基于知识图谱的变电站安全隐患动态分析方法  

Dynamic Analysis Method for Hidden Dangers in Substation Based on Knowledge Graph

作　　者：郭素芹 郑建宁 陈坤 林瑞安 张勃波 宗鑫 GUO Suqin;ZHENG Jianning;CHEN Kun;LIN Rui’an;ZHANG Bobo;ZONG Xin(Fujian Yili Electric Power Technology Co.,Ltd,Fuzhou 350003,China)

机构地区：[1]福建亿力电力科技有限责任公司,福州350003

出　　处：《电力系统及其自动化学报》2021年第12期125-133,共9页Proceedings of the CSU-EPSA

摘　　要：由于变电站安全隐患非结构化的文本格式导致无法进行隐患知识提取与推理,难以挖掘潜在隐患间的关系和规律,提出一种变电站安全隐患动态分析方法。首先,对非结构化的隐患文本数据解析抽取,构建了基于ElasticSearch弹性分布式隐患数据搜索引擎。其次,利用隐马尔科夫模型对引擎内数据进行分词训练,结合维特比算法求解隐藏的状态序列以完成隐患实体分词标注。最后,采用Neo4j图数据库动态生成变电站安全隐患知识图谱。以某地区变电站安全隐患数据进行算例分析,证明该方法的有效性。Due to the unstructured text format of hidden dangers in a substation,it is impossible to extract and reason about the hidden dangers,and it is difficult to explore the relationships and laws for potential hidden dangers.To solve this problem,a dynamic analysis method for hidden dangers in the substation is proposed.First,the unstructured hidden danger text data is analyzed and extracted,and a flexible distributed hidden danger data search engine based on ElasticSearch is built.Second,the hidden Markov model is used to train the data in the engine,and the Viterbi algorithm is combined to solve the hidden state sequence in order to complete the word segmentation labeling of hidden danger entities.Finally,the Neo4j graph database is used to dynamically generate a knowledge map of hidden dangers in the substation.An example analysis is performed based on the hidden danger data of a substation in a certain area,and results prove the effectiveness of the proposed method.

关 键 词：变电站安全 知识图谱 搜索引擎 隐马尔科夫分词模型 图数据库 

分 类 号：TM73[电气工程—电力系统及自动化]