Research on UAV Transmission Line Inspection Technology Based on CNN

Abstract

With the rapid development of China's electric power industry, the line patrol inspection is faced with the situation of high operation intensity, long cycle, and bad environment of some lines. The traditional manual patrol inspection method is inefficient and has high risk coefficient, so it is necessary to introduce unmanned aerial vehicle (UAV) technology to maintain the lines efficiently. This paper combines the characteristics of transmission lines, integrates improved convolutional neural network (CNN) with UAV technology, and studies the feature extraction and feature matching of the images taken by UAV patrol inspection combined with the feature target recognition algorithm of transmission lines. By using the cloud scene technology to represent the corresponding model and scene representation of the transmission line, the example proves that the point cloud recognition and cutting extraction of the transmission line conductor and ground wire have reached a high modeling accuracy, which is verified to be feasible by the example. UAV technology can greatly improve the efficiency of line patrol and inspection.