电力系统架空线巡线机器人对于提高架空输电线的维护效率进而保障电力系统的安全稳定运行具有重要作用。针对架空线巡线机器人巡线过程中需要有效识别线路中的附属障碍物并采取相应障碍规避动作的问题，文章研究了基于深度学习的巡线机器人架空线附属障碍物识别方法。论述了基于深度学习的架空线巡线机器人系统的整体结构，在整体系统结构的基础上分析了YOLOv8(you only look once version 8)模型的结构及其应用，进而基于增强的架空线障碍物数据集对所述方法进行了有效性验证。实验结果表明，文中所述基于YOLOv8的巡线机器人障碍物识别模型具有更快的识别速度和更高的识别率，能够更好满足巡线机器人的避障需求。
英文摘要:
The overhead line patrol robot plays an important role in improving the line maintenance efficiency and ensuring the safe and stable operation of power system. Aiming at the problem that the overhead line patrol robot needs to effectively identify affiliated obstacles and adopt corresponding obstacle-avoidance actions, this paper studies the deep learning-based affiliated obstacle identification method of overhead transmission line for patrol robots. The overall structure of overhead line patrol robot based on deep learning is discussed. The YOLOv8(you only look once version 8) model and its application in obstacle identification are analyzed based on the overall structure. Furthermore, the effectiveness of the proposed method is verified using an augmented obstacle data set of overhead line. Experimental results show that the proposed overhead line obstacle recognition method has a faster recognition speed and a higher recognition rate, which can meet the obstacle avoidance needs of patrol robots.