The 3D reconstruction technology based on laser point clouds plays a vital role in the digitization of power systems. However, the accuracy of reconstructed results is often compromised by sensor noise and other interferences affecting the collected point cloud data. To address this issue, an improved point cloud denoising method based on enhancing PointCleanNet is proposed. Firstly, the radius filter is introduced into the PointCleanNet algorithm to further enhance the denoising efficacy. Secondly, experiments and validation are conducted using point cloud data from the ShapeNet dataset. Finally, a comparative analysis is performed between denoised and non-denoised point cloud data. Additionally, the PF-Net algorithm is utilized for 3D reconstruction, indicating that the optimized PointCleanNet method enhances the quality and precision of 3D reconstruction. The combination of this study with the power system can improve the security, reliability and efficiency of the power system and promote the progress and development of power system technology.