Point Cloud Processing And Hole Identification Method Of Attached Chip Blades Based On Structural Light Measurement

Aeroengine blade is the core component of engine.With the application and requirements of thin-walled curved parts represented by aeroengine blade becoming higher and higher,it is of great significance to consider the intelligent inspection integration technology in machining.Because of its high precision and high efficiency,structured light detection technology is widely used for on machine detection of such thin-walled blades.Due to the huge amount of point cloud data in the machining environment,the measurement accuracy is inaccurate,which affects subsequent point cloud operations,and the point cloud data needs to be simplified;Due to the thermal viscosity of aero-engine blade materials during the machining process,there is a phenomenon that chips attach to the blade surface in the dry milling scene,leading to the decoding error of structured light,which forms holes and bulges in the point cloud,affecting the precision of structured light measuring the workpiece surface.Therefore,in the process of structured light detection,it is particularly important to study the point cloud reduction method,chip formed holes and bulges repair technology.Firstly,according to the modeling and structural characteristics of aeroengine blades,the requirements for on board inspection were studied,and an optical inspection experimental platform was built.The key components and detection principle of monocular structured light equipment were described in detail,the point cloud data obtained from structured light detection were classified,and the point cloud data preprocessing methods and characteristics were introduced.According to the on machine detection requirements of aeroengine blades and the principle of structured light detection,an optical detection experimental platform was built,and the motion module,detection module and control module in the optical detection platform were analyzed.Secondly,aiming at the problem that the point cloud data sampling of thinwalled blades will lead to feature disappearance,an improved index space method was proposed to achieve accurate point cloud reduction.First,the normal vector of the point cloud was calculated,and the outer contour of the blade was extracted according to the angle threshold of the normal vector and Euclidean distance.Secondly,the average curvature and Gaussian curvature of the point cloud were calculated,the threshold and divide the point cloud were seted into molecular sets,and the index space method was used to simplify the point cloud data of different subsets,so that the aeroengine blade point cloud features can be well preserved.Thirdly,in order to solve the problem of holes in the point cloud data obtained by structured light measurement due to the chip adhering to the surface during the processing of thin-walled blades,a hole recognition method based on the combination of eight neighborhood of grid point cloud and machine vision was proposed.First,according to the view point characteristics of the processing environment,the foreground area of the gray image was acquired,and the gray value of the extracted area was binarized.Then,the image morphology was used to detect the burrs and identify whether the chips were attached to the outer contour of the blade.Then,set the threshold value through the average distance of the point cloud.After the point cloud was locally projected,calculate whether the distance between the seed point cloud and its eight neighbors exceeds the threshold value,complete the hole point extraction,and judge whether the point cloud’s outer contour data was retained according to the image recognition feedback information.Finally,the point cloud data with holes and bulges were repaired to complete the on machine reconstruction measurement of the morphology.First,used RANSAC method to repair the convex point cloud.Secondly,the properties of cubic B-spline were analyzed,and the linear point cloud was extracted through the grid point cloud feature straight through filtering.Finally,cubic B-spline was used to fit linear point cloud,and threshold value was set to judge whether the distance between point cloud data was too large to supplement point cloud,completed hole repair,and achieved the integrity of point cloud data,which provided theoretical and application basis for on machine inspection of aeroengine blade parts.