Micro bearing rapid batch online detection method and system

Abstract

The application discloses a rapid batch online detection method and system for micro bearings, and relates to the technical field of bearings.The method comprises the following steps: cooperatively collecting micro bearings to generate a multidimensional detection data set; performing dynamic space registration on key detection areas to locate the key detection areas of the micro bearings; performing defect feature extraction on the key detection areas to generate a defect probability distribution graph, performing dynamic segmentation on the defect probability distribution graph, and generating a defect type label group; performing online sorting on the micro bearings based on the defect type label group to obtain a sorting instruction set, generating a detection quality report, and feeding back the detection quality report to a production control system.The application solves the technical problems of low detection efficiency, insufficient sampling accuracy, and difficulty in real-time online detection and quality control of the micro bearings in the prior art, realizes rapid batch online detection of the micro bearings, and improves the detection efficiency and accuracy of the micro bearings.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of bearing, in particular to a rapid batch online detection method and system for micro bearings. BACKGROUND

[0002] Micro bearings are widely used in aerospace, medical devices, electronic equipment and precision instruments and many other high-end manufacturing fields. The performance and reliability of the equipment put forward very high requirements on the precision, quality and stability of the micro bearings. Traditional micro bearing detection methods mostly rely on manual sampling inspection or offline laboratory detection equipment. However, manual sampling inspection is not only low in efficiency and difficult to meet the rhythm of large-scale production, but also the detection results are easily affected by subjective factors of the detection personnel, and there is a large probability of misjudgment and omission. Although offline laboratory detection can provide more accurate detection results, the detection process takes a long time and cannot realize real-time monitoring and feedback of products on the production line, which is difficult to adapt to the needs of modern production for efficient and rapid quality detection, thereby affecting the detection efficiency and accuracy in the production process of micro bearings.

[0003] Therefore, in the related art at present, there are technical problems of low detection efficiency of micro bearings, insufficient sampling inspection accuracy, and difficulty in real-time online detection to control bearing quality. SUMMARY

[0004] The present application provides a rapid batch online detection method and system for micro bearings, which solves the technical problems of low detection efficiency of micro bearings, insufficient sampling inspection accuracy, and difficulty in real-time online detection to control bearing quality in the prior art, realizes rapid batch online detection of micro bearings, and achieves the technical effect of improving the detection efficiency and accuracy of micro bearings.

[0005] The present application provides a rapid batch online detection method for micro bearings, which comprises: cooperatively collecting micro bearings to generate a multi-dimensional detection data set; dynamically spatially registering the multi-dimensional detection data set to locate the key detection area of the micro bearing; traversing the key detection area to extract defect features and generate a defect probability distribution map; dynamically segmenting the defect probability distribution map to generate a defect type label group; based on the defect type label group, online sorting the micro bearing to obtain a sorting instruction set and generate a detection quality report feedback to a production control system.

[0006] The application also provides a rapid batch online detection system for micro bearings, comprising: a cooperative acquisition module for cooperatively acquiring micro bearings to generate a multi-dimensional detection data set; a dynamic spatial registration module for dynamically registering the multi-dimensional detection data set to locate key detection areas of the micro bearings; a dynamic segmentation module for traversing the key detection areas to extract defect features and generate a defect probability distribution map, and dynamically segmenting the defect probability distribution map to generate a defect type label group; and an online sorting module for sorting the micro bearings based on the defect type label group to obtain a sorting instruction set and generate a detection quality report to be fed back to a production control system.

[0007] The rapid batch online detection method and system for micro bearings provided by the application cooperatively acquire micro bearings to generate a multi-dimensional detection data set, traverse key detection areas to dynamically register the key detection areas and locate key detection areas of the micro bearings, traverse the key detection areas to extract defect features and generate a defect probability distribution map, dynamically segment the defect probability distribution map to generate a defect type label group, sort the micro bearings based on the defect type label group to obtain a sorting instruction set, and generate a detection quality report to be fed back to a production control system. The technical problems of low detection efficiency, insufficient sampling accuracy and difficulty in real-time online detection to control bearing quality in the prior art are solved, rapid batch online detection of micro bearings is achieved, and the technical effects of improving the detection efficiency and accuracy of micro bearings are achieved. BRIEF DESCRIPTION OF DRAWINGS

[0008] In order to more clearly illustrate the technical solutions of the embodiments of the present disclosure, the drawings of the embodiments of the present disclosure will be briefly introduced below. The flowcharts are used to illustrate the operations performed by the system according to the embodiments of the present disclosure. It should be understood that the foregoing or the following operations are not necessarily performed in sequence. On the contrary, various steps can be processed in reverse order or simultaneously according to needs. Meanwhile, other operations can be added to these processes, or one or more steps can be removed from these processes.

[0009] Figure 1 The rapid batch online detection method for micro bearings provided by the embodiments of the present application is shown in the flowchart.

[0010] Figure 2 The rapid batch online detection system for micro bearings provided by the embodiments of the present application is shown in the structural diagram.

[0011] Legend: cooperative acquisition module 10, dynamic spatial registration module 20, dynamic segmentation module 30, online sorting module 40. DETAILED DESCRIPTION

[0012] The above description is only a summary of the technical solutions of the present application. In order to make the technical means of the present application more clearly understood, and to be implemented according to the content of the description, and in order to make the above and other purposes, characteristics and advantages of the present application more obvious and easy to understand, the following specific embodiments of the present application are described.

[0013] In order to make the purposes, technical solutions and advantages of the present application more clear, the present application will be further described in detail below with reference to the accompanying drawings. The described embodiments should not be regarded as a limitation of the present application. All other embodiments obtained by those skilled in the art without creative labor fall within the scope of protection of the present application.

[0014] In the following description, "some embodiments" are related to a subset of all possible embodiments, but it can be understood that "some embodiments" can be the same subset or different subsets of all possible embodiments, and can be combined with each other without conflict. The term "first\second" is only to distinguish similar objects, and does not represent the specific order of the objects. The terms "include" and "have" and any variations, are intended to cover non-exclusive inclusion, for example, a process, method, system, product or server including a series of steps or units does not have to be limited to those steps or units clearly listed, but can include other steps or modules that are not clearly listed or inherent to these processes, methods, products or devices. Unless otherwise defined, all technical and scientific terms used herein have the same meaning as understood by those skilled in the art to which the present application belongs. The terms used herein are only for the purpose of describing the embodiments of the present application.

