基于深度学习的无人机电力巡检目标识别技术研究

打开文本图片集

摘要：本文结合深度学习与无人机技术提出了基于改进YOLOv5的电力设备缺陷检测方法。该方法在绝缘子、杆塔等关键设备缺陷识别上平均准确率达93.7%，较传统方法提升15.2%，为提升电力巡检效率和安全性提供了新的技术方法。

关键词：深度学习；无人机；电力巡检；目标识别；YOLOv5

doi：10.3969/J.ISSN.1672-7274.2025.03.001

中图分类号：TM 7；TP 3；V 279+.2        文献标志码：A           文章编码：1672-7274（2025）03-000-03

Research on Target Recognition Technology for UAV Power Inspection

Based on Deep Learning

LIU Hao1， LU Zhenghe1， BASANG Zerenang1， REN Zexin2， WANG Bangxing2

（1. State Grid Sichuan Electric Power Company Liangshan Power Supply Company， Xichang 615000， China;

2. Zhongke Fangcun Zhiwei （Nanjing） Technology Co.， Ltd.， Nanjing 211135， China）

Abstract： This paper proposes a power equipment defect detection method combining deep learning and unmanned aerial vehicle （UAV） technology based on the improved YOLOv5. This method achieves an average accuracy rate of 93.7% in the identification of defects in key equipment such as insulators and transmission towers， which is 15.2% higher than that of traditional methods， providing a new technology for improving the efficiency and safety of power inspection.

Keywords： deep learning; unmanned aerial vehicle （UAV）; power inspection; target recognition; YOLOv5

0   引言

电力系统是国民经济的重要基础设施，其安全稳定运行对社会发展至关重要。（剩余4094字）