Research And Application For Surface Quality Inspection Method Of High-voltage Cable Joints Based On Three-dimensional Point Cloud

As limited by the current manufacturing process and transportation and other factors,the maximum manufacturing length of a single high-voltage cable is 10km.Therefore,when the long-distance electric energy transmission is to be realized,the cable joint needs to be polished by technicians on site to realize the connection between cables.Because the quality of the joints polished by various technicians is different,surface quality detection of the joints polished on site is an essential link before they are connected to the power system.Aiming at the problems such as strong subjectivity and poor accuracy in surface quality detection of laying cable joints by manual,this thesis proposes a surface quality detection method of high-voltage cable joints based on three-dimensional point cloud,which realizes that parameter measurement of cable joints and main insulation surface defect detection.Aiming at the parameter measurement of cable joint,this thesis proposes a parameter measurement algorithm of cable joint based on three-dimensional point cloud segmentation.Firstly,the algorithm adjusts the spatial pose of the joint point cloud based on data fitting.Then,the rough segmentation is realized according to the mutation characteristic of the ratio of radius variance of adjacent fitting circles at the regional junction,so as to obtain multiple local point clouds containing regional junction points.Then,strip division of each local point cloud is realized according to the angle relationship between points and coordinate axis in each local point cloud.On this basis,the intersection lines of adjacent regional planes are obtained by using the random sample consensus algorithm and Lagrange multiplier method,and the preliminary segmentation results are obtained according to the distance relationship between points on the strip point cloud and the intersection lines.Finally,based on the residual distance between the points of the strip point cloud and the fitting plane,the initial segmentation results are corrected to obtain the final segmentation results,and on this basis,the cable joint parameter measuredment is achieved.Aiming at the defect detection of main insulation surface of cable joint,this thesis proposes an algorithm based on point cloud dimensionality reduction mapping for detection of sag and protrusions on main insulation surface of cable joint.The algorithm first reduces the dimension of the main insulating point cloud based on coding mapping and data fitting,and obtains the two-dimensional mapping image.Then,according to the difference of pixel gray value between the defect area and the non-defect area in the mapping image,and the pixel eight neighborhood relationship of defect area in the mapping image,the extraction of each defect is completed.Nextly,the point information contained in each defect is obtained according to the coding relationship between the point cloud and the mapping image.Finally,the volume and area of each defect are quantified based on mathematical knowledge such as Helen’s formula and point cloud processing methods such as greedy projection triangulation algorithm.The experimental results of more than 50 artificially polished crosslinked polyethylene high voltage cable joints show that the absolute error of the proposed measurement method of12 parameters such as axial height and outer diameter is less than 1.5mm,and the relative error is less than 6.7%.The defect recognition rate of the proposed defect detection method is 93.5%.The absolute error of volume is less than 3.1mm~3,and the relative error is less than 4.3%.The absolute error of area is less than 8.5mm~2,and the relative error is less than 2.6%.Indicating that the proposed method has high robustness and measurement accuracy for the surface quality detection of the high-voltage cable joints.