Research On Bearing Cloud Detection Method Based On Machine Vision

The bearing industry plays an important role in the manufacturing industry,and the bearing is an indispensable basic part of mechanical equipment.The bearing rings are affected by the production process is prone to cause surface defects with the result that after processing and assembling the bearing,it affects the normal operation of mechanical equipment.The company usually adopts manual visual inspection of bearings,which has disadvantages such as low efficiency and uncontrollable factors.This paper studies bearing defect detection based on machine vision method.The main contents are as follows.First,the design of the bearing defect detection device was finished.By analyzing the bearing structure and production process,the types of defects that were prone to occur in the actual production process of bearing rings were summarized.Combined with the factory inspection requirements,a set of bearing defect inspection system was proposed and the structural design and selection design of its mechanical parts and core components were carried out respectively.By formulating the work flow of the detection device,the design goal of image acquisition was completed.Secondly,the bearing defect detection algorithm was designed and completed.Traditional defect detection algorithms had high requirements on the prior knowledge of target defects,while merely using the defect classification algorithm of deep learning,but it was not conducive to timely modification of the effective setting parameters for defining defects。Through image engineering,the traditional detection algorithm and convolutional neural network were combined to design and complete the detection algorithm suitable for bearing defects.Thirdly,the improved bearing defect classification model was built.Through image acquisition,data expansion and proportional division,the bearing defect data set was established.Using deep learning knowledge,improving convolution neural network,selecting optimizer,activating function and setting super parameters,experiments showed that the improved classification model had higher accuracy.Finally,the bearing detection test bench was built,and the "cloud" platform framework for bearing defect detection was designed.In order to realize the "cloud" sharing of bearing defect data sets and solve the problems of improving order execution,customer acquisition,enterprise credit,employee piecework system and financing demand of manufacturing small and micro enterprises,the "cloud" platform system framework was designed and relevant tests were carried out.Automatic detection of bearing defects based on machine vision can not only reduce the labor intensity of workers,but also improve the bearing production process,upgrade the bearing production line and improve the production efficiency of enterprises.