基于YOLOv10深度学习的无人机接触网缺陷检测研究OA

As the core power supply component of electrified railway systems,the catenary's operational state directly af-fects train operation safety.During UAV inspection for catenary component defects,the components are featured with numerous small targets,complex backgrounds and multi-scale characteristics.To address this problem,this paper pro-poses a UAV catenary defect detection scheme based on YOLOv10 deep learning.A dataset including four typical de-fects(insulator damage,conductor strand breakage,connector loosening and foreign matter hanging)is constructed.The CBAM attention mechanism is introduced into the YOLOv10 backbone to suppress background interference and en-hance small-target feature extraction,while the BiFPN is adopted in the neck network to optimize multi-scale feature fusion and reduce missed/false detections.Experimental results show that the mAP@0.5 of the improved model reaches 92.3%,which effectively improves the UAV inspection capability for catenary defects and provides reliable technical support for intelligent catenary detection.