[0015] The embodiments of the present application provide a rapid batch online detection method for micro bearings, as shown in Figure 1 The method comprises the following steps:

[0016] In step S100, the micro bearing is cooperatively collected to generate a multi-dimensional detection data set.

[0017] Preferably, the micro bearing is cooperatively collected by the high-speed linear array camera and the annular multi-spectral light source, that is, the surface and internal structure images of the micro bearing on the assembly line are collected respectively to generate a multi-dimensional detection data set. Specifically, the high-speed linear array camera can quickly acquire images. When detecting the micro bearing on the assembly line, the high-speed linear array camera can quickly scan and image the surface of the micro bearing in a dynamic motion state, can capture images at a very high frame rate, and can ensure that various features of the surface of the micro bearing, such as the texture of the surface, whether there are scratches, wear, cracks and other defect information, are recorded completely and clearly during the movement of the micro bearing. The annular multi-spectral light source can emit light rays of different wavelengths, has different penetration abilities, and can penetrate the surface of the micro bearing to different depths inside the micro bearing when detecting the micro bearing, and can reflect, refract or scatter when encountering internal structure features (such as internal geometry, material inhomogeneity, internal defects, etc.). The annular multi-spectral light source surrounds the micro bearing and can irradiate it from various angles, so that the high-speed linear array camera can collect images containing internal structure information of the micro bearing, reflect the internal situation of the micro bearing, and help detect potential internal defects. The collected surface images and internal structure images provide information of the micro bearing from different dimensions to form a multi-dimensional detection data set, so that the quality of the micro bearing can be evaluated from multiple angles.

[0018] Further, step S100 further includes step S110 of arranging an adjustable focus annular light source array on both sides of the bearing transmission track and setting an initial spectral combination; step S120 of acquiring a bearing rotating speed in real time; step S130 of triggering the high-speed linear array camera to collect the micro bearing according to a bearing rotating period in combination with the bearing rotating speed when the micro bearing enters an imaging area to obtain an axial image data set of the micro bearing; step S140 of cooperatively collecting internal structures of the micro bearing based on the initial spectral combination to obtain an internal structure image data set; and step S150 of performing space-time alignment on the axial image data set and the internal structure image data set to obtain the multi-dimensional detection data set.

[0019] Preferably, the bearing transmission track is arranged on both sides of the array of annular light sources with adjustable focal length, which can provide uniform illumination, and the initial spectral combination is set to be visible light and near-infrared light alternately irradiated, the visible light can clearly show the color, texture and other features of the bearing surface, and the near-infrared light is used to detect the structural features of a certain depth under the bearing surface; and the rotation speed of the bearing is monitored in real time by using a rotary encoder, wherein the rotary encoder can convert the angle and speed of the rotating object into an electrical signal, specifically, when the micro bearing enters the pre-set imaging area, according to the real-time bearing speed, the high-speed linear array camera collects 360° image of the bearing according to the rotation period of the bearing, ensures that the image information of the entire circumferential surface of the micro bearing is collected, so as to obtain the axial image data set containing the surface features of each angle of the bearing in the axial direction, which is used to detect the defects, wear and other conditions of the bearing surface.

[0020] Preferably, according to the initial spectral combination of visible light and near-infrared light alternately irradiated, the internal structure of the micro bearing is cooperatively collected by X-ray transmission, and the internal structure image data set containing detailed information of the internal structure of the micro bearing is obtained, which is further used to detect whether there are cracks, pores, material unevenness and other defects; then the axial image data set (reflecting the bearing surface information) and the internal structure image data set (reflecting the bearing internal information) are spatio-temporally aligned, that is, the two sets of data correspond to each other in time and space, ensuring the consistency and accuracy of the data; and then the multi-dimensional detection data set is integrated, and the accurate detection of the micro bearing is ensured.

[0021] Step S200, dynamically space registering the multi-dimensional detection data set to locate the key detection area of the micro bearing.

[0022] Preferably, the multi-dimensional detection data set is dynamically spatially registered, i.e. the micro bearing image data collected from different sources and different perspectives are spatially aligned to eliminate differences in different perspectives and coordinate systems. For example, based on feature point registration, feature points are extracted in different images, which have unique positions and feature descriptions (such as the edges of the inner and outer rings, the contours of the balls, etc.), and then by matching the feature points, the transformation relationship between the images is calculated, such as translation, rotation, scaling, etc., so as to accurately register the detection images. Different parts of the micro bearing have different effects on performance and quality. For example, the raceway of the bearing, the contact area between the ball and the raceway, and the sealing ring mounting part are prone to defects such as wear and tear and cracks, and then by analyzing the registered multi-dimensional detection data set, the key detection area is determined, such as establishing a three-dimensional model of the micro bearing, matching the registered image data with the three-dimensional model, and finding the corresponding area in the image data according to the set key area; or based on machine learning (such as object detection), a recognition model is constructed, a large number of micro bearing image data with annotations are trained, the features of the key detection area of the bearing are learned, so as to automatically recognize and locate the key detection area of the micro bearing, while ensuring the accuracy and efficiency of batch detection of the micro bearing.

[0023] Further, step S200 further comprises step S210 of reconstructing a point cloud based on the multi-dimensional detection data set to generate bearing three-dimensional point cloud parameters; step S220 of setting a set of geometric constraint conditions of the micro bearing, dynamically spatially registering the bearing three-dimensional point cloud parameters according to the set of geometric constraint conditions, and constructing a spatial transformation matrix; step S230 of performing registration residual analysis according to the spatial transformation matrix to identify a plurality of local deformation difference areas, the plurality of local deformation difference areas containing a plurality of deformation gradient amplitudes, and the plurality of deformation gradient amplitudes having a corresponding relationship with the plurality of local deformation difference areas; and step S240 of performing weight distribution on the plurality of local deformation difference areas according to the plurality of deformation gradient amplitudes, constructing a regional weight distribution map, and performing detection positioning according to the regional weight distribution map to determine the key detection area.

