数字孪生黄河知识平台建设初探

打开文本图片集

关键词：知识平台；大语言模型；知识图谱；历史场景匹配；数字孪生黄河中图分类号：TP393;TV882.1 文献标志码：A doi：10.3969/j.issn.1000-1379.2025.08.007引用格式：，，，等.数字孪生黄河知识平台建设初探[J].人民黄河，2025，47（8）：32-38.

Preliminary Study on the Construction of the Digital Twin Yellow River Knowledge Platform

ZHU Binhao 1，2，3 ， XIA Runliang4，LI Tao1，2.3，WU Dan 1，2，3 ， DING Yukai 4 ， LI Bing 1，2，3 （ （1.YellowRiver InstituteofHydraulicResearch，YRCC，Zhengzhou4503，China;.YelowRiverLaboratory，Zhengzhou40003， China;3.Engineering Technology Research Center of Henan Province Smart Water Conservancy， Zhengzhou 450o03，China; 4.Information Center，Ministry ofWaterResources，Beijing1OOo53，China）

Abstract：Inordertoimprovethefcencyandintellgencelevelofwateresoucesbusinessdecision-makinginthYelloRiverBasi， thisstudyitroducedlargelanguagemodel（LLM）andknowledgegraphtechnologytobuilddigitaltwinYellwRiverknowledgeplatfor. Aiingattheisuoftatickolegestorage，sutabilttrocssustrucueddataindaicusinseiosoftdiioal knowledgegraphs，awaterconservancyknowedgequestionansweringmechanismandmodelcaling mechanismbasedonLLMwerepro posed，andfuctiossucsuntfistantfloopretio“fourpr”santsstocaleatngndodgee trievalweredvelopd.Aweighedstatisticaladaptivedistacecalulatiotodaproosdtduceteisusfdimensialfer encesandetrvauiteferceiulti-atrutesilitatngPrctalalicatiosowstattlafosigfatl theresponsespeedoffdctroldecisiosandteacuracyofhistcalseemating.Thauacyofhdroloalodelcalse from 43% to 72% ，andthedistance domainiscompressed from50o.00-410o.00 to0.48-0.72，providing eficientandreliable knowledge service support for business decision-making in the Yellow River Basin.

Keywords：knowledge platform；LargeLanguageModel；knowledgegraph；historicalscene matching；digitaltwinYellowRiver

0 引言

随着计算机科学技术的发展，新一批人工智能技术为黄河流域精准决策提供了支撑。（剩余8316字）