Research On Workpiece Surface Damage Recognition Based On 3D Laser Scanning

When the equipment is in operation, friction and wear phenomena may appear on the workpiece surface, causing damage/defect to the workpiece surface, which can have important effect on the health of workpiece surface and the performance of equipment. If the degree of damage/defect is not in the normal range, mechanical failure and safety accidents will happen, resulting in unnecessary personal harm and economic loss. So it is very necessary to achieve high accuracy recognition of the workpiece surface damage/defect size, position and so on. In this paper, the high-speed sliding electrical contact rail is the research workpiece. According to the characteristics of the rail surface damage, the non-contact measurement method based on laser triangulation measurement principle was mainly studied, then a 3D measurement information system experimental platform was designed by the high resolution 2D laser scanner for precision identification of rail surface damage information.The 3D information measurement system is used to measure the information of the damage morphology information of different material guide rails at different positions,and the measured 3D point cloud data is the core processing data of damage identification.Based on the study of 3D point cloud data processing technology, the feature extraction method of point cloud in damage area is studied to realize that the damage region can be accurately detected and located. According to the detection of damage information, the damage volume loss and quality is accurately calculated, furthermore recognition of the damage types can be realized.Firstly, a method of workpiece surface damage detection is proposed which is based on point cloud mean curvature estimation.Then the mean curvature of point clouds which represents the "external bending" geometric feature is estimated and the OTSU method is used to set the threshold curvature. Furthermore, the accurate extraction of damage point cloud and precise location of damage region can be realized by the said method. Secondly,in order to verify the detection method is accurate and feasible, a method based on point cloud depth mapping color is put forward. The point cloud depth color mapping model is constructed, then the point cloud depth of damage morphology is mapped into the red,green and blue(RGB) information, and a one-dimensional maximum entropy method is used to set the optimal color threshold, realizing the accurate extraction of the damaged area. Then, according to the characteristics of the damage extraction and point cloud data,the volume of damage and the quality of the damage is estimated. Finally, according to the characteristics of damage types, as well as in order to simplify the classification, binary tree pattern recognition method is used to establish damage classification model. Besides,by extracting the damage characteristic parameter and setting up the corresponding classification rules, the recognition and classification of micro damage on the guide rail surface is realized at last.Experimental results show that the two methods are effective and feasible, and as well as show that the minor damage of which mass loss is less than 1 gram detection rate can reach more than 98%; mass loss detection accuracy is up to Milligram; pits and scratches damage identification rate can reach more than 85%.