Research On 3D Point Cloud Data Recognition Technology Of Substation Equipment Based On Laser Scanning

The 3D coordinate information of the object surface can be acquired quickly by the laser scanner, which can help to accelerate the reconstruction of the 3D model.With the application of laser scanner in the electric power industry, the establishment of the 3D simulation model of substation becomes the focus of research gradually. At present in the reconstruction of 3D model of substation, it is mainly through human eyes to identify the 3D point, determine the type of equipment, and then find out the corresponding 3Dmax model. This method not only wastes a lot of human resources,but also is inefficient. So it is of practical significance to study how to identify the cloud data of equipment fast and accurately. A 500 kV substation is selected as experimental scene. The 3D information of equipment is collected by laser scanner and the simplification algorithm of point cloud data is studied. It focuses on the solution of the problem of feature extraction and recognition of the point cloud data.At last the encapsulation of Matlab algorithm and the function of the model database management software are introduced.In view of the above research objectives, the specific work contents of this paper are as follows:(1) A 500 kV substation is selected and the point cloud data of all the equipment is acquired by Faro laser scanner. Then the data of the single device can be extracted using Geomagic software. The simplification method based on octree encoding is used and the number of point cloud data can be reduced greatly.(2) The curvature of the boundary points of the device in the three coordinate plane point set can express its shape information. In this paper, two methods of extracting boundary points are introduced, then analyzing their advantages and disadvantages. A rolling method is then proposed to extract outer boundary points of point set. This method accelerates the speed of extracting the boundary points. Finally,the curvature of the boundary point is acquired by point-to-chord distanceaccumulation and it is preserved as the feature in the model database.(3) In this paper, a projection contour features(PCF) extraction method is adopted. Comparing the feature of target cloud data and model cloud data firstly, the equipments whose shapes are similar are selected. So it can reduce the number of models to match. Then matching cloud data of target and model uses ICP algorithm,finding the model whose matching error is smallest. This model type is the target cloud data belongs.(4) The encapsulation of Matlab recognition algorithm is introduced firstly. After the completion of the encapsulation, dll file can be generated. Then the method for calling dll file by C# is provided. At last, the model database management software is introduced in detail, including the functions of add, delete, modify and query.