[0024] Preferably, the multi-dimensional detection data set is reconstructed into a point cloud, specifically, the feature points of the axial image of the micro bearing surface are projected into a three-dimensional space, the internal structure image is projected and converted according to the penetration characteristics of X-rays and the imaging geometric relationship, preliminary point cloud data is obtained, and then the point cloud data of the axial image and the internal structure image is fused to generate complete micro bearing three-dimensional point cloud data. Basic parameters of the point cloud are calculated, including calculating the average value of the coordinates of all points to obtain the centroid coordinates of the point cloud, calculating the minimum value and maximum value of the boundary to obtain the boundary box parameters of the point cloud; the geometric feature parameters of the micro bearing are calculated according to the point cloud data, such as determining the radius of the inner and outer rings of the bearing, the diameter of the ball, etc. by fitting the circular or spherical region in the point cloud data; the angle of the ball distribution is calculated, and the angle relationship of the ball is determined by analyzing the position distribution of the ball in the point cloud; the surface feature parameters are calculated using the point cloud data, such as estimating the surface roughness parameters by calculating the height change of the point cloud in the local area, and describing the surface texture characteristics by analyzing the local density change and curvature change of the point cloud.

[0025] Preferably, the set constraint conditions of the micro bearing are set, including the ball distribution angle tolerance, the inner and outer ring concentricity threshold, and the assembly stress distribution characteristics, wherein the ball distribution angle tolerance specifies the allowable deviation range of the distribution angle of the ball in the bearing, the inner and outer ring concentricity threshold represents the standard of the concentricity between the inner and outer rings, and the assembly stress distribution characteristics describes the reasonable distribution of the stress borne by each part of the bearing during assembly; then, according to the set geometric constraint conditions, the generated bearing three-dimensional point cloud parameters are dynamically spatially registered, that is, the point cloud parameter points are spatially transformed so that the point cloud data conforms to the geometric constraint conditions, and then a spatial transformation matrix is constructed, which contains transformation information such as translation, rotation, and scaling, and is used to transform the point cloud data to the correct spatial position and attitude.

[0026] Preferably, after the three-dimensional point cloud parameters are registered using the spatial transformation matrix, there may be differences between the actual point cloud data and the ideal state conforming to the geometric constraint conditions, i.e. registration residual error, then the registration residual error is analyzed, i.e. the difference between each point in the point cloud after transformation and the ideal position is calculated using the Euclidean distance, and then the registration residual error threshold is set according to the bearing manufacturing tolerance, detection accuracy, etc. The points with registration residual error greater than the registration residual error threshold are marked as points with deformation, and the local deformation difference areas on the micro bearing are identified (i.e. local deformation difference areas), the deformation degree in each local deformation difference area is not uniform, and contains multiple deformation gradient amplitudes, wherein the deformation gradient amplitude describes the change rate of deformation in the area, and different local deformation difference areas correspond to different deformation gradient amplitudes.

[0027] Preferably, since the deformation gradient amplitudes of different local deformation difference areas are different, the influence degree of the deformation gradient amplitudes on the performance and quality of the micro bearing is also different, then according to the size of the deformation gradient amplitude, a weight is assigned to each local deformation difference area, the larger the amplitude is, the higher the weight is, indicating that the possibility of the area having a problem is greater, and the influence on the bearing quality is also greater, and then a regional weight distribution map is constructed to intuitively show the risk degree of each area on the micro bearing; finally, detection positioning is performed according to the regional weight distribution map to determine a high-risk detection area, that is, a key detection area, so as to facilitate more accurate detection and further ensure that potential quality problems can be found in time.

[0028] Further, step S210 further includes step S211 of performing edge enhancement processing on the axial image data set to obtain three-dimensional space coordinate parameters, performing interlayer interpolation processing on the internal structure image data set, and extracting internal crack feature point cloud parameters according to the continuum data; step S212 of calling an assembly gap measurement value of the micro bearing to construct a gap distribution matrix, and constructing gap space point cloud parameters from the gap distribution matrix; and step S213 of fusing the three-dimensional space coordinate parameters, the internal crack feature point cloud parameters, and the gap space point cloud parameters to generate the bearing three-dimensional point cloud parameters.

[0029] Preferably, the edge enhancement processing on the axial image data set includes detecting regions with large gray scale changes in the image, that is, edges, using an image processing algorithm (such as a Sobel operator, a Canny operator, etc.), thereby highlighting the edge information of the micro bearing in the image, making the profile of the bearing more clear and obvious, and then combining an imaging model (such as a pinhole camera model) of the camera and camera parameters (such as focal length, shooting angle, etc.) to convert the edge points in the two-dimensional image to three-dimensional space to obtain three-dimensional space coordinates, and through space conversion of all points in the axial image data set, three-dimensional space coordinate parameters of the surface of the micro bearing are obtained to describe the geometric shape and position information of the bearing surface. The interlayer interpolation processing on the internal structure image data includes generating more intermediate layer data between discrete layered internal structure images using linear interpolation, spline interpolation, etc., filling the blank areas in the middle, making the data more continuous and smooth, and then obtaining more complete internal structure data; and then using image analysis processing (such as threshold segmentation) to identify and extract internal crack information, and converting the related points in the crack region into point cloud data to extract feature parameters of the point cloud, such as the position, shape, and size of the point cloud, to describe the characteristics of the internal crack, that is, the internal crack feature point cloud parameters.

[0030] Preferably, the assembly gap of the micro bearing is measured using a laser interferometer, by measuring the assembly gap at different positions to obtain multiple measurement values, and arranging the measurement values in order of the spatial positions of the measurement points to construct a gap distribution matrix to reflect the distribution of the assembly gap at different positions of the micro bearing; then according to the information in the gap distribution matrix, each measurement point and its corresponding gap value is converted into point cloud data in three-dimensional space, each point having three-dimensional coordinates (indicating the position of the measurement point) and an attribute representing the gap value, and a gap space point cloud parameter is constructed to describe the distribution characteristics of the assembly gap in space. Finally, the obtained three-dimensional spatial coordinate parameters (describing the geometric shape of the bearing surface), internal crack feature point cloud parameters (describing the internal crack features), and gap space point cloud parameters (describing the distribution of the assembly gap) are integrated to form a complete three-dimensional point cloud parameter set containing the information of the surface, internal structure and assembly gap of the micro bearing, i.e. the final bearing three-dimensional point cloud parameter, which is used for efficient and accurate detection and analysis of the micro bearing.

