面向数字孪生低压配电网的基于Gabor-YOLO算法的架空线高效识别方法研究

从各种测量装置中实时获取和更新电力线路运行状态是低压配电网数字孪生的基础,获取电力线路运行状态的首要任务是对电力线进行精准识别。本文针对低压配电架空线路航拍图像背景复杂、遮挡严重、目标特征微弱的问题,提出了基于Gabor-YOLO的算法,用于低压架空电力线的高效提取。首先,对图像进行灰度化和高斯滤波等预处理后利用改进后的Gabor算子进行特征提取,在图像中分割出前景区域；其次,在改进YOLO网络模块中,对电力线及辅助目标进行定位和识别最终提取出电力线。实验结果表明,改进的Gabor算子可以快速提取出图像前景区域,改进的YOLO网络可以在前景区域中准确提取出电力线。实验结果证明所提方法相比于yolov4等方法具有最高的准确率和提取速度,mAP值可达93.6%,满足实际工作需要。

Research on efficient identification method of overhead lines based on Gabor-YOLO algorithm for digital twin low-voltage distribution network

Real-time acquisition and update of power line operating status from various measurement devices is the basis of the digital twin of low-voltage distribution networks. The primary task of obtaining power line operating status is to accurately identify power lines. This paper proposes an algorithm based on Gabor-YOLO for the efficient extraction of low-voltage overhead power lines, aiming at the problems of complex background, serious occlusion and weak target features in aerial images of low-voltage power distribution overhead lines. First, after preprocessing the image, such as grayscale and Gaussian filtering, the improved Gabor operator is used for feature extraction, and the foreground area is segmented in the image; secondly, in the improved YOLO network module, the power lines and auxiliary targets are located. And identify the final extracted power lines. The experimental results show that the improved Gabor operator can quickly extract the foreground area of the image, and the improved YOLO network can accurately extract the power lines in the foreground area. The experimental results show that the proposed method has the highest accuracy and extraction speed compared with other methods such as yolov4, and the mAP value can reach 93.6%, which meets the needs of practical work.