Research On Classification Method Of Insulators Based On Point Cloud

As the main transmission facility for power inspection in China,overhead transmission lines ensure that its safe and stable operation plays a vital role in China's power development.The insulator on the power pole is an important component in the transmission line and is also the main target in the power inspection process.In order to ensure the safe and reliable operation of the entire transmission line,it is necessary to timely and effectively classify and inspect the insulators on the power tower and eliminate the faults.Therefore,this paper studies the problem of point cloud information image processing collected by UAV equipped with Light Detction and Ranging(LiDAR)technology during power inspection,focusing on the preprocessing method of point cloud information of power pole insulator and the point-based method.Distributed primary direction Robust Principal Component Analysis(RPCA)point cloud segmentation algorithm,and in-depth study of the new application Bag of Words(BoW)dictionary model in this field.In this paper,the data needs to be processed for the object to be processed.Because the laser LiDAR technology is affected by the terrain complex factors in the acquisition process,the acquired point cloud information has measurement error,so the data is outliers and the voxel gate is used.The lattice method is used for denoising,which effectively reduces the time consumption of subsequent work.Secondly,an RPCA point cloud segmentation algorithm based on point distributed main direction calculation is proposed,which solves the problem of extremely high outlier sensitivity in traditional segmentation algorithm,and has significant improvement in computational efficiency and stability.Finally,for the case that the insulator features in the power tower are few and difficult to classify,a classification method based on the BoW model is proposed.A large number of feature points of 3d point cloud data were extracted through PointCNN deep learning network,a word bag dictionary with a large number of words was obtained by clustering with k-means method,and the purpose of classification was achieved by SVM classifier.Experiments show that the above method is feasible for the classification of insulator point cloud of power tower,and has good classification accuracy and efficiency.