数字孪生水利建设中的人工智能大模型应用探索  

Exploration of artificial intelligence large model applications in digital twin water conservancy construction

作　　者：舒全英 马媛 陈亮[3] 李磊[4] 郭磊[5] 吴健柏 Shu Quanying;Ma Yuan;Chen Liang;Li Lei;Guo Lei;Wu Jianbai

机构地区：[1]浙江省水利水电勘测设计院有限责任公司,杭州310002 [2]太湖流域管理局水文局(信息中心),上海200434 [3]黄河水利委员会信息中心,郑州450000 [4]水利部信息中心,北京100053 [5]广东省水利水电科学研究院,广州510635 [6]新疆维吾尔自治区水利厅网络信息中心,乌鲁木齐830099

出　　处：《中国水利》2025年第6期14-30,共17页China Water Resources

基　　金：国家重点研发计划“流域多源异构信息智能融合与数据底板构建”(2023YFC3209201);国家自然科学基金长江水科学研究联合基金计划“基于数字孪生技术的长江下游感潮河网地区多目标调度研究项目”(U2340221)。

摘　　要：人工智能大模型可为数字孪生水利建设提质增效提供新动能。按照“分析定位、探索路线、摸清需求、落地应用、推广模式”思路,在分析数字孪生水利建设面临的问题挑战、人工智能大模型发展及行业应用情况、大模型在数字孪生水利建设应用的必要性与可行性基础上,从技术、业务、管理视角对大模型在数字孪生水利业务场景中的落地应用进行了探索思考。以“场景数字化、模拟智能化、决策精准化”为技术路径,梳理了动态数字场景构建、复杂系统智能模拟、人机协同精准决策等关键技术;按照“2+N”业务需求清单并以防洪预报、预警、预演、预案“四预”应用、网络安全防护场景为例,阐述人工智能大模型在具体业务场景落地应用的思路和应用步骤;提出了人工智能大模型在数字孪生水利应用中的共建共享模式,给出“共建共享、统分结合、合作推广”的建设思路。研究成果可为人工智能大模型在数字孪生水利建设中的功能定位和业务场景落地应用提供参考和借鉴。Artificial intelligence(AI)large models can provide new momentum for enhancing the quality and efficiency of digital twin water conservancy construction.Following the approach of“positioning analysis,route exploration,needs assessment,implementation,and model promotion”,this study analyzes the challenges in digital twin water conservancy,the development and industry applications of AI large models,and the necessity and feasibility of their application in this field.From the perspectives of technology,business,and management,the study explores the application of AI large models in digital twin water conservancy scenarios.Using the technical path of“scenario digitization,intelligent simulation,and precise decision-making”,the study outlines key technologies such as dynamic digital scenario construction,intelligent simulation of complex systems,and precise human-machine collaborative decision-making.Based on a“2+N”business demand framework,the application routes and steps of AI large models are illustrated through specific scenarios,such as the“four pre”(forecasting,warning,rehearsal,and planning)flood control applications and network security protection.The study proposes a co-construction and sharing model for the application of AI large models in digital twin water conservancy,offering a“co-construction and sharing,unified and distributed,collaborative promotion”development approach.The findings provide references for the functional positioning and business scenario implementation of AI large models in digital twin water conservancy construction.

关 键 词：数字孪生水利 人工智能 大模型 智能模拟 人机协同 

分 类 号：TV[水利工程] TP39[自动化与计算机技术—计算机应用技术]