Research on Nonlinear Surveying and 3D Modeling Methods for Underground Distribution Pipelines in Large Cities

With the advancement of urbanization, accurate modeling and reconstruction of underground distribution pipelines have become increasingly important in power system planning. Traditional linear modeling methods are insufficient to meet the demands of complex underground environments. Therefore, this study proposes a data collection scheme that combines LiDAR (Light Detection and Ranging) and UAV (Unmanned Aerial Vehicle) imagery, along with a nonlinear modeling method that integrates Convolutional Neural Networks (CNN) and Long Short-Term Memory Networks (LSTM). This approach can accurately capture both the static and dynamic characteristics of underground distribution pipelines, enhancing modeling accuracy and adaptability. Experimental results demonstrate that the proposed model outperforms traditional methods, especially in environments with pipeline interferences and complex terrains, showing higher accuracy and stability. This method provides effective support for the 3D modeling, optimized design, and efficiency improvement of power transmission in underground pipelines.