[0031] Further, step S220 further comprises step S221 of constructing a reference parameter feature tree of the micro bearing, performing multi-dimensional analysis through the reference parameter feature tree, performing assembly stress analysis according to the multi-dimensional analysis data, and constructing the set of geometric constraint conditions; step S222 of performing feature analysis on the bearing three-dimensional point cloud parameter to obtain a surface feature data set and a structure feature data set; step S223 of extracting edge contour feature points of the micro bearing based on the surface feature data set; and step S224 of performing rigid registration of the edge contour feature points and the structure feature data set according to the set of geometric constraint conditions to construct the space transformation matrix.

[0032] Preferably, a reference parameter feature tree of the micro bearing is constructed, i.e. various reference parameters (such as size parameters and structure parameters) of the micro bearing are organized into a tree structure in a tree structure, for example, the basic size (inner and outer ring diameter, ball diameter, etc.) of the bearing, the number of balls, the distribution angle, etc. are taken as nodes of the tree, and the relationship between the nodes reflects the hierarchical structure of the parameters. Then, through the reference parameter feature tree, multiple parameters of the micro bearing are analyzed comprehensively from different dimensions (such as size dimension and structure dimension) to analyze the mutual influence of the bearing parameters, for example, the influence of the number of balls and the distribution circumferential angle on the bearing load capacity, and the influence of the inner and outer ring diameter tolerance band on the assembly accuracy, etc., to obtain multi-dimensional analysis data including the number of balls N, the distribution circumferential angle θ, and the inner and outer ring diameter tolerance band ΔD, etc. Then, the assembly stress of the micro bearing is analyzed using finite element simulation, by discretizing the bearing model into multiple small units to calculate the stress distribution of each unit in the assembly process, thereby obtaining the assembly stress state of the entire bearing.

[0033] Preferably, finally according to the analysis results, the set of geometric constraints is determined, including the allowable range of the distribution angle deviation of the balls, the absolute value of the distribution angle deviation of each ball should be less than or equal to the distribution circumferential angle θ divided by 2N, to ensure the uniform distribution of the balls on the circumference, thereby ensuring the normal operation and carrying capacity of the bearing; the inner and outer ring center offset should be less than or equal to one-tenth of the inner and outer ring diameter tolerance band ΔD, to ensure the concentricity of the inner and outer rings, and to avoid bearing wear and performance degradation due to insufficient concentricity; the stress concentration coefficient of the contact area should be less than or equal to δ, wherein δ is the allowable stress concentration coefficient threshold value determined according to the bearing material and design requirements, to prevent material fatigue and damage due to excessive stress concentration.

[0034] Preferably, the three-dimensional point cloud parameters of the bearing are analyzed for features, specifically, by calculating the normal vector, curvature and other geometric properties of the point cloud, information related to surface features is extracted from the point cloud parameters to form a surface feature data set (such as surface roughness, texture, edge profile, etc.); At the same time, by analyzing the depth information and spatial distribution of the point cloud, structural features are extracted, information related to internal structural features is extracted to form a structural feature data set (such as internal cracks, material density distribution, etc.). In the surface feature data set, edge detection algorithms (such as Canny operator, Sobel operator, etc.) and contour extraction are used to identify and extract the edge profile of the micro bearing, that is, to detect the area with large changes in surface normal vector in the point cloud data, that is, the edge area, and to extract the feature points on the edge profile. Then according to the set of geometric constraints, the extracted edge profile feature points are rigidly registered with the structural feature data set, that is, the shape and size of the object do not change during the registration process, only translation and rotation operations are performed, and the position and attitude of the edge profile feature points are adjusted to make them conform to the geometric constraints related information in the structural feature data set; At the same time, the translation and rotation parameters required for matching and alignment are calculated to form a spatial transformation matrix, which is used to transform the edge profile feature points from the current position and attitude to the position and attitude that conform to the geometric constraints, and is also used to uniformly transform the entire point cloud data.

[0035] Step S300, traversing the key detection area to extract defect features, generating a defect probability distribution map, dynamically segmenting the defect probability distribution map to generate a defect type label group.

[0036] Preferably, the key detection area is traversed to extract defect features, that is, feature information representing defects is extracted from the point cloud data of the key detection area. Specifically, for surface defects, the height difference between each point and its neighborhood points is calculated, and the surface roughness is obtained by statistical analysis; the surface curvature is estimated by fitting a local surface or calculating the normal vector of the point cloud, and high curvature areas may represent edges, corners or potential defect locations; edge detection algorithms such as Canny operator, Sobel operator, etc. are used to identify edge points in the point cloud and determine edge irregularities, etc.; for internal defects, the point cloud density is extracted, that is, the point cloud density in each small area is counted, such as pores, loose, etc., the point cloud data is divided into a plurality of voxels (three-dimensional pixels), and the number or quality distribution of point clouds in each voxel is calculated. Based on support vector machine (SVM) and naive Bayes classifier, a defect probability model is established, the extracted defect features are taken as input, and the probability of each point or area belonging to a defect is predicted by training, and then the point cloud data of the key detection area is input into the defect probability model to calculate the probability of occurrence of defects, and a defect probability distribution map is generated to more clearly show the trend and characteristics of the probability distribution.

[0037] Preferably, the defect probability distribution map is dynamically segmented. Specifically, the gray value distribution of the defect probability distribution map is analyzed, and a suitable threshold is determined by empirical method, maximum inter-class variance method (OTSU), etc. For example, it is found that the defect probability is mainly concentrated in the area with high gray value, and the gray value 0.6 is taken as the initial threshold, the area with gray value greater than 0.6 in the probability distribution map is regarded as the possible defect area, and the area with gray value less than the value is regarded as the normal area; then according to the determined threshold, the image is segmented into foreground (possible defect area) and background (normal area). For the foreground area after preliminary segmentation, its related features are extracted, such as area, perimeter, centroid coordinates, gray mean value, etc., a suitable clustering algorithm is selected, such as K-means clustering algorithm, the foreground area is clustered according to the features, it is assumed that there are three different types of defects according to the morphology and probability distribution of the defects, the K value is set to 3, that is, the foreground area is divided into three different clusters, each cluster represents a possible defect type; and the corresponding area is assigned a defect type label, for example, the label "D1" is assigned to the area representing surface scratch, "D2" is assigned to the area representing internal crack, and "D3" is assigned to the area representing wear, etc. Finally, the defect type label group is output to intuitively show the defect distribution of the micro bearing.

