架空输电线路巡检机器人系统设计  

Design of inspection robot system for overhead transmission line

作　　者：王鹏[1] 盛宇军 高笃良 蔡永挚 闫付乐 Wang Peng;Sheng Yujun;Gao Duliang;Cai Yongzhi;Yan Fule(State Grid Beijing Electric Power Company,Beijing 100051,China)

机构地区：[1]国网北京市电力公司,北京100051

出　　处：《电子测量技术》2022年第21期117-122,共6页Electronic Measurement Technology

基　　金：国家自然科学基金(62005077);河南省科技攻关计划(工业领域)项目(192102210141);河南省高等学校青年骨干教师培养计划(2019GGJS082)项目资助。

摘　　要：架空输电线路机器人可以代替人工执行电力巡检任务,解决人工巡检中效率低、劳动强度大、智能化程度不足等问题。提出了档间巡检的设计思路,开发了低成本架空输电线路巡检机器人系统,通过设计机械结构、优化识别算法等方式,实现了机器人的自主行走、无线充电、图像采集、数据处理、智能巡检等功能。在软件方面基于C#开发了专用程序及界面,并建立了基于架空输电线金具及缺陷数据集,在YOLOv4目标检测算法基础上,通过数据增强、边界框优化和模型结构改进,建立YOLOv4-M模型,优化了电力金具识别检测性能,提升了系统应用的环境适应性。通过机器人在线路上的运行和测试,识别算法在速度可以达到45fps,平均检测精度达到97.6%。Overhead transmission line robot can replace artificial power inspection tasks, solve the low efficiency in manual inspection Labor intensity The problem such as intelligent degree is insufficient Put forward between your checking design train of thought, through the design of mechanical structure optimization identification algorithm, realize the independent walking of the robot Wireless charging image acquisition intelligent inspection In the aspect of software, a special program and interface are developed based on C#, and a data set based on overhead transmission line hardware and defect is established. On the basis of YOLOv4 target detection algorithm, through data enhancement The boundary box optimization and model structure improvement, the establishment of YOLOV4-M model, optimize the performance of power hardware identification and detection, improve the environmental adaptability of the system application through the robot online operation and testing, the identification algorithm in the speed of 45 fps, the average detection accuracy of 97.6%.

关 键 词：架空输电线路 自主移动机器人 目标检测 自主巡检 无线充电 

分 类 号：TP242.6[自动化与计算机技术—检测技术与自动化装置]