2}\end{array}\right.\end{array}\end{array}$
式中:f(${x}_{i}^{f}$,${t}_{i}^{f}$)为对流-弥散方程的左侧残差,描述了模型输出与PDE约束的偏差,配点 $\{{x}_{i}^{f},{t}_{i}^{f}{\}}_{i=1}^{{N}_{f}}$位于计算域的内部。u(${x}_{i}^{u}$,${t}_{i}^{u}$)和g(${x}_{i}^{g}$,${t}_{i}^{g}$)为模型在初始和边界条件下的输出,配点$\{{x}_{i}^{u},{t}_{i}^{u}{\}}_{i=1}^{{N}_{u}}$$\{{x}_{i}^{g},{t}_{i}^{g}{\}}_{i=1}^{{N}_{g}}$对应计算域的边界和初始条件。公式中的误差项包含两个部分:一是初始和边界条件的均方误差项,用于度量模型在边界和初始条件下的拟合误差;二是方程残差项,用于衡量模型输出与PDE残差的偏差,以确保模型满足方程的物理约束条件。通过最小化上述3个均方误差项,使模型能够在满足对流-弥散方程的同时,保证输出满足边界条件和初始条件。
分别测试传统PINN模型和加入随机失活处理的PINN模型在本研究中的训练效果,得到的训练迭代次数与损失值关系见图4。在传统PINN模型中进行1 000次迭代后,损失值达到0.083 582;而在加入随机失活处理的PINN模型中,进行800次迭代损失值便可下降到0.083 281。加入随机失活处理后的模型表现出更高的泛化能力,并有效提高了训练效率。
在本研究中,对一维对流-弥散方程进行了有限差分数值模拟,并将解析解、数值解与PINN解的结果进行对比,结果见图5。可以看出,数值解能较好地描述铀浓度随时间和空间的变化;但与解析解相比,数值解在部分区域存在一定偏差,尤其是在较长时间范围内误差逐渐增大。这主要是由于有限差分方法在处理复杂边界条件时的局限性,以及网格剖分与方程离散造成的损失,导致解的精度随时间的推移出现下降趋势。
相比之下,PINN解在各时间节点下与解析解的吻合度更高。4个时间点下数值解、PINN解与解析解误差对比见表1。可以看出,PINN解与解析解的误差在t=0.25 s时为9.23×10-4,受初期训练参数设置和网络权重初始化的影响,该误差略高于其他时刻;但在后续时刻表现出更好的稳定性。在t=0.50 s时,受网格剖分精度的影响,数值解的误差增大至0.004 5,而PINN解的误差仅为7.29×10-4。在t=1.00 s时,PINN解的误差为7.62×10-4,优于数值解的误差。这表明PINN方法在长时间模拟中能更好地保持与解析解的一致性,表现出更高的精度和稳定性。由于PINN方法不依赖于时间迭代计算,能够规避数值方法中时间推进带来的累积误差,因此在0.25~1.00 s内,PINN解未受到初始误差的影响,依然能够很好地贴合解析解。
对训练完成的模型进行时间外推预测能力的测试,PINN预测解与解析解的对比结果见图6。在非训练数据的时间范围内,PINN的预测解在一定时间内与解析解吻合度高,并保持了较高的精度。当t=1.5 s时,PINN预测解与解析解的MSE为4.23×10-4;在t>1.5 s时,PINN预测解的精度有所下降,但其依然能够较好地保持与解析解在形态上的一致性。
与传统的数值模拟方法相比,PINN通过将物理规律直接嵌入神经网络的损失函数中,能够对未参与训练的时间或空间区域进行物理一致性的预测,在对流-弥散过程中的表现更优,尤其在处理长时间演化和复杂边界条件时,PINN的求解精度和稳定性更高,与解析解有效逼近。PINN具有独特的外推和泛化能力。这种结合数据驱动与物理约束方法,使得PINN不仅在已知数据范围内表现出色,还在数据不足或不可获得的情况下,具备良好的外推能力。PINN能在超出训练范围时仍然遵循物理定律,从而保持结果的可信性和形态一致性。
通过PINN求解对流-弥散方程,验证了PINN方法在模拟对流-弥散现象中的适用性和精确性。研究结果表明,PINN能够在无需网格剖分的前提下,较好地模拟一维对流-弥散过程,且与解析解、数值解的误差较小。在训练过程中,加入随机失活处理的PINN模型比未加入处理的模型收敛速度更快,且泛化能力更强;加入随机失活的PINN模型能在更少的迭代次数下,达到更低的损失值,提升了模型的泛化能力与收敛速度,进一步证明了PINN模型在对流-弥散问题中的有效性与稳定性。
另外,有限差分方法虽然可以通过提高阶数来增加精度,但这种方法在处理高维问题时,计算量和计算时间都会显著增加。相比之下,PINN方法具有强大的外推能力。通过将物理规律直接嵌入神经网络的损失函数中,PINN能够在训练数据之外的时间和空间区域进行预测,并保持与已知数据一致的物理规律。这种外推能力使得PINN能够有效应对未知或不可获取的边界条件和数据,避免了传统数值方法中由于缺乏数据或边界条件的影响而产生的预测不准确。这种结合物理约束和数据驱动的特性,使得PINN在复杂系统的长时间模拟中展现出显著优势,在复杂物理系统模拟中展现了一定的应用前景。未来可以通过进一步优化网络结构和参数设置,以提升模型在多维问题和长时间模拟中的表现。
  • 中核集团基础研究项目(CNNC-JCYJ-202333)
  • 中核集团基础研究项目(电驱动强化CO2+O2地浸机理研究)
参考文献 引证文献
排序方式:
[1]
柳琪, 李春光, 刘龙成, 等. 地浸采铀工艺数值模拟进展及电动地浸采铀工艺数值模型建立[J]. 铀矿冶, 2023, 42(1):1-9.
LIU Qi, LI Chunguang, LIU Longcheng, et al. Progress of numerical simulation for in-situ leaching uranium and establishment of numerical model for electrokinetic in-situ leaching uranium[J]. Uranium Mining and Metallurgy, 2023, 42(1):1-9 (in Chinese).
[2]
李洋, 刘腾, 陈佳. 原地浸出采铀的辐射环境影响研究[J]. 环境科学与管理, 2021, 46(7):14-18.
LI Yang, LIU Teng, CHEN Jia. Analysis of radiation environmental impact of in-situ leaching uranium mining[J]. Environmental Science and Management, 2021, 46(7):14-18 (in Chinese).
[3]
王嗣晨, 杜志明, 刘正邦, 等. 含矿含水层非均质结构对浸出液运移影响的模拟研究[C]// 中国核学会.中国核科学技术进展报告(第八卷)中国核学会2023年学术年会论文集:第1册. 北京: 核工业北京化工冶金研究院,2023:8.
[4]
袁新. 蒙其古尔铀矿CO2+O2地浸中的溶质运移与孔隙演化特征[D]. 抚州: 东华理工大学, 2023.
[5]
刘春雨. 酸法地浸过程中铀的迁移特征实验研究[D]. 抚州: 东华理工大学, 2023.
[6]
王兵. 巴彦乌拉铀矿酸法地浸体系的反应溶质运移模拟[D]. 抚州: 东华理工大学, 2022.
[7]
杜洋. 721矿尾矿库中核素(铀、钍)迁移特征及其对库区水环境影响[D]. 抚州: 东华理工大学, 2014.
[8]
田振夫. 一维对流扩散方程的四阶精度有限差分法[J]. 宁夏大学学报(自然科学版), 1995(1):30-35.
TIAN Zhenfu. Fourth-order finite difference methods for one-dimensional convection-diffusion equations[J]. Journal of Ningxia University(Natural Science Edition), 1995(1):30-35 (in Chinese).