[0038] Further, step S300 further comprises step S310, performing cross-modal feature fusion on the multi-dimensional detection data set to obtain multi-scale semantic features; step S320, performing defect analysis based on the multi-scale semantic features by traversing the key detection area, and performing dynamic weighting according to the defect analysis result to generate a defect feature tensor; and step S330, performing spatial up-sampling on the key detection area based on the defect feature tensor to generate the defect probability distribution map.

[0039] Preferably, the data features in the multi-dimensional detection data set are integrated, for example, the texture features in the optical image are combined with the three-dimensional geometric features obtained by laser scanning, and then the fused features are processed by using a convolutional neural network to extract multi-scale semantic features, i.e., semantic features of different scales. Specifically, small-scale features can capture detailed information of the bearing surface, such as micro scratches and wear; large-scale features can reflect the overall structure and morphology information of the bearing, such as the profile and overall deformation of the bearing; and the state of the bearing can be described more comprehensively. Then, according to the multi-scale semantic features, the key detection areas (such as the inner and outer ring surfaces of the bearing, the contact area between the ball and the raceway, etc.) are analyzed one by one, and by comparing the features with the defect mode or feature library, it is determined whether each area has defects and the possible defect type and severity; and according to the results of the defect analysis, different weights are given to different features, wherein if the features of a certain area indicate that there may be a serious crack defect, the weight of the features related to the crack is increased; and the weight of the features that are not likely to be related to the defect is reduced; through dynamic weighting, the features that are important to defect judgment are highlighted, the accuracy of defect detection is improved, and finally a defect feature tensor is generated, which contains various feature information of the key detection area and the corresponding weight, and can more compactly and accurately describe the defect features of the bearing.

[0040] Preferably, spatial up-sampling is performed on the key detection area based on the defect feature tensor, i.e., the defect feature tensor is enlarged or restored in the spatial dimension so that it can match the spatial resolution of the original key detection area, so as to more accurately locate the position of the defect on the surface or inside of the bearing, and then according to the up-sampled defect feature tensor, the probability of each spatial position appearing defect is calculated, and finally the probability value is represented in the form of an image to generate a defect probability distribution map, in which different colors or gray scales represent different probability values, so as to intuitively show the possible distribution of defects in the key detection area of the bearing.

[0041] Further, step S300 further comprises step S340 of performing graph feature analysis based on the defect probability distribution map to obtain histogram features, performing boundary demarcation based on the histogram features in combination with the defect feature tensor to generate a dynamic segmentation threshold; step S350 of performing morphological analysis on the defect probability distribution map, and calculating a plurality of adjacent defect regions according to a result of the morphological analysis; step S360 of performing dynamic segmentation identification on the defect probability distribution map according to the dynamic segmentation threshold in combination with the plurality of adjacent defect regions to generate geometric feature parameters of a plurality of adjacent regions; and step S370 of matching the geometric feature parameters with a preset defect type to generate the defect type label group.

[0042] Preferably, the defect probability distribution map is subjected to graph feature analysis, that is, the probability value distribution of pixels in the map is analyzed, and histogram features are generated by counting the frequency of different probability values, for example, counting the number of pixels in each interval such as 0 to 0.1, 0.1 to 0.2, etc. of the probability value to obtain a histogram of the probability distribution map to reflect the distribution of the probability value in the entire map; and the histogram features are combined with the defect feature tensor to consider the feature information of different regions, and the boundary between the defect and non-defect regions is demarcated according to the histogram features. Specifically, the distribution peak value, valley value, etc. of the probability value in the histogram are analyzed to determine a suitable probability value as the dynamic segmentation threshold. Then, morphological image processing methods such as erosion, dilation, opening operation, closing operation, etc. are used to perform morphological analysis on the defect probability distribution map, mainly to analyze the shape, size, connectivity, etc. of the defect regions in the map, for example, the erosion operation can remove isolated points on the boundary of the defect region, and the dilation operation can fill small holes in the defect region to optimize the morphological representation of the defect region. Finally, a plurality of adjacent defect regions are recognized and calculated according to the result of the morphological analysis, and the pixel regions whose probability value is higher than the dynamic segmentation threshold and are adjacent in space are divided into a defect region by judgment, and finally a plurality of different defect regions are obtained, each of which represents a possible defect position or range.

[0043] Preferably, the defect probability distribution map is converted into a binary image according to a dynamic segmentation threshold, wherein pixels with probability values higher than the threshold are marked as 1 (representing defect regions), and pixels with probability values lower than the threshold are marked as 0 (representing non-defect regions), and in the binary image, each connected domain, i.e. a defect region connected to each other, is marked, and for each marked connected domain, its geometric feature parameters such as area, perimeter, centroid coordinates, major axis length, minor axis length, etc. are calculated to describe the size, shape and position of each adjacent defect region. Finally, the geometric feature parameters of each adjacent defect region are matched with the pre-set feature templates of different defect types, wherein the pre-set surface scratch defect may have specific length and width ratio characteristics, the internal crack may have specific shape and position characteristics, etc. By comparing and matching, it is determined which pre-set defect type each defect region belongs to, and a label group containing defect position (such as centroid coordinates) and type is generated.

[0044] Step S400, based on the defect type label group, the micro bearing is sorted online, a sorting instruction set is obtained, and a detection quality report is generated and fed back to the production control system.

[0045] Preferably, the micro bearing on the production line is sorted online by using the defect type label group. Specifically, the sorting device grasps, moves and places the micro bearing according to the defect type label group, and sorts each bearing to the corresponding area. For example, if the label of a bearing shows "qualified", the sorting device guides it to the qualified product storage area; if it is a "surface scratch" defect bearing, it is guided to the repair or defective product processing area; and for a "internal crack" bearing with serious defects, it is directly guided to the waste area. Moreover, during the online sorting process, the control system generates a plurality of sorting instructions to clearly indicate how the sorting device operates the micro bearing. For example, the instructions may include "move bearing No. 1 to the waste area", "place the qualified bearing No. 4 on the 3rd layer of the qualified product storage rack", etc. All the sorting instructions for different bearings constitute a sorting instruction set, which is used to guide the sorting device to process each bearing in the correct order and manner, ensuring the efficiency and accuracy of the sorting work; and a micro bearing detection quality report is generated, which may include the total number of bearings detected, the number of qualified products, the number of bearings of different types of defects and their proportions, etc. It may also include analysis of defect distribution, such as the frequency of defects in different parts of the bearing, etc. Finally, the generated detection quality report is sent to the production control system, and the production process parameters are adjusted in time, such as checking the quality of raw materials and adjusting the precision of processing equipment, to improve product quality and reduce the generation of defective products.

