应用EW+YOLOv8耐张线夹DR图像缺陷检测识别技术
Application in DR image defect detection and identification technology of tension clamp based on EW-YOLOv8
汪灵姿 1刘桂雄 1钟飞 2张国才1
作者信息
- 1. 华南理工大学机械与汽车工程学院,广州 510640
- 2. 广东粤电科试验检测技术有限公司,广州 510640
- 折叠
摘要
输耐张线夹在电线路中起到连接导线及运载电流作用,其压接质量直接关系电网安全.针对耐张线夹压接DR缺陷检测存在操作复杂、人员要求高等问题,提出一种EW+YOLOv8应用EW+YOLOv8耐张线夹DR图像缺陷检测识别方案,选用准确性好、实时性佳的YOLOv8n作为检测基准网络,再添加高效通道注意力机制ECA突出缺陷特征图关键信息,并应用基于动态非单调聚焦机制Wise-IoU损失函数代替CIoU损失函数,降低标注样本中低质量锚框影响.基于数据集准备与实验评价指标分析,开展相关消融实验及对比实验,表明了 EW+YOLOv8n在保证较高检测精度时计算速度快、模型参数少,应用于电力耐张线夹压接缺陷检测识别mAP@0.5、FPS分别达到97.4％、50张,能满足工程实际检测需求.
Abstract
The tension clamp plays the role of connecting wires and carrying current in the power line,and its crimping quality is directly related to the safety of the power grid.In order to solve the problems of complex operation and high personnel requirements in the DR defect detection of tension clamp crimping,an EW+YOLOv8 application EW+YOLOv8 tension clamp DR image defect detection and identification scheme was proposed,which selected YOLOv8n with good accuracy and good real-time performance as the detection reference network,and then added the efficient channel attention mechanism ECA to highlight the key information of the defect feature map,and applied the Wise-IoU loss function based on the dynamic non-monotonic focusing mechanism to replace the CIoU loss function.Reduce the impact of low-quality anchor frames in labeled samples.Based on the data set preparation and the analysis of test evaluation indexes,relevant ablation experiments and comparative experiments were carried out,which showed that EW+YOLOv8n had fast calculation speed and few model parameters when ensuring high detection accuracy,and was applied to the detection and identification of power tension clamp crimping defects with mAP@0.5 and FPS of 97.4％and 50 sheets,respectively,which could meet the actual detection needs of the project.
关键词
耐张线夹/缺陷检测/DR图像/YOLOv8Key words
tension clamp/defect detection/DR image/YOLOv8引用本文复制引用
基金项目
2023南网科技专项(GDYDKKJ2023-02)
出版年
2024
NETL
NSTL
万方数据