面向土石坝历史病险特征的知识图谱构建与应用OA

Earth-rock dams are the most widely distributed and numerous dam type in China,and their safe operation is directly related to basin-level flood control and water supply security.The Haihe River basin is characterized by a high density of earth-rock dam projects,generally aging dam structures,complex types of defects and risks,and increasingly frequent extreme weather events,all of which pose severe challenges to engineering safety.A large volume of unstructured or semi-structured textual materials has been accumulated in this basin,including defect and risk treatment records,safety evaluation reports,and operation logs,which encompass extensive engineering practice knowledge and treatment experience.However,due to the limitations of decentralized storage,semantic heterogeneity,and implicit associations,traditional manual review methods have difficulty in achieving effective reuse of this empirical knowledge.To address this issue,this paper introduced knowledge graph technology and,based on a large earth-rock dam in the Haihe River basin,and constructed a defect-and-risk knowledge graph for earth-rock dams using multi-year treatment records,organized around the framework of"engineering component,defect and risk characterization,and treatment measure".Unlike existing knowledge graph studies that primarily rely on static ontology descriptions,this paper proposed a dynamic ontology model incorporating multi-dimensional attributes of"time,space,and state",enabling semantic modeling of the entire process from defect and risk discovery and development to treatment.In the knowledge extraction phase,the approach integrated a large language model with domain rule constraints,balancing text comprehension capability with standardized domain terminology expression to improve extraction accuracy and consistency.Finally,the Neo4j graph database was employed for knowledge storage and retrieval.The application results demonstrate that the constructed knowledge graph enables rapid positioning of earth-rock dam defects and risks and significantly improves knowledge reuse efficiency through associated retrieval of defect and risk type distribution,treatment response time,and actual effectiveness,thereby providing technical support for intelligent decision-making in the reinforcement of reservoir dams.