摘要
为解释全国电力安全事故的演化规律并为预控事故的发生,提出电力安全事故事理图谱的构建与推演方法。以国家能源局发布的全国电力安全生产情况为数据源,使用规则模板提取事件槽对;利用Word2vec训练词向量,通过事件槽相似度计算泛化事件槽,并以热力图矩阵记录泛化权重;采用Neo4j构建事故事理图谱,以泛化权重和事件链的方向预测事故走向。结果表明:方法能够较好地揭示事故演化的关键节点、预演电力安全事故发展路径,为事故的预控提供1种有效的知识服务方案。
In order to explain the evolution laws of power safety accidents across the country and prevent the occurrence of accidents,a construction and deduction method of event evolutionary graph for power safety accidents was proposed.Taking the national power work safety situation released by the National Energy Administration as a data source,the rule template was used to extract the event slot pair.The Word2vec was used to train the word vectors,then the generalization event slot was calculated by the similarity of event slot,and the generalized weights were recorded by the thermodynamic diagram matrix.The event evolutionary graph of accidents was constructed by using Neo4j,and the accident trend was predicted by using the generalized weights and the direction of event chain.The results showed that the method could better reveal the key nodes of accident evolution,preview the development path of power safety accidents,and provides an effective knowledge service solution for the pre-control of accidents.
作者
林穿
徐启峰
黄奕钒
LIN Chuan;XU Qifeng;HUANG Yifan(School of Electrical Engineering and Automation,Fuzhou University,Fuzhou Fujian 350108,China)
出处
《中国安全生产科学技术》
CAS
CSCD
北大核心
2021年第10期39-45,共7页
Journal of Safety Science and Technology
基金
国家自然科学基金项目(51977038)。
关键词
事理图谱
电力安全事故
事故预控
演化
event evolutionary graph
power safety accident
accident pre-control
evolution
