摘要
基于深度学习网络和知识图谱技术,提出了一种关联电力设备多模态信息故障诊断方法。对采集的数据提取融合并构建一个多模态信息的知识图谱,利用YOLOv4算法提取电气设备故障库的先验框参数,将多模态信息知识图谱与YOLOv4算法视觉检测结合并应用到电气主设备进行故障诊断。试验结果证明,所提方法可以实现电气主设备故障智能化诊断,相比其他诊断算法精度提高约18.2%,能够提高电网运行维护效率。
A multi-modal information fault diagnosis method for the associated power equipment was proposed based on deep learning network and knowledge graph technology.The collected data was extracted and fused to construct a knowledge graph of multimodal information.The YOLOv4 algorithm was used to extract the prior box parameters of the electrical equipment fault library.The multimodal information knowledge graph was combined with the YOLOv4 algorithm for visual detection and applied to the diagnosis of main electrical equipment faults.The experimental results show that the proposed method can achieve intelligent diagnosis of main electrical equipment faults,with an accuracy improvement about 18.2%compared to that of other diagnostic algorithms,and can raise the efficiency of power grid operation and maintenance.
作者
尚明远
罗锋
魏艳霞
许陈德
邓祺
Shang Mingyuan;Luo Feng;Wei Yanxia;Xu Chende;Deng Qi(Guangzhou Power Supply Bureau,Guangdong Power Grid Co.,Ltd.,Guangzhou Guangdong 510735,China)
出处
《电气自动化》
2024年第6期100-102,105,共4页
Electrical Automation
关键词
深度学习
知识图谱
多模态
电气设备
智能诊断
deep learning
knowledge graph
multimodal
electrical equipment
intelligent diagnosis
