嵌入式智慧电网绝缘子缺失检测系统设计与实现  

Design and Implementation of Embedded Smart Grid Insulator Missing Detection System

作　　者：焦双健[1] 宫中飞 JIAO Shuangjian;GONG Zhongfei(Faculty of Engineering,Ocean University of China,Qingdao 266100,China)

机构地区：[1]中国海洋大学工程学院,山东青岛266100

出　　处：《电视技术》2022年第12期46-50,62,共6页Video Engineering

摘　　要：针对输电线路绝缘子长期暴露在自然环境中容易遭到侵蚀而自爆缺失的问题,设计一种基于深度学习的嵌入式绝缘子缺失检测系统。为了提升对于小目标物体的检测精度并增强模型的泛化能力,对YOLOv5s算法融合CBAM注意力机制,并将PANet替换为BiFPN。改进后的CBBi-YOLOv5s算法检测m AP值为95.2%,相比于YOLOv5s算法提升了4.6%。将训练好的最优模型部署至英伟达Jetson TX2并搭载在无人机上,可以对缺失的绝缘子进行自动检测,并将绝缘子缺失信息传输至电力系统维护部门,实现绝缘子缺失的自动化巡检。Aiming at the problem that transmission line insulators are easily eroded and self-detonated due to long-term exposure to natural environment, this paper designs an embedded insulator missing detection system based on deep learning. In order to improve the detection accuracy of the model in small targets and complex environments, the YOLOv5s algorithm integrates the CBAM attention mechanism, and improves PANet to BiFPN, and the improved algorithm model mAP is 95.2%, which is 4.6% higher than the YOLOv5s algorithm, and the optimal weight of the model is deployed to Jetson TX2 and mounted on the UAV to perform online real-time inspection of missing insulators, which has high application value for the protection of transmission lines.

关 键 词：深度学习 YOLOv5s 嵌入式系统 智慧电网 注意力机制 

分 类 号：TP311.5[自动化与计算机技术—计算机软件与理论]