• CSCD核心库收录期刊
  • 中文核心期刊
  • 中国科技核心期刊

电力建设 ›› 2022, Vol. 43 ›› Issue (3): 66-74.doi: 10.12204/j.issn.1000-7229.2022.03.008

• 人工智能在电网故障诊断和定位中的应用·栏目主持 王小君教授、罗国敏副教授、石访副教授· • 上一篇    下一篇

基于深度学习和知识图谱的变电站设备故障智能诊断

肖发龙1(), 吴岳忠1(), 沈雪豪1(), 何震凯2(), 秦烨3()   

  1. 1.湖南工业大学轨道交通学院,湖南省株洲市412007
    2.湖南旭瑞智能技术有限公司,湖南省株洲市412007
    3.国网江苏省电力有限公司仪征市供电公司,江苏省扬州市211400
  • 收稿日期:2021-05-19 出版日期:2022-03-01 发布日期:2022-03-24
  • 通讯作者: 吴岳忠 E-mail:1544149279@qq.com;wuyuezhong@hut.edu.cn;2515847126@qq.com;13974117836@139.com;qingye1981@163.com
  • 作者简介:肖发龙(1996),男,硕士研究生,主要研究方向为视觉理解、变电站设备故障智能诊断等,E-mail: 1544149279@qq.com;
    沈雪豪(1995),男,硕士研究生,主要研究方向为深度学习、变电站设备故障智能诊断等,E-mail: 2515847126@qq.com;
    何震凯(1977),男,硕士,工程师,主要从事工业互联网、自动化控制与过程检测工作,E-mail: 13974117836@139.com;
    秦烨(1981),女,本科,工程师,主要从事电力信息通信技术工作,E-mail: qingye1981@163.com
  • 基金资助:
    国家重点研发计划项目(2019YFE0122600);湖南省自然科学基金项目(2021JJ50050)

Intelligent Fault Diagnosis of Substation Equipment on theBasis of Deep Learning and Knowledge Graph

XIAO Falong1(), WU Yuezhong1(), SHEN Xuehao1(), HE Zhenkai2(), QIN Ye3()   

  1. 1. College of Traffic Engineering, Hunan University of Technology, Zhuzhou 412007, Hunan Province, China
    2. Hunan Xurui Intelligent Technology Co., Ltd., Zhuzhou 412007, Hunan Province, China
    3. State Grid Jiangsu Electric Power Company Yizheng Power Supply Company,Yangzhou 211400, Jiangsu Province, China
  • Received:2021-05-19 Online:2022-03-01 Published:2022-03-24
  • Contact: WU Yuezhong E-mail:1544149279@qq.com;wuyuezhong@hut.edu.cn;2515847126@qq.com;13974117836@139.com;qingye1981@163.com
  • Supported by:
    National Key R&D Program of China(2019YFE0122600);Natural Science Foundation of Hunan Province(2021JJ50050)

摘要:

及时发现并诊断变电站运行中设备存在的问题,是保障电网安全运行的关键手段之一。基于深度网络与知识图谱技术,提出一种关联变电站设备多模态信息的故障智能诊断方法。利用深度学习技术和知识图谱方法对采集的多模态数据进行知识提取和融合,构建一个多模态信息融合的语义知识图谱;使用YOLOv4算法对故障样本聚类并提取先验框参数;将多模态知识图谱和YOLOv4视觉检测相结合,应用到变电站场景中,实现变电站设备的自主预警诊断。实验表明,该模型可以实现故障诊断决策智能化的目标,从而提高电网的日常运行、维护和管理效率。

关键词: 多模态, 智能诊断, 深度学习, 知识图谱

Abstract:

Finding and diagnosing problems in the operation of substation equipment in time is one of the key means to ensure the safe operation of power grids. On the basis of the research of deep network and knowledge graph, this paper proposes an intelligent fault diagnosis method that correlates multi-modal information of substation equipment. Deep learning technology and knowledge graph method are used to extract and fuse knowledge from the collected multi-modal data, and a semantic knowledge graph of multi-modal information fusion is constructed. YOLOv4 algorithm is used to cluster fault samples and extract prior bounding box parameters. The multi-modal knowledge map and YOLOv4 visual inspection are combined together and applied to the substation scene to realize the autonomous early warning and diagnosis of the substation equipment. Experiments show that the model can achieve the goal of intelligent decision-making of fault diagnosis, thereby improving the efficiency of daily operation, maintenance and management of power grid.

Key words: multi-modal, intelligent diagnosis, deep learning, knowledge graph

中图分类号: