3D Reconstruction Of Power Transmission Pylons For UAV Fine Inspection Based On LiDAR Point Clouds

As an indispensable foundation of modern social activities,the stable operation of power system is the guarantee of social and economic operation and people’s life.The loss caused by the damage of transmission equipment is immeasurable,so it is indispensable to carry out regular monitoring and maintenance of transmission corridors.Pylons,as the key facilities in the transmission corridors,are the main objects in the fine inspection and the accurate 3D reconstruction of pylons plays a vital role in the safe operation and maintenance of power systems,such as the automatic generation of fine inspection routes.Meanwhile,the 3D reconstruction of pylon is fundamental for the construction of smart grid.And it is an important part of the digital twin of the power system.As a new type of active sensor,LiDAR can quickly and accurately collect massive 3D point cloud data.LiDAR is widely used in transmission corridor inspection and it provides the foundational data for the construction of smart grid.However,due to the limitations of acquisition equipment performance and environment,the actual original point cloud usually has problems such as noise interference,data loss and density anisotropy.At the same time,pylons have various types and the complex steel structures.These bring great challenges to the three-dimensional reconstruction of the pylons.This paper presents an automatic and efficient reconstruction method for pylons from airborne LiDAR data,which combines both data-driven and model-driven strategies.Based on the structural characteristics of pylons,this paper decomposes the components of pylons and uses different strategies to achieve independent reconstruction and assembly for different structures so as to realize accurate 3D reconstruction of pylons.The main research work of the paper is as follow:(1)This dissertation introduces the domestic and international research progress of image-based and point cloud-based 3D reconstruction methods,and the current status of the application of the two methods in the 3D reconstruction of transmission pylons,respectively,and summarizes the shortcomings and problems of the existing pylon reconstruction algorithms.In view of these difficulties,the research objectives and contents of this paper are determined.(2)Structural decomposition for pylons.The structure of pylon head and pylon body varies greatly,this paper proposes a Transformer combined with geometric statistical features for pylon identification and pylon segmentation method.Firstly,an algorithm based on PCA is proposed to realize redirection and construct the local coordinate system.After that,Swin Transformer is proposed to identify the type of pylon.Meanwhile,the “projection ratio” and “fill rate” are designed to describe the structural characteristics of the pylons.Finally,based on the pylon type and structural features,the pylon can be decomposed into two parts: the pylon body and head.(3)Pylon body 3D reconstruction.The reconstruction of pylon body is based on prior structural knowledge and a data-driven strategy.According to the structural characteristics of pylons,the key points of external structures of pylon body are extracted,and the 2D clustering algorithm is used to extract the key point of internal structures.Then,the 2D linear fitting algorithm is used to fit steel frame structures for reconstructing the model of pylon body.(4)Pylon head 3D reconstruction.A hybrid-driven strategy is adopted to reconstruct the pylon head.With the aid of a pre-built pylon head parameter model library,the coherent point drift(CPD)algorithm is used to determine the topological relationship among key points of pylon head structures and the initial value of model parameters.Then,through the parameters are optimized by simulated annealing algorithm,the reconstruction of pylon head model can be achieved.(5)The proposed algorithm is tested and analyzed on the airborne LiDAR data of3,398 power pylons of 8 types.The average reconstruction time of pylon is 1.10 s,the average reconstruction accuracy of pylon head is 0.398 m,and the average fitting accuracy of tower body is 0.026 m.The experiment demonstrates that the proposed method has high accuracy and applicability.Based on the constructed accurate three-dimensional pylon model,the automatic generation of fine inspection routes for transmission lines is realized,which greatly improves the efficiency of transmission line inspection work and enhances the level of intelligence and automation of power grid operation and maintenance work.