摘要
在运绝缘子的憎水性检测是电力巡检工作中的难点,为此设计一种基于无人机平台和深度学习模型的绝缘子憎水性检测系统。首先由喷雾无人机对绝缘子表面实施喷雾,然后采集绝缘子图像并通过深度学习模型对图像进行处理,完成憎水性分类检测。系统通过无人机遥控器调用边缘检测设备,实现图像互传,并且在遥控器上完成用户交互界面,与现有的巡检工作之间可友好适配,所提出的深度学习分类模型采用FasterNet主干网络融合通道注意力机制(efficient channel attention,ECA),具有良好的检测性能。通过数据集验证和现场试验证明,所提的系统运行良好,憎水性平均判别准确率达98.4%,可满足绝缘子日常巡检的要求。
Hydrophobicity detection of the insulators in operation is a difficulty in power inspection work.To address this issue,an insulator hydrophobicity detection system based on UAV platform and the deep learning model is designed.First,the spray UAV implements spray on the insulator surface,then collects the insulator images and processes the images through the deep learning model to complete hydrophobicity classification detection.The system calls the edge detection device through the drone remote control to achieve image exchange,and completes the user interaction interface on the remote control.The system can be friendly adapted to the existing inspection work.The proposed deep learning classification model adopts the FasterNet backbone network fused with efficient channel attention(ECA)mechanism and has good detection performance.The dataset validation and field experiments prove that the proposed system runs well with an average discrimination accuracy of 98.4%for hydrophobicity,which can meet the requirements of daily inspection of insulators.
作者
杨传凯
边少聪
辛蕾
张雷
杜建超
YANG Chuankai;BIAN Shaocong;XIN Lei;ZHANG Lei;DU Jianchao(State Grid Shaanxi Electric Power Research Institute,Xi’an,Shaanxi 710100,China;School of Telecommunications Engineering,Xidian University,Xi’an,Shaanxi 710071,China)
出处
《广东电力》
北大核心
2025年第11期46-53,共8页
Guangdong Electric Power
基金
国网陕西省电力有限公司科技项目(5226KY24000H)。
关键词
电力巡检
绝缘子
憎水性
无人机
深度学习
power inspection
insulator
hydrophobicity
unmanned aerial vehicle(UAV)
deep learning
