Research On Bend Size Detection And Flaw Detection Path Planning Based On Point Cloud

Large diameter hot induction bend plays a key role in oil and gas pipeline transportation,such as connecting,changing direction and diverting.The dimension detection and nondestructive testing of bend are indispensable in its production process.Traditional hot-simmering bend size detection mainly relies on special detection tools and manual marking,which is cumbersome and has low accuracy.Therefore,the main content of this thesis is to improve the size detection accuracy of large-diameter hot-simmering bend and improve the efficiency and accuracy of nondestructive detection..The main research work of this thesis is as follows:Firstly,the establishment of automatic measurement and automatic nondestructive testing platform for large diameter hot induction bend and the pretreatment of initial bend point cloud are completed.According to the characteristics of large pipe diameter of457-1422 mm and the maximum pipe length of 10 m,the appropriate robot and scanner for bend data acquisition are selected.Direct filtering,statistical filtering and Euclidean clustering were used to extract the bend point cloud from the scanning data of the workbench with a lot of noise.Secondly,a method is proposed to complete the missing point cloud of bend into a complete point cloud of bend by using projection bend contour.The curve point cloud data was projected on the workbench plane to extract the boundary line,and the contour line close to the 3d laser scanner was extracted by calculating the four vertices of the boundary line by convex hull algorithm.The curve point cloud was sliced along the contour line and ellipse fitting was carried out to get ellipse parameters.According to the ellipse parameters of each section,the complete bend point cloud data is reconstructed.RANSAC algorithm was used to extract the straight line segment of hot-simmering bend from the projected contour line and measure the length of the straight line segment.The diameter,bending Angle and ellipticity of the bend are calculated on the basis of the projected contour line and the length of the straight line segment,so as to realize the automatic dimension measurement of the large diameter hot induction bend.Finally,the path planning and attitude calculation of automatic bend detection are completed with part of bend point cloud data.After extracting the flaw detection area from the point cloud of the curved surface of the bend,it is divided into equal arc length and path feature point search.The normal vector estimated by the rectangular neighborhood point cloud is used to replace the k-nearest neighbor normal vector as flaw detection pose.The results of path planning and attitude are quantitatively analyzed from two aspects: the coverage of flaw detection area and the distance between probe and bend surface.Three kinds of bend tubes with diameter D of 1016 mm,610mm and 457 mm,bending Angle of 30-45 ° and bending radius of 5D were selected as verification objects for dimension measurement,flaw detection path planning and attitude calculation.The error of the bend point cloud is within the acceptable range,and the dimensional measurement accuracy meets the application requirements.The waypoint scanning area realizes the full coverage of the predetermined area,and the distance between any point on the exploration frame and the bend surface at each waypoint is within the acceptable range.The experimental results verify that the proposed method meets the requirements of precision,effectiveness and efficiency in the application of automatic dimension detection and nondestructive flaw detection.