A Vehicle Bearing Ring Surface Defect Detection System Based On Machine Vision

Bearings are used in a very high proportion in the automotive industry,such as automotive transmissions,wheels,chassis differentials and other parts.Its quality directly determines the performance and stability of equipment and products,so the surface defect detection of bearing rings has become an important part.Traditional method of rings surface defects is manual visual inspection,which easily leads to problems such as low efficiency and easy to miss detection.The inspection technology based on modern machine vision is more and more used in defect inspection due to its simple process,fast inspection speed,high efficiency and strong adaptability.In this subject,a machine vision method was used to design an online inspection system for outer surface defects of bearing rings.The specific research work is as follows:(1)According to the inspection content and design goals,the overall scheme design of the inspection system was completed.Besides,the design of feeding mechanism,manipulator mechanism,turnover mechanism,rotating platform mechanism,and elimination system was completed and participated in the debugging.(2)Complete the hardware selection of the vision acquisition system for each station.The area scan camera is used to obtain images on the end faces of the ferrule,and the line scan camera is used to obtain images on the inner and outer sides of the ferrule.In the end-face image acquisition part,the method of using a spherical integral light source and a low-angle shadowless ring light source to cooperate is proposed,and the end-face image and the chamfered image are almost simultaneously obtained in one station.(3)The defects on the surface of the ferrule are classified,different algorithms are used for different types of defects,and the optimal plan is finally established through the comparison of multiple treatment plans.An adaptive threshold Canny algorithm is proposed for edge detection,which effectively extracts the edges of the ferrule.Aiming at the situation that the defect in slender ring area at the chamfer cannot be accurately segmented by the adaptive threshold value,the ring area is divided into lots of blocks,the gray average value of each block is calculated separately,and the low gray average area is found according to the judgment conditions,regard these regions as seed regions,and use the region growing algorithm to extract the target region.(4)Completed the design of the computer software.It includes functions such as interface design,result display,PLC communication,camera control,data management.Make the whole system run smoothly,and the detection results can be displayed in the screen immediately.Experimental test results show that the comprehensive detection accuracy rate of the system is 96.7%,the detection speed reaches every 4 seconds a piece,and the detection accuracy reaches 0.4mm.The detection system is at the advanced level of the industry and has high engineering application value.