人民黄河2025，Vol.47Issue(8)：32-38,7.DOI:10.3969/j.issn.1000-1379.2025.08.007
数字孪生黄河知识平台建设初探
Preliminary Study on the Construction of the Digital Twin Yellow River Knowledge Platform
摘要
Abstract
In order to improve the efficiency and intelligence level of water resources business decision-making in the Yellow River Basin,this study introduced large language model(LLM)and knowledge graph technology to build digital twin Yellow River knowledge platform.Aiming at the issues of static knowledge storage,insufficient ability to process unstructured data in dynamic business scenarios of traditional knowledge graphs,a water conservancy knowledge question answering mechanism and model calling mechanism based on LLM were pro-posed,and functions such as document office assistant,flood prevention"four pre"assistants,historical scene matching and knowledge re-trieval were developed.A weighted statistical adaptive distance calculation method was proposed to reduce the issues of dimensional differ-ences and extreme value interference in multi-attribute similarity matching.Practical application shows that the platform significantly improves the response speed of flood control decisions and the accuracy of historical scene matching.The accuracy of hydrological model calls increase from 43%to 72%,and the distance domain is compressed from 500.00-4 100.00 to 0.48-0.72,providing efficient and reliable knowledge service support for business decision-making in the Yellow River Basin.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Key words
knowledge platform/Large Language Model/knowledge graph/historical scene matching/digital twin Yellow River分类
信息技术与安全科学引用本文复制引用
祝宾皓,夏润亮,李涛,吴丹,丁昱凯,李冰..数字孪生黄河知识平台建设初探[J].人民黄河,2025,47(8):32-38,7.基金项目
国家重点研发计划项目(2023YFC3209203,2023YFC3209305) （2023YFC3209203,2023YFC3209305）
国家自然科学基金资助项目(U2453320) （U2453320）
黄河水利科学研究院基本科研业务费专项(HKY-JBYW-2024-12,HKY-YF-2024-05,HKY-JBYW-2023-20,HKY-JBYW-2025-17) （HKY-JBYW-2024-12,HKY-YF-2024-05,HKY-JBYW-2023-20,HKY-JBYW-2025-17）
青海省基础研究计划项目(2025-ZJ-721) （2025-ZJ-721）
