基于深度学习与毫米波雷达的输电线路防外破系统研究

Research on Transmission Line Anti-external Damage System Based on Deep Learning and Millimeter-wave Radar

为提高复杂背景下的外破隐患识别率,采取深度学习与毫米波雷达融合的方式,将毫米波雷达的高精度距离信息与监拍相机采集的丰富环境信息相结合,通过传感器标定,实现毫米波雷达和监拍相机的时间同步与空间融合,基于YOLOv5目标识别算法对监拍相机捕获的图像进行深度处理,准确地识别图像中的目标物体,并输出相应的检测框和检测结果。同时,将毫米波雷达探测到的目标点映射到图像中,生成感兴趣区域,使得算法能够更专注于这些可能存在外破隐患的区域,提高了检测的效率和准确性。最后,通过目标融合算法,结合监拍相机和毫米波雷达识别检测结果,对输电线路外破隐患目标进行了精确识别和分类,有效地解决了光线、距离等因素对传统监拍装置的影响,实现了对复杂背景下输电线路外破隐患目标的精准识别。

In order to improve the identification rate of external hidden damage targets under complex backgrounds,a fusion approach combining deep learning and millimeter-wave radar is adopted.and it combines the high-precision distance information from millimeter-wave radar with the rich environmental information captured by monitoring camera.Through sensor calibration,time synchronization and spatial integration between millimeter-wave radar and monitoring camera are achieved.Based on the YOLOv5 target identification algorithm,the images captured by the monitoring camera are processed in-depth to accurately identify the target objects in the images and output corresponding detection frames and results.Meanwhile,the target points detected by the millimeter-wave radar are mapped to the images to generate regions of interest,enabling the algorithm to focus more on these areas that may contain external hidden damage,thus improving the efficiency and accuracy of detection.Finally,through the target fusion algorithm and combining the identification and detection results of monitoring camera and millimeter-wave radar,the external hidden damage targets of the power transmission line are precisely identified and classified,effectively solving the impact of light,distance and other factors on traditional monitoring devices,and achieving precise identification of external hidden damage targets of the power transmission line under complex backgrounds.