山东科学 ›› 2021, Vol. 34 ›› Issue (2): 81-89.doi: 10.3976/j.issn.1002-4026.2021.02.011

• 其他研究论文 • 上一篇    下一篇

基于Jetson-TX2的输电线路设备实时巡检系统

杨学杰1,陈文栋1,许荣浩2*,李宋林1,李建业1   

  1. 1.国网淄博供电公司,山东 淄博 2550222.国网智能科技股份有限公司,山东 济南 250002
  • 收稿日期:2020-04-07 出版日期:2021-04-13 发布日期:2021-04-14
  • 通信作者: 许荣浩,男,工程师,研究方向为输电线路智能识别。Tel:15615713601,E-mail:lnint_xuronghao@163.com E-mail:zbsdyjs@163.com
  • 作者简介:杨学杰,(1970—),男,工程师,研究方向为输电线路运维检修。E-mail:zbsdyjs@163.com
  • 基金资助:
    国网山东省电力公司资助科技项目(2020A-108

Real-time inspection system for transmission line equipment based on Jetson-TX2

YANG Xue-Jie1,CHEN Wen-Dong 1,XU Rong-Hao2*, LI Song-Lin1, LI Jian-Ye   

  1. 1. State Grid Zibo Electric Power Company, Zibo 255032,China;2. State Grid Intelligence Technology Co., Ltd., Jinan 250002, China
  • Received:2020-04-07 Online:2021-04-13 Published:2021-04-14

摘要: 针对输电线路及设备巡检效率低的问题,设计了一种基于Jetson-TX2的输电线路设备实时巡检系统。该系统包括基于YOLO v3算法的Jetson-TX2主控模块和云台相机控制模块。Jetson-TX2主控模块通过TensorRT加速库,对YOLO v3算法模型进行优化加速,完成视频流目标实时识别与定位;采用 PID算法控制云台(PTZ)相机,实现设备的高清图像采集。该系统对输电线路设备整体识别准确率达95%,可实现对视频流的实时检测,有效提高输电线路巡检效率。

关键词: 输电线路巡检, 视频实时检测, 算法加速, 云台控制, Jetson-TX2, YOLO v3, 目标定位

Abstract: To solve the problem of low efficiency of inspection of transmission line equipment, this paper designs a real-time inspection system for transmission line equipment based on Jetson-TX2. The main control module of Jetson-TX2 based on YOLO v3 and a Pan/Tilt/Zoom(PTZ) camera control module are included in this system. The YOLO v3 algorithm model is accelerated by TensorRT, which is transplanted in Jetson-TX2 to complete the real-time identification and positioning of the device target in the video stream. The PTZ camera is controlled by Jetson-TX2 to realize the high-definition image acquisition of the device through the PID algorithm. The system's overall recognition accuracy for transmission line equipment can reach 95%, which can realize real-time video detection. It can effectively improve the efficiency of transmission line inspection.

Key words: transmission line maintenance, real-time video detection, algorithm acceleration, PTZ control, Jetson-TX2, YOLO v3; target position

中图分类号: 

  • TM755