Research On The Composite Sensing Measurement Method Of Bearing Cylindrical Roller Surface Defects

As a core component of contemporary machinery and equipment,bearings play an important role in supporting national economic development and the construction of basic industries.Roller is an important part of the bearing,its surface defects will lead to bearing operation failure,directly affecting the accuracy and service life of the bearing.At present,the surface quality of bearing rollers mainly relies on processing technology to ensure that the detection of defects rely on manual visual sampling and other traditional methods.The method is inefficient and cannot fully guarantee the accuracy of the inspection,nor can it meet the requirements of industrial inspection.Accordingly,based on machine vision inspection technology,this paper designs a composite sensing measurement method combining two-dimensional images and threedimensional data to achieve accurate identification of roller surface defects,with the following main research content.(1)Analyze the types of surface defects that are likely to occur during the actual production of rollers,select the core components such as sensors with integrated image acquisition module and grating projection module,motion control device and mechanical motion structure to build a hardware inspection platform,and design a software control system for the entire composite measurement method of bearing roller surface defects to achieve information acquisition,data processing and human-computer interaction.(2)A two-dimensional image processing algorithm was designed to enable the extraction and localization of defects.Firstly,the image is pre-processed by Gaussian filtering and the processed multi-bit pose image is stitched together,then the defects are extracted by using the threshold segmentation method combined with morphological operations,and finally the Blob analysis method is used as a means to obtain the characteristic parameters and centre point coordinates of all suspected defects to avoid the miss detection rate,completing the first step of the composite measurement method for roller surface defects.(3)Using 3D point cloud data processing techniques and using depth values as evaluation indicators,defects are accurately identified as genuine when located in the image.Firstly,the original roller point cloud data is denoised and streamlined,and then the method proposed in this thesis is used to detect the defect depth information.The principle is that after the 3D data is converted to 2D plane,the extracted plane boundary points are fitted by three times B-sample curves to quantitatively calculate the defect depth value,excluding the actual non-existent pseudo-defects caused by external factors and reducing the false detection rate.(4)Experiments to verify the feasibility and reliability of the composite measurement method.For the defect composite detection method features,using the depth value as an indicator to screen unqualified rollers.By using different diameter rollers for experimental verification,the accuracy of this thesis’s composite measurement method to identify whether the rollers are qualified can reach 99.23%,especially the accuracy of the identification of unqualified rollers reaches 100%,providing a theoretical and experimental basis for the research of cylindrical roller surface defect detection instruments.