长江大保护EPC工程管理知识图谱构建方法及应用

钟千有; 刘兴宁; 郭先强; 来亦姝; 张涵需; 吴学明

doi:10.13928/j.cnki.wrahe.2025.S1.011

水利水电技术（中英文） ›› 2025, Vol. 56 ›› Issue (S1) : 58 -66. DOI: 10.13928/j.cnki.wrahe.2025.S1.011

知识驱动的长江大保护智慧EPC管控技术专栏

长江大保护EPC工程管理知识图谱构建方法及应用

钟千有 ¹^,² ,
刘兴宁 ³ ,
郭先强 ⁴ ,
来亦姝 ¹^,² ,
张涵需 ¹^,² ,
吴学明 ³

作者信息 +

Construction and application of a knowledge graph for EPC engineering management in the Yangtze River Protection Project

Qianyou ZHONG ¹^,² ,
Xingning LIU ³ ,
Xianqiang GUO ⁴ ,
Yishu LAI ¹^,² ,
Hanxu ZHANG ¹^,² ,
Xueming WU ³

Author information +

文章历史 +

PDF (10542K)

摘要

在现代工业环境中，EPC管理领域面临着前所未有的复杂数据管理与决策挑战。随着数据体量的不断增长，传统的数据处理方法已难以满足高效、高精度的知识提取与管理需求。探讨了基于深度学习的长江大保护EPC工程管理知识图谱构建技术，通过结合BERT-BiLSTM-CRF模型进行实体识别，并采用改进的BERT-BiGRU-Attention模型进行实体间关系抽取，对EPC管理领域中标准规范、管理文档等信息进行知识图谱构建与系统化分析。研究结果表明，改进的知识抽取模型在EPC工程管理领域中的实体识别率和关系抽取率分别高达93.5%、92.9%,显著提高了EPC知识管理的效率，为知识综合应用打下基础。最后，将知识抽取与EPC管理平台集成，操作界面友好，且支持知识动态更新与知识查询。综上，研究成果为EPC工程管理领域的专业人士提供了一个动态更新、内容丰富的知识图谱资源库，并通过EPC管理平台增强了项目管理过程中的复杂决策能力，推动了EPC项目管理的智能化发展。

Abstract

In the modern industrial environment, the field of water EPC management is faced with unprecedented complex data management and decision-making challenges. With the continuous growth of data volume, traditional data processing method have been difficult to meet the requirements of efficient and high-precision knowledge extraction and management. The construction technology of knowledge graph was discussed for water conservancy EPC enterprise management based on deep learning. Entity identification is carried out by combining BERT-BiLSTM-CRF model, and inter-entity relationship extraction is carried out by using improved BERT-BiGRU-Attention model. The knowledge graph construction and systematic analysis are carried out on the information of standards, specifications and management documents in the field of water conservancy EPC management. The result show that the entity recognition rate and relation extraction rate of the improved knowledge extraction model in the field of water conservancy engineering management are as high as 93.5% and 92.9% respectively, which significantly improves the efficiency of EPC knowledge management and lays a foundation for comprehensive knowledge application. Finally, knowledge extraction was integrated with the water conservancy EPC management platform, which has a friendly operation interface and supports dynamic knowledge update and knowledge query. To sum up, the study findings provide a dynamically updated and rich knowledge graph resource base for professionals in the field of water conservancy engineering management, enhances the complex decision-making ability in the process of EPC project management through the water conservancy EPC management platform, and promotes the intelligent management of water conservancy EPC projects.