摘要
绝缘子是输电线路的重要元件,绝缘子缺陷会增大输电线路的故障停运风险,因此,对绝缘子缺陷状况的早期判别十分重要。本文提出一种基于YOLOv4模型的玻璃绝缘子自爆缺陷辨识方法。首先,通过无人机采集及数据增强获取大量详实的现场绝缘子图像;其次,通过采用迁移学习的训练策略训练YOLOv4网络并改进网络的输入图像以提高辨识的准确性;最后,通过实验验证改进策略提高了网络性能。实验结果表明,所提的方法可准确、有效地实现对绝缘子缺陷的辨识。
Insulators are important components of transmission lines,and their failure will increase the risk of outages of transmission lines.Therefore,it is very important to distinguish the defects of insulators in the early stage.A method for identifying self-explosion defects of glass insulators based on the YOLOv4 model is proposed in this paper.Firstly,a large number of detailed images of field insulators are obtained through unmanned aerial vehicle acquisition and data enhancement.Secondly,the training strategy of transfer learning is adopted to train the YOLOv4 network and improve the input image of the network to achieve the accuracy of identification.Finally,the improvement of network performance is verified by experiments.The experimental results show that the proposed method can accurately and effectively realize the identification of insulator defects.
作者
周宸
高伟
郭谋发
ZHOU Chen;GAO Wei;GUO Moufa(College of Electrical Engineering and Automation,Fuzhou University,Fuzhou 350108)
出处
《电气技术》
2021年第5期38-42,49,共6页
Electrical Engineering
关键词
输电线路
绝缘子
缺陷辨识
YOLOv4网络
多阶段迁移学习
transmission line
insulators
defect identification
YOLOv4 network
multi-stage transfer learning
