On-line Inspection Key Technology Research For The Train Wheelset Geometric Parameters

According to production problems from train wheelset manufacture enterprises,and combined with key scientific and technological development projects in Jilin Province Science and Technology Development Project named "On-line Inspection Key Technology Research for Train Wheelset Production Quality"(project No.20160204005GX),wheelset production on-line detection technology is researched including wheel profile arc radius,wheel rolling circle diameter,wheel radial runout,other geometric parameters and engraved characters in the shaft end surface.According to design requirements of train wheelset on-line inspection,based on machine vision and laser detection technology,the on-line inspection system is designed and developed to detect train wheelset geometric parameters and engraved characters in shaft end;and two kinds of orientation support mechanism are designed taken wheelset center hole and bearing surface as datum respectively,meanwhile universal joint rotary drive mechanism,high precision horizontal / vertical movement mechanism,gantry base support mechanism and other mechanical structures are also designed;The structured light sensor with the AT cameras and line laser,data acquisition card,servo motor,Servo amplifier,motion control card and other data acquisition and electrical control components are chosen;the developed train wheelset production on-line inspection project prototype are tested at the production site;The results including wheel profile arc curve radius,wheel rolling circle diameter,wheel radial runout and other geometric parameters and engraved characters in the shaft end,can meet the design requirements.In the visual detection camera model,the transformation relationship between four coordinate systems,camera coordinate system,world coordinate system,image pixel coordinate system and image physical coordinate system are analyzed and the mathematical model for line-structured light sensor inspection system is researched.Regarding the practical problems that the inspection blind area blocked by the wheel rim,the technical scheme with two line-structured light sensors is applied and the calibration of two cameras is completed by calibration tools in Halcon software.The method for detecting the arc radius of each section on the wheel curve is studied.Regarding BC and EF segment arcs are short and with few measurement data,the error of the radius would be larger if calculated by the least squares method.The arc radius calculation method is proposed that the taking the arc endpoint and the theoretical circle center as constraint conditions,which improves the accuracy for circular radius measurement effectively.The algorithm that searching multi-circle edge by applying artificial bee colony algorithm is proposed.Taking advantage of artificial bee colony algorithm,the searching efficiency of the algorithm is improved.For the curve composed of multi-segment arcs,artificial bee colony algorithm is applied to optimize the detection data of the wheelset profile.The results show that the proposed algorithm can effectively improve the wheelset rim curve radius detection accuracy.A method is presented to register the 3D point cloud data of the wheelset profile in the frequency domain,through compiled MATLAB programs to complete the accurate registration for the 3D point cloud of the wheelset rim profile,acquired by the two line-structured light sensors,and engineering problems are solved that single line-structured light sensor can only capture some parts of the wheelset surface data because the camera is blocked by wheelset rim block.For the problem of blurring of contours caused by different depth and illumination angle,two kinds of character recognition methods are proposed.One method is based on the principle of the gray gradient mutation on the character image edge and the niche artificial bee colony algorithm is used to simulate the population behavior of the colony to find the nectar,the edge of the character image is acquired by colony iterative search method.According to the characteristics of the character image,in the artificial bee colony algorithm,the niche technology is introduced to control the bee group convergence to each corresponding character by setting the niche radius guidance,and get the balance of the global optimal solution and the precision of the solution.The other method is for engraved characters in the wheelset shaft end surface and without color difference background but with the depth of information characteristics,character point cloud information is acquired by line-structured light sensor,combined with total least squares plane fitting method,outline of the character information is extracted from the character point cloud depth data.The validity of the two methods is verified by the experiment results.Experiments show that the researched and the developed wheelset production quality on-line inspection system holds the higher detection accuracy,and can meet the wheelset production on-line inspection speed requirements.The results of this paper have theoretical and practical value for intelligent detection technology applied in the train wheelset on-line manufacturing quality.