Research On 3D Modeling And Feature Extraction Of Weld Groove Based On Laser Scanning

Aiming at the three-dimensional complex spatial curved surface structure of the intersecting groove of offshore wind turbine jacket,a weld groove scanning system based on laser scanning is built,and the feature extraction and three-dimensional reconstruction methods of weld image are studied to realize the three-dimensional modeling and feature extraction of weld groove,which provides a theoretical basis for all-position automatic welding technology of large-diameter jacket intersecting groove.In order to realize the feature extraction and three-dimensional modeling of large weld groove,firstly,a platform including robot and line laser vision sensor is built.The mathematical model is established and the parameters of the image sensor are calibrated.The robot hand-eye calibration and the overall calibration of the measurement system are completed,and the hardware basis for obtaining the surface data of the weld groove is constructed.Secondly,the laser stripe image of the weld is obtained by the structured light system,and the image is thresholded.The Steger method based on the Hessian matrix is used to extract the sub-pixel position of the light strip.Combined with the clustering method,the feature points extracted by the Steger algorithm are optimized to improve the accuracy of the light strip center extraction and make the recognition result more stable.A multi-region multi-segment scanning strategy was developed,and a mathematical model was used to convert the two-dimensional stripe centerline into a three-dimensional point cloud to obtain point cloud data of multiple sets of jacket grooves.Finally,in order to obtain a complete jacket groove point cloud model,a point cloud registration method based on multiple local feature matching is studied.The SIFT-3D key points are used to establish the multi-neighborhood local feature descriptors of point clouds,and the transformation matrix is solved by feature matching and TrICP algorithm.The multi-group groove point clouds are transformed into a unified base coordinate system to obtain the complete jacket groove point cloud data.The errors in the X,Y,and Z directions are all less than 0.5 mm,and the average error of the overall weld groove is 0.416 mm,with high accuracy.According to the point cloud normal vector,the groove features are segmented,and the welding path points of the jacket weld are generated,and the weld depth is measured.The average error is 0.421 mm,and the three-dimensional modeling and feature extraction of the weld groove are realized.In this paper,the laser scanning system of jacket intersecting groove weld is constructed,and the accurate extraction of laser stripe line of weld image is realized.The three-dimensional model of complex spatial curved surface structure of jacket intersecting groove is constructed,and the feature model extraction of jacket weld is realized.It is of positive significance to the automatic welding technology of offshore wind power jacket.The research results provide a theoretical basis for robot automatic welding technology and three-dimensional modeling of weld trajectory,and provide a new method for the welding of jacket intersection in large steel structural parts such as offshore wind power,pressure vessel and bridge.