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。
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.
作者
杨学杰
陈文栋
许荣浩
李宋林
李建业
YANG Xue-Jie;CHEN Wen-Dong;XU Rong-Hao;LI Song-Lin;LI Jian-Ye(State Grid Zibo Electric Power Company, Zibo 255032,China;State Grid Intelligence Technology Co., Ltd., Jinan 250002, China)
出处
《山东科学》
CAS
2021年第2期81-89,共9页
Shandong Science
基金
国网山东省电力公司资助科技项目(2020A-108)。