摘要
梯级泵站群运维知识通常以文本形式记录,存在数据量大、碎片化严重等特点,难以进行数据挖掘应用。为此,提出一种从非结构化数据中智能挖掘泵站群运维知识的方法。结合双向编码器表示(BERT)、双向长短期记忆网络(BiLSTM)、条件随机场(CRF)以及字词向量(Word Vector),建立梯级泵站群运维实体识别模型,识别文本中的实体类型;利用BERT-BiGRU-Attention模型抽取实体间关系;利用Neo4j数据库构建梯级泵站群运维知识图谱,将大规模的非结构化数据转换为结构化知识,明确知识之间的逻辑关系。以南水北调东线山东干线梯级泵站群为例,进行模型及图谱的实际应用。应用情况表明;所提出方法能够准确识别与抽取泵站群运维文本中的实体与关系,基于知识图谱实现知识的可视化与高效利用,以及对梯级泵站群运行方案进行智能推荐,为提高泵站群运维效率提供了支撑。
Operational knowledge of cascade pumping station groups is typically recorded in textual form,characterized by large volumes and severe fragmentation,making data mining applications difficult.To address this issue,a method for intelligently mining pumping station group operational knowledge from unstructured data was proposed.By integrating Bidirectional Encoder Representations from Transformers(BERT),Bidirectional Long Short-Term Memory networks(BiLSTM),Conditional Random Fields(CRF),and Word Vectors,an entity recognition model for the operation and maintenance of cascade pumping station groups was established to identify entity types in the text.The BERT-BiGRU-Attention model was used to extract relationships between entities.By utilizing the Neo4j database,a knowledge graph for the operation and maintenance of cascade pumping station groups was constructed,converting large-scale unstructured data into structured knowledge and clarifying the logical relationships between knowledge elements.The model and knowledge graph were practically applied by using the cascade pumping station showes of the Shandong main line in the eastern route of the South-to-North Water Diversion Project as an example.Application situation showes that the proposed method can accurately identify and extract entities and relationships in the operation and maintenance text of pumping station groups.Based on the knowledge graphs,it realizes the visualization and efficient utilization of knowledge,intelligently recommends optimization schemes for the operation of cascade pumping station groups,and provides support for improving the efficiency of pumping station group operation and maintenance.
作者
谢津平
李喆
钟千有
付超
刘敏
XIE Jinping;LI Zhe;ZHONG Qianyou;FU Chao;LIU Min(China Water Resources Beifang Investigation,Design and Research Co.,Ltd.,Tianjin 300222,China;State Key Laboratory of Hydraulic Engineering Intelligent Construction and Operation,Tianjin University,Tianjin 300072,China;School of Civil Engineering,Tianjin University,Tianjin 300072,China)
出处
《人民黄河》
北大核心
2025年第11期133-138,共6页
Yellow River
基金
水利部重大科技项目(SKS-2022133)
天津大学自主创新基金资助项目(2023XJD-0065)
中国长江三峡集团有限公司企业科研项目(202103551)。
关键词
梯级泵站群
运维
实体识别
关系抽取
知识图谱
智能
南水北调东线山东干线
cascade pumping station group
operation and maintenance
entity recognition
relation extraction
knowledge graph
intelligence
Shandong main line in eastern route of South-to-North Water Diversion Project
