| Gear is a commonly used key transmission component,widely used in transportation,aerospace,agricultural production and other fields.And it plays an indispensable role in the development and upgrading of the manufacturing industry.In recent years,with the rapid development and technological changes of China’s manufacturing industry,the production quantity of gear parts has continued to grow.The processed gears need to pass the quality inspection before they can be used in actual industrial scenarios.The gears put into use also need to be regularly inspected to ensure normal performance.The development of gear measurement and defect detection technology in China is relatively lagging behind,mainly relying on contact measurement equipment.This article is based on laser scanning technology in non-contact measurement to obtain three-dimensional point clouds of gears,and develops a gear point cloud processing system to achieve noncontact detection of gears.The main research content includes:(1)Collection and preprocessing of gear point cloud data.Select the scanning equipment based on the target features to complete the acquisition of the original point cloud,and then carry out preprocessing operations on the original point cloud,including:point cloud radius filtering,smoothing based on laplace operator,sampling strategy combining octree and uniform Downsampling,and estimation of point cloud normal vector.(2)Extraction of gear feature parameters.Based on the preprocessed gear point cloud data,the RANSAC algorithm is used to fit the feature point cloud plane.According to the characteristics of the gear feature parameters,the extraction methods of the tooth tip circle diameter,tooth number,tooth width and other parameters are designed,including:extracting the outer contour based on the normal and search radius,using statistical filtering algorithm to remove the noise of the outer contour point cloud,and extracting boundary point based on the convex hull algorithm,Fitting the boundary point to form a two-dimensional circle to obtain the location coordinates of the tooth tip circle parameters and the circle center;Calculate the centroid of the point cloud of the gear outer profile plane,sort the points in the outer profile point cloud counterclockwise around the centroid,calculate the distance from the sorted points to the centroid,and form a statistical curve.Determine the number of teeth based on the angle between the peaks of the statistical curve;Finally,the plane equation of the gear’s two end faces is extracted using the characteristics of the two end faces,and the distance between the end faces is calculated as the tooth width.(3)Extraction and classification of gear defect point clouds.Based on the extracted gear parameters,an ideal gear model is constructed,and an efficient point cloud registration algorithm based on a plane is proposed.The theoretical point cloud is registered with the actual point cloud,and the threshold deviation is set.The point cloud clustering method is used to extract the gear defect point cloud.Propose a classification and recognition method for gear defect point clouds: extract the length of radial defects,project the defect point cloud along the radial plane,segment the projected image along the edge to obtain a two-dimensional image,calculate the Hu invariant moment and density of the two-dimensional image,and classify the defect point cloud based on these parameters.Finally,perform gradient coloring based on the size of the defect deviation value to obtain a colored defect point cloud.(4)Development and validation of a gear point cloud processing system.Based on the above research results,the programming implementation of the gear point cloud processing system has been completed,and the system can display the gear point cloud and defects.Design validation experiments for gear parameter extraction and gear defect point cloud extraction and classification,respectively.The experimental results demonstrate that the system can achieve the preset gear point cloud processing function. |
