Identification method of insulators in complex overhead transmission line scenarios based on three-dimensional point cloud features

In the project of automatic inspection of overhead transmission lines by unmanned airborne vehicles (UAVs), the inspection and acceptance of insulator strings is an important part. Using point cloud data captured by UAV light detection, and ranging (LiDAR) system to identify insulator position information plays a key role in power line stability and improving inspection efficiency. Aiming at the effective extraction of small power component insulators in point clouds of complex overhead transmission line scenarios, this paper proposes an insulator identification method based on three-dimensional point cloud features. First, the point cloud downsampling and statistical filtering algorithms are used to preprocess the original point cloud data, and the DBSCAN clustering algorithm is used to roughly extract the tower point cloud combined with the elevation difference information. Use multiple eigenvalue features of the local point cloud to construct an insulator point cloud feature set, standardize each feature through the entropy weight method and determine the weight of each feature, thereby obtaining the feature matrix of the insulator point cloud and using it to train a random forest classifier. Complete point cloud identification of insulators by adaptively adjusting the parameters of the classifier. Experimental results show that this method is suitable for rapid and effective extraction of insulators from complex point cloud scenes collected by UAVs, and provides important technical and data support for location selection and fault detection of automatic insulator inspections.