[0046] Further, step S400 further comprises step S410, establishing a defect type-severity mapping table according to the defect type label set, constructing an initial sorting decision tree according to the defect type-severity mapping table; step S420, introducing batch detection speed of micro bearing to determine sorting time extreme value, performing detection timing analysis on the initial sorting decision tree according to the sorting time extreme value, and generating a detection descending order queue; step S430, formulating sorting priority according to the detection descending order queue, updating the initial sorting decision tree according to the sorting priority, and obtaining a sorting decision tree; and step S440, mapping the defect type label set to the sorting decision tree for multi-level classification, and generating the sorting instruction set.

[0047] Preferably, according to the defect type label set of micro bearing, different defect types such as crack, scratch and rust are defined with grade judgment rules, so as to establish a defect type-severity mapping table. For example, for crack, three grades of slight, moderate and severe are divided according to length, depth and other factors; scratch is divided into grades according to width, length and density; rust is graded according to area and rust degree. Then, the initial sorting decision tree is constructed based on the defect type-severity mapping table. Specifically, the defect type is taken as a tree node, branches correspond to different severity grades, and leaf nodes can be corresponding sorting decisions, such as sending bearings with slight defects to a repair area, sending bearings with moderate defects to a substandard product area, and sending bearings with severe defects to a waste product area.

[0048] Preferably, considering the batch detection speed of the micro bearing, the maximum sorting decision time allowed is calculated according to the line speed, that is, the sorting time extreme value, for example, if the production line produces 10 bearings per minute, the detection and sorting time of each bearing cannot exceed 6 seconds, and 6 seconds is the sorting time extreme value; then the initial sorting decision tree is analyzed in terms of detection time according to the sorting time extreme value, the time required for each decision path is analyzed, and the paths with longer detection time are arranged in front to generate a descending queue, for example, the detection and judgment of crack defects may require more time, while scratch defects relatively consume less time, and the detection path related to crack is arranged in front of the descending queue. Then, according to the descending queue, the sorting priority of the initial sorting decision tree is determined, specifically, the defect type and severity corresponding to the detection path arranged in front of the descending queue have higher sorting priority, and the decision tree depth and branch factor are optimized according to the sorting priority, which may include increasing the depth of the decision tree to optimize the high-priority branch and further classify the decision; simplify and merge low-priority branches to reduce decision time; and further obtain a more efficient sorting decision tree. Finally, the defect type label group is mapped to the sorting decision tree for multi-level classification, including judging the defect type and severity of each bearing according to the branches and nodes of the decision tree, and finally generating a specific sorting instruction set, for example, a bearing with moderate scratch defects is mapped to the sorting decision tree according to the corresponding branches and decision rules, and a sorting instruction for sending it to the defective product area is generated. Each bearing corresponds to a sorting instruction, which further constitutes a sorting instruction set for guiding the sorting of micro bearings.

[0049] Further, step S300 further comprises step S350 of obtaining a plurality of defect type occurrence frequencies according to the matching of the defect types in the historical sorting period according to the sorting instruction set, and constructing a defect distribution thermodynamic map according to the plurality of defect type occurrence frequencies; step S360 of performing time domain correlation identification according to the defect distribution thermodynamic map and the production process parameters of the micro bearing to determine a parameter abnormal correlation mode; step S370 of performing defect prediction based on the parameter abnormal correlation mode, drawing a defect prediction trend chart, compensating the production process parameters according to the defect prediction trend chart, and generating a process optimization suggestion; and step S380 of correlating and integrating the parameter abnormal correlation mode, the defect prediction trend chart and the process optimization suggestion to generate the detection quality report.

[0050] Preferably, the sorting instruction set is used to match the defect type corresponding to each instruction in the historical sorting period, and then the number of occurrences of different defect types is counted to calculate the occurrence frequency of multiple defect types, that is, to obtain the occurrence frequency of multiple defect types, and then a defect distribution heat map is constructed in a graphical manner. Specifically, different colors in the heat map represent different defect occurrence frequencies, and the deeper the color, the higher the frequency of the defect type, and the red area may represent a higher frequency of a certain defect, while the blue area represents a lower frequency, thereby intuitively displaying the distribution of various defects in the historical sorting. The defect distribution heat map is associated with the production process parameters (such as temperature, pressure, speed, processing time, etc.) of the micro bearing in the time dimension, and the relationship between the change of the production process parameters and the occurrence frequency of the defect type in different time periods is analyzed, and then the abnormal association mode between the production process parameters and the defect type is determined, for example, when the pressure exceeds a certain set value, the occurrence frequency of internal crack defects increases significantly, thereby determining the parameter abnormal association mode, which helps to identify which changes in production process parameters may cause defects.

[0051] Preferably, based on the determined parameter abnormal association mode, the future defect situation is predicted, that is, according to the current and historical production process parameters, the frequency and trend of different defect types that may occur in the future are predicted, and then a defect prediction trend chart is drawn to intuitively display the prediction results, and then according to the defect prediction trend chart, the production process parameters that may lead to an increase in defects are compensated and adjusted, for example, if it is predicted that an increase in temperature will lead to an increase in a certain defect, the temperature set value is appropriately reduced; at the same time, process optimization suggestions are generated according to the compensation adjustment, including specific parameter adjustment values, adjustment times, etc., to guide the optimization of the production process. Finally, the parameter abnormal association mode, the defect prediction trend chart, and the process optimization suggestions are associated and integrated to generate a detection quality report, including the historical defect distribution, the association between the production process parameters and the defects, the future defect prediction, and the corresponding process optimization measures, thereby improving the production efficiency and product quality of the micro bearing.

[0052] In the foregoing, the Figure 1 The micro bearing rapid batch online detection method according to the embodiment of the present application is described in detail. Next, the Figure 2 The micro bearing rapid batch online detection system according to the embodiment of the present application will be described.

[0053] The micro bearing rapid batch online detection system according to the embodiment of the present application is used to solve the technical problems of low detection efficiency, insufficient sampling accuracy, and difficulty in real-time online detection and control of bearing quality in the prior art, to realize micro bearing rapid batch online detection, and to achieve the technical effect of improving the detection efficiency and accuracy of micro bearings. As Figure 2As shown, the micro-bearing-oriented rapid batch online detection system comprises a cooperative acquisition module 10, a dynamic spatial registration module 20, a dynamic segmentation module 30, and an online sorting module 40.

