Automatically segmenting the main body of transmission towers from three-dimensional point cloud data is a crucial step in the automation modeling of power transmission lines, bearing significant importance. However, existing point cloud segmentation methods suffer from issues such as high costs and low accuracy. Designing an automated point cloud segmentation method based on deep learning holds paramount significance. To address this issue, this paper integrates the Dynamic Graph Convolutional Neural Network (DGCNN) model to achieve automated semantic segmentation of laser scanning point cloud data from real transmission lines. Experimental validation demonstrates that the proposed method achieves higher accuracy on the test dataset compared to existing methods.