视觉辅助的高压线路巡检无人机高精度追踪系统  

High-precision Tracking System for High-voltage Line Inspection UAV with Visual Assistance

作　　者：杜家振 赵科 龚淼 DU Jiazhen;ZHAO Ke;GONG Miao(State Grid Shanghai Municipal Electric Power Company,Shanghai 200000,China)

机构地区：[1]国网上海市电力公司,上海200000

出　　处：《微型电脑应用》2025年第2期261-264,共4页Microcomputer Applications

摘　　要：通过灰度化、几何校正和对比度亮度调整等图像预处理技术优化无人机(UAV)视觉系统捕获的数据,以提升后续特征识别的准确性和效率。针对高压线路巡检,采用改进的YOLOv4算法和路径聚合网络,提升目标定位和类别预测的准确性。实验结果表明,这些改进使得最佳精确率和召回率为95.1%和95.3%,降落偏差的减少验证了系统在提高巡检精度方面的有效性。无人机巡检仅需12 min,减少了传统方法的风险,有效提高了效率,实现了安全、快速且高效的巡检模式。The study optimizes data captured by unmanned aerial vehicle(UAV)vision systems through image preprocessing techniques such as gray scaling,geometric correction and adjustments to contrast and brightness,thereby enhancing the accuracy and efficiency of subsequent feature recognition.Specifically for high-voltage line inspection,an improved YOLOv4 algorithm and a path aggregation network are employed to enhance the accuracy of target localization and category prediction.Experimental results indicate that these improvements have result in an optimal accuracy rate of 95.1%and an optimal recall rate of 95.3%,respectively.Furthermore,significant reductions in landing deviations are observed during flight tests,which enhances the precision of inspection.The UAV inspection requires only 12 minutes,significantly improving efficiency and reducing the risks associated with traditional methods,thus achieving a safe,rapid and efficient inspection mode.

关 键 词：YOLOv4 无人机巡检 高压线路 视觉辅助 

分 类 号：TM775[电气工程—电力系统及自动化]