[0054] The cooperative acquisition module 10 is configured to cooperatively acquire the micro-bearing to generate a multi-dimensional detection data set. The dynamic spatial registration module 20 is configured to dynamically spatially register the multi-dimensional detection data set to locate a key detection area of the micro-bearing. The dynamic segmentation module 30 is configured to traverse the key detection area to extract defect features and generate a defect probability distribution map, and dynamically segment the defect probability distribution map to generate a defect type label group. The online sorting module 40 is configured to sort the micro-bearing based on the defect type label group to obtain a sorting instruction set and generate a detection quality report for feedback to a production control system.

[0055] Next, the specific configuration of the cooperative acquisition module 10 will be described in detail. The cooperative acquisition module 10 further comprises an adjustable focus ring-shaped light source array arranged on both sides of a bearing transmission track, and an initial light spectrum combination is set. The bearing rotation speed is acquired in real time, and when the micro-bearing enters the imaging area, the high-speed line array camera is triggered to capture the micro-bearing according to the bearing rotation period in combination with the bearing rotation speed to obtain an axial image data set of the micro-bearing. The internal structure of the micro-bearing is cooperatively acquired based on the initial light spectrum combination to obtain an internal structure image data set. The axial image data set and the internal structure image data set are spatio-temporally aligned to obtain the multi-dimensional detection data set.

[0056] Next, the specific configuration of the dynamic spatial registration module 20 will be described in detail. The dynamic spatial registration module 20 further comprises point cloud reconstruction based on the multi-dimensional detection data set to generate bearing three-dimensional point cloud parameters. A set of geometric constraints of the micro-bearing is set, and the bearing three-dimensional point cloud parameters are dynamically spatially registered according to the set of geometric constraints to construct a spatial transformation matrix. Registration residual analysis is performed according to the spatial transformation matrix to identify a plurality of local deformation difference areas, the plurality of local deformation difference areas contain a plurality of deformation gradient amplitudes, and the plurality of deformation gradient amplitudes have a corresponding relationship with the plurality of local deformation difference areas. The plurality of local deformation difference areas are weight-distributed according to the plurality of deformation gradient amplitudes to construct a regional weight distribution map, and detection positioning is performed according to the regional weight distribution map to determine the key detection area.

[0057] Next, the specific configuration of the dynamic spatial registration module 20 will be described in detail. The dynamic spatial registration module 20 further comprises: performing edge enhancement processing on the axial image data set, obtaining three-dimensional spatial coordinate parameters, performing interlayer interpolation processing on the internal structure image data set, and extracting internal crack feature point cloud parameters according to the continuum data; calling the assembly gap measurement value of the micro bearing to construct a gap distribution matrix, and constructing a gap space point cloud parameter from the gap distribution matrix; fusing the three-dimensional spatial coordinate parameters, the internal crack feature point cloud parameters, and the gap space point cloud parameters to generate the bearing three-dimensional point cloud parameters.

[0058] Next, the specific configuration of the dynamic spatial registration module 20 will be described in detail. The dynamic spatial registration module 20 further comprises: constructing a reference parameter feature tree of the micro bearing, performing multi-dimensional analysis through the reference parameter feature tree, performing assembly stress analysis according to the multi-dimensional analysis data, and constructing the set of geometric constraints; performing feature analysis on the bearing three-dimensional point cloud parameters to obtain a surface feature data set and a structure feature data set; extracting edge contour feature points of the micro bearing based on the surface feature data set; and performing rigid registration of the edge contour feature points and the structure feature data set according to the set of geometric constraints to construct the spatial transformation matrix.

[0059] Next, the specific configuration of the dynamic segmentation module 30 will be described in detail. The dynamic segmentation module 30 further comprises: performing cross-modal feature fusion on the multi-dimensional detection data set to obtain multi-scale semantic features; performing defect analysis based on the multi-scale semantic features by traversing key detection regions, performing dynamic weighting according to the defect analysis results, and generating a defect feature tensor; performing spatial upsampling on the key detection regions based on the defect feature tensor to generate the defect probability distribution map.

[0060] Next, the specific configuration of the dynamic segmentation module 30 will be described in detail. The dynamic segmentation module 30 further comprises: performing graph feature analysis based on the defect probability distribution map to obtain histogram features, performing boundary demarcation based on the histogram features in combination with the defect feature tensor to generate a dynamic segmentation threshold, performing morphological analysis on the defect probability distribution map, and calculating a plurality of adjacent defect regions according to the graph morphological analysis results; performing dynamic segmentation identification on the defect probability distribution map according to the dynamic segmentation threshold in combination with the plurality of adjacent defect regions to generate a plurality of geometric feature parameters of adjacent regions; and matching the geometric feature parameters with a preset defect type to generate a defect type label group.

[0061] Next, the specific configuration of the online sorting module will be described in detail. The online sorting module further comprises: establishing a defect type-severity mapping table according to the defect type label group, constructing an initial sorting decision tree according to the defect type-severity mapping table; introducing batch detection speed of micro bearings to determine sorting time extreme value, performing detection timing analysis on the initial sorting decision tree according to the sorting time extreme value, and generating a detection descending order queue; formulating sorting priorities for the initial sorting decision tree according to the detection descending order queue, updating the initial sorting decision tree according to the sorting priorities, and obtaining a sorting decision tree; and mapping the defect type label group to the sorting decision tree for multi-level classification, and generating the sorting instruction set.

[0062] Next, the specific configuration of the online sorting module will be described in detail. The online sorting module further comprises: establishing a defect type-severity mapping table according to the defect type label group, constructing an initial sorting decision tree according to the defect type-severity mapping table; introducing batch detection speed of micro bearings to determine sorting time extreme value, performing detection timing analysis on the initial sorting decision tree according to the sorting time extreme value, and generating a detection descending order queue; formulating sorting priorities for the initial sorting decision tree according to the detection descending order queue, updating the initial sorting decision tree according to the sorting priorities, and obtaining a sorting decision tree; and mapping the defect type label group to the sorting decision tree for multi-level classification, and generating the sorting instruction set.

[0063] The rapid batch online detection system for micro bearings provided in the embodiments of the present application can perform the rapid batch online detection method for micro bearings provided in any of the embodiments of the present application, and has the corresponding function modules and beneficial effects of performing the method.

