In view of the problem that traditional methods are difficult to accurately and quickly monitor and identify the status of transmission lines, this article proposes a method for constructing an intelligent identification system for transmission lines. This article introduces research results in related fields and explains the application prospects of deep learning algorithms in transmission line fault detection. Then, this article introduces in detail how to use DJI Mavic 2 Pro (DJI Mavic 2 Pro with Hasselblad Camera) drone to collect images, and use OpenCV (Open Source Computer Vision Library) to perform image enhancement processing and add noise and other data enhancement methods to expand the quantity and quality of training samples. This article designs and uses a convolutional neural network (CNN) to implement feature extraction and recognition of transmission line images. Finally, this article proves through experimental tests that the CNN recognition model has high recognition accuracy and training effect, and the recognition accuracy of accessory faults in transmission lines reaches 98.56%. This article aims to provide a new deep learning algorithm application idea for transmission line fault detection, and provide reference for academic research in related fields.