Research On On-line Inspection System Of Bearing Ring End Face Based On Machine Vision

As an important basic part of the machinery industry,the quality of bearing affects the performance,reliability and service life of equipment.In order to meet the need of the market,bearing manufacturers also pay more attention to the control of bearing quality,the detection of bearing inner and outer ring ends is the key link.However,the traditional manual visual inspection method is not only inefficient,but also can easily lead to false detection and missed detection,which cannot meet the need of enterprises.Therefore,many enterprises urgently need an automatic inspection device instead of manual completion of the bearing ring end face detection to reduce the influence of human factors,and achieve strict control of product quality.In this thesis,an on-line inspection system,based on the actual need of the cooperative enterprise,has been designed to realize the automatic detection of defects on the ring end face and the rejection of defective products by machine vision technology.The detection system is composed of two parts,hardware and software.The hardware part mainly includes the construction of visual system and the design of the supporting hardware.The software part includes the design and implementation of functional modules such as the system authority management and parameter setting and the defect detection algorithms.By analyzing the characteristics of all kinds of defects on the bearing ring end face,the defects are divided into three types,the shape defect,the defect below the target gray level(called as low-grayscale defect)and the defect above the gray level of the target(called as high-grayscale defect).Based on this defect type and detection order,the defect detection algorithm is divided into shape defect detection algorithm,low-grayscale defect detection algorithm and high-grayscale defect detection algorithm.In the shape defect detection algorithm,a fast adaptive threshold Canny algorithm is proposed for edge detection,and the number of edges is used to detect some defects of the shape.Then the inner circle and outer circle of the end face are fitted by the least square method,and the end face location and extraction are completed.Finally,the detection of the remaining shape defects is realized according to the fitting results and prior knowledge.In the low-grayscale defect detection algorithm,the Otsu method is used to obtain the best threshold of the annular region of the ring face,and the segmentation and extraction of the low-grayscale defect can be realized.In view of the fact that the uneven illumination may cause the error detection of the algorithm,the end face is divided into eight regions,and the minimum gray value of each sub-region and the defect area are introduced as a criterion to improve the detection accuracy of the algorithm.In the high-grayscale defect detection algorithm,the end face is divided into four regions in order to reduce the influence of uneven illumination and consider the efficiency of the algorithm.Based on the seed pixels of each region,the region growing method is used to extract the target area,and the defect discrimination is completed according to the shape characteristics of each region.The on-line inspection system of bearing ring end face based on machine vision has realized the automation of the appearance detection process,improves the detection efficiency and detection quality,and contributes to the improvement of the product quality of the bearing industry in China.The off-line experimental result shows that the system's comprehensive detection accuracy can attain to 99.1%,and the average detection time does not exceed 300ms.The proposed system can achieve 100%detection of the bearing ring and has a high engineering application value.