[0064] Although the present application makes various references to certain modules in the system according to the embodiments of the present application, however, any number of different modules can be used and run on the user terminal and / or server, and the various units and modules are only divided according to functional logic, but are not limited to the above division, as long as the corresponding functions can be realized; in addition, the specific names of each functional unit are only for the convenience of mutual differentiation, and do not limit the protection scope of the present application.

[0065] The above specific embodiments do not constitute a limitation on the protection scope of the present application. Those skilled in the art should understand that various modifications, combinations and substitutions can be made according to design requirements and other factors. Any modifications, equivalent substitutions and improvements made within the spirit and principles of the present application shall be included in the protection scope of the present application.

Claims

1. A rapid batch online detection method for micro bearing, characterized in that, The method comprises: Coordinated collection of micro bearings generates multidimensional detection data sets; Dynamic spatial registration of the multidimensional detection data sets locates key detection areas of micro bearings; Defect feature extraction in the key detection areas generates defect probability distribution maps, and dynamic segmentation of the defect probability distribution maps generates defect type label groups; Based on the defect type label groups, online sorting of micro bearings is performed to obtain a sorting instruction set, and a detection quality report is generated and fed back to a production control system; Coordinated collection of micro bearings generates multidimensional detection data sets, the method comprising: An adjustable focus ring light source array is arranged on both sides of a bearing transmission track, and an initial light spectrum combination is set; The bearing rotation speed is acquired in real time, and when a micro bearing enters an imaging area, a high-speed linear array camera is triggered to collect the micro bearing according to the bearing rotation period in combination with the bearing rotation speed, thereby obtaining an axial image data set of the micro bearing; Based on the initial light spectrum combination, the internal structure of the micro bearing is collected in coordination to obtain an internal structure image data set; The axial image data set and the internal structure image data set are spatio-temporally aligned to obtain the multidimensional detection data set; Dynamic spatial registration of the multidimensional detection data sets locates key detection areas of micro bearings, the method comprising: Point cloud reconstruction is performed based on the multidimensional detection data sets to generate bearing three-dimensional point cloud parameters; Geometric constraint condition sets of micro bearings are set, and dynamic spatial registration of the bearing three-dimensional point cloud parameters is performed according to the geometric constraint condition sets to construct a spatial transformation matrix; Registration residual analysis is performed according to the spatial transformation matrix to identify a plurality of local deformation difference areas, the plurality of local deformation difference areas contain a plurality of deformation gradient amplitudes, and the plurality of deformation gradient amplitudes have a corresponding relationship with the plurality of local deformation difference areas; According to the plurality of deformation gradient amplitudes, weight distribution is allocated to the plurality of local deformation difference areas to construct a region weight distribution map, and detection positioning is performed according to the region weight distribution map to determine the key detection areas; Defect feature extraction in the key detection areas generates defect probability distribution maps, the method comprising: Cross-modal feature fusion is performed on the multidimensional detection data sets to obtain multi-scale semantic features; Based on the multi-scale semantic features, defect analysis is performed in the key detection areas, and a defect feature tensor is generated according to the defect analysis results; Based on the defect feature tensor, spatial up-sampling is performed on the key detection areas to generate the defect probability distribution maps; Dynamic segmentation of the defect probability distribution maps generates defect type label groups, the method comprising: Histogram feature analysis is performed based on the defect probability distribution maps to obtain histogram features, and dynamic segmentation threshold values are generated by combining the histogram features with the defect feature tensor to demarcate boundaries; Morphological analysis is performed on the defect probability distribution maps, and a plurality of adjacent defect areas are calculated according to the morphological analysis results; Dynamic segmentation of the defect probability distribution maps is performed according to the dynamic segmentation threshold values in combination with the plurality of adjacent defect areas to generate geometric feature parameters of a plurality of adjacent areas; According to the matching of the geometric feature parameters and the preset defect type, the defect type label group is generated.

2. The method for rapid batch on-line detection of micro-bearings according to claim 1, wherein, Based on the multi-dimensional detection data set, a point cloud reconstruction is performed to generate a bearing three-dimensional point cloud parameter, and the method comprises: An edge enhancement processing is performed on the axial image data set to obtain a three-dimensional space coordinate parameter, and an interlayer interpolation processing is performed on the internal structure image data set to extract an internal crack feature point cloud parameter according to the continuum data; A micro bearing assembly gap measurement value is called to construct a gap distribution matrix, and the gap distribution matrix is constructed into a gap space point cloud parameter; The three-dimensional space coordinate parameter, the internal crack feature point cloud parameter and the gap space point cloud parameter are fused to generate the bearing three-dimensional point cloud parameter.

3. The method for rapid batch on-line detection of micro-bearings according to claim 1, wherein, A set of geometric constraint conditions of the micro bearing is set, and a dynamic space registration is performed on the bearing three-dimensional point cloud parameter according to the set of geometric constraint conditions to construct a space transformation matrix, and the method comprises: A reference parameter feature tree of the micro bearing is constructed, a multi-dimensional analysis is performed through the reference parameter feature tree, an assembly stress analysis is performed according to the multi-dimensional analysis data, and the set of geometric constraint conditions is constructed; A feature analysis is performed on the bearing three-dimensional point cloud parameter to obtain a surface feature data set and a structure feature data set; An edge contour feature point of the micro bearing is extracted based on the surface feature data set; The edge contour feature point is rigidly registered with the structure feature data set according to the set of geometric constraint conditions to construct the space transformation matrix.

4. The method for rapid batch on-line micro bearing inspection as claimed in claim 1, wherein, Based on the defect type label group, an online sorting of the micro bearing is performed to obtain a sorting instruction set, and the method comprises: A defect type-severity mapping table is established according to the defect type label group, and an initial sorting decision tree is constructed according to the defect type-severity mapping table; A sorting time extreme value is determined according to the batch detection speed of the micro bearing, a detection time sequence analysis is performed on the initial sorting decision tree according to the sorting time extreme value to generate a detection descending queue; A sorting priority is formulated for the initial sorting decision tree according to the detection descending queue, the initial sorting decision tree is updated according to the sorting priority to obtain a sorting decision tree; The defect type label group is mapped to the sorting decision tree for multi-level classification to generate the sorting instruction set.

5. The method for rapid batch on-line micro bearