The Study Of Substation Equipment Affiliated Facilities Segmentation Algorithms Based On 3D Point Cloud

Segmenting affiliated facilities from the point cloud is the key of 3D identification and reconstruction of the equipment.The effect of segmentation will have a great impact on identification and reconstruction.The computer is used to achieve the segmentation of equipment ancillary facilities,and then the identification algorithm is used to identify the equipment type,and then determine the location of the equipment.Finally,the distribution of the entire substation equipment is restored to achieve the reconstruction of substation equipment.In order to improve the reconfiguration efficiency of substation equipment and the recognition rate of equipment,a cone search method is proposed to segment the wire from substation equipment based on three-dimensional cloud.At the same time,a method is proposed to segment equipment's base from the equipment according to the change trend of the horizontally-projected areas of the layers formed by layering the equipment.As far as the main technical contributions of this paper are concerned,they can be summarized as follows:(1)Wire segmentation in substation equipment.In this paper,a cone search method is proposed to segment the wire from substation equipment.Specifically,the implementation process of the algorithm is expressed as follows.First,change the direction of the device by using rotating.Two possible wire endpoints,which can be confirmed by the characteristic of the wire endpoints in subsequent step,can be found in this step.Second,using K-means clustering algorithm and the nearest neighborhood search algorithm can find the position of other wire endpoints.Third,judge whether the possible wire endpoints are real wire endpoints according to the characteristic of the wire endpoints.Finally,use cone search algorithm to confine the search direction,and find out the data of wire and remove the wire points.(2)Segment wire from substation equipment.In this paper,a method is proposed to segment an equipment's base from the equipment according to the change trend of the horizontally-projected areas of the layers formed by layering the equipment.First,the test equipment point cloud data is preprocessed to obtain appropriate experimental data.Second,remove the test equipment base according to the location of separated line for the first time.Third,several preselected equipment can be obtained by using the projected features of the test equipment with the rest base,then each matching degree between each preselected equipment and the test equipment with the rest base using the improved ICP iteration algorithm is calculated.The preselected equipment with the highest matching degree is regarded as the current identified equipment.The above operation is marked as one loop and the final identified result is the one with the highest matching degree within 10 times loop.At the same time,the position of the separated line corresponding to the maximum matching degree is the exact position of the segmentation line.Remove the parts below the separated line to realize the partition of equipment base.(3)Improved ICP iteration algorithm.The biggest drawback of ICP is that it needs too many iterative times for identifying point cloud with a large amount of data,which will increase matching time.The specific method is that if any difference between continuous 10 errors is less than 0.01,the last several steps of iteration are not operated,and the ICP iteration is jumped out earlier to shorten the matching time.(4)Solve the defective wire point.In the process of searching wire point cloud data,when the point cloud is defective,the solution is to use a liner equation of the 3D space which is formed by clustering center points searched previously,to get a fitted line segment.Then extend the fitted line segment forward by a fixed step length to get the next clustering center point.In this way,the problem of defective point cloud of wire can be solved.As the result of testing 303 unidentified data provided by Henan Tenglong Information Engineering Company Limited,the base segmentation algorithm in this paper can improve the recognition rate from 56.77% to 85.8%.After solving the problem of point cloud defective,the line segmentation algorithm can segment 275 equipment wires completely.The improved ICP algorithm for 137 substation equipment can save about 96 minutes under the same conditions.Experiments show that the proposed segmentation algorithm can reduce manual removal of the base and wire operations,while improving the recognition rate of substation equipment greatly.