[9]
王明镜, 田芳, 郭亚妮. 求解对流扩散反应方程的高阶指数型组合紧致差分格式[J]. 西南师范大学学报(自然科学版), 2023, 48(8):41-54.
WANG Mingjing, TIAN Fang, GUO Yani. A high-order exponential combination compact finite difference scheme for convection-diffusiom-reaction equation[J]. Journal of Southwest China Normal University(Natural Science Edition), 2023, 48(8):41-54 (in Chinese).
[10]
魏剑英, 葛永斌. 一种求解三维非稳态对流扩散反应方程的高精度有限差分格式[J]. 应用数学和力学, 2022, 43(2):187-197.
WEI Jianying, GE Yongbin. A high-order finite difference scheme for 3D unsteady convection diffusion reaction equations[J]. Applied Mathematics and Mechanics, 2022, 43(2):187-197 (in Chinese).
[11]
RAISSI M, PERDIKARIS P, KARNIADAKIS G E. Physics-informed neural networks:a deep learning framework for solving forward and inverse problems involving nonlinear partial differential equations[J]. Journal of Computational Physics, 2018,378:686-707.
[12]
张振, 苏欣荣, 袁新. 基于物理信息神经网络的气膜冷却湍流模型反演学习[J]. 动力工程学报, 2024, 44(9):1459-1465.
ZHANG Zhen, SU Xinrong, YUAN Xin. PINN-based inversion learning of turbulence model for film cooling[J]. Journal of Chinese Society of Power Engineering, 2024, 44(9):1459-1465 (in Chinese).
[13]
闵建, 傅卓佳, 郭远. 课程-迁移学习物理信息神经网络用于曲面长时间对流扩散行为模拟[J]. 应用数学和力学, 2024, 45(9):1212-1223.
MIN Jian, FU Zhuojia, GUO Yuan. Curriculum-transfer-learning-based physics-informed neural networks for simulating long-term-evolution convection-diffusion behaviors on curved surfaces[J]. Applied Mathematics and Mechanics, 2024, 45(9):1212-1223 (in Chinese).
[14]
陈豪龙, 唐欣越, 柳兆涛, 等. 基于物理信息神经网络的非线性瞬态热传导正/反问题研究[J]. 重庆大学学报, 2024, 47(12):124-136.
CHEN Haolong, TANG Xinyue, LIU Zhaotao, et al. Solving nonlinear transient heat conduction forward/inverse problem based on physical-information neural networks[J]. Journal of Chongqing University, 2024, 47(12):124-136 (in Chinese).
[15]
ZHENG C M, GORDON D B. Applied contaminant transport modeling[M]. 2th ed. New York: John Wiley & Sons,2002:109-111.
[16]
GUERRERO J S P, PIMENTEL L C G, SKAGGS T H, et al. Analytical solution of the advection-diffusion transport equation using a change-of-variable and integral transform technique[J]. International Journal of Heat & Mass Transfer, 2009, 52(13/14):3297-3304.
[17]
NOORISHAD J, TSANG C F, PERROCHET P, et al. A perspective on the numerical solution of convection-dominated transport problems:A price to pay for the easy way out[J]. Water Resources Research, 1992, 28(2):551-561.
[18]
OGATA A, BANKS R B. A solution of the differential equation of longitudinal dispersion in porous media[J]. Geological Survey Professional, 1961, 411-A(37).
[19]
CUOMO S, DI COLA V S, GIAMPAOLO F, et al. Scientific machine learning through physics-informed neural networks: where we are and what’s next[J]. Journal of Scientific Computing, 2022, 92(3):88.
[20]
曲钧天, 丁进良. 基于多层感知器神经网络的精矿品位预报[J]. 控制工程, 2014, 21(S1):78-84.
QU Juntian, DING Jinliang. Forecasting of concentrate grade based on multilayer perceptron neural network[J]. Control Engineering of China, 2014, 21(S1):78-84 (in Chinese).
2025年第44卷第2期
PDF下载
393
166
引用本文
BibTeX
文章信息
doi: 10.13426/j.cnki.yky.2024.09.09
  • 接收时间:2024-09-30
  • 首发时间:2025-07-04
  • 出版时间:2025-05-20
补充材料
相关文章
文章信息
作者
出版历史
  • 收稿日期:2024-09-30
基金
中核集团基础研究项目(CNNC-JCYJ-202333)
中核集团基础研究项目(电驱动强化CO2+O2地浸机理研究)
作者信息
    核工业北京化工冶金研究院, 北京 101149

