Application of Deep Convolutional Neural Network in Remote Protection of Transmission Lines

In order to alleviate the pressure of line operation and maintenance, it is necessary to use advanced image recognition methods to identify hidden line hazards. In this paper, we extract features from hidden targets and perform target recognition using deep convolutional neural network. Firstly, we collect a large number of photos with hidden dangers for transmission line. Secondly, we classify those samples according to the detection targets. Thirdly, the deep convolutional neural network model is trained and tuned by using the sample images of various detection objects. When deployed on the server, identification of hidden dangers such as cranes, tower cranes, excavators, and mountain fires reaches acceptable accuracy.