通讯作者:

白云龙(1989-),男,山西大同人,博士,高级工程师,主要从事铀矿采冶与环境修复技术研究;
刘龙成(1965-),男,江苏宿迁人,博士,教授,主要从事人工智能在铀矿冶中的应用与核素迁移技术研究。
参考文献
分享链接
https://castjournals.cast.org.cn/joweb/yky/CN/10.13426/j.cnki.yky.2024.09.09
分享至
全文二维码

扫描看全文

引用本文
BibTeX
本文的引用情况
2种不同金属材料的力学参数

Family
属数
Number of
genus
种数
Number of
species
占总种数比例
Percentage of
total species (%)

Genus
种数
Number of
species
占总种数比例
Percentage of total
species (%)
鹅膏菌科Amanitaceae 2 11 5.26 鹅膏菌属 Amanita 10 4.78
小菇科 Mycenaceae 2 12 5.74 丝盖伞属 Inocybe 5 2.39
多孔菌科 Polyporaceae 8 14 6.70 蜡蘑属 Laccaria 5 2.39
红菇科 Russulaceae 3 23 11.00 小皮伞属 Marasmius 6 2.87
小菇属 Mycena 11 5.26
光柄菇属 Pluteus 5 2.39
红菇属 Russula 17 8.13
栓菌属 Trametes 5 2.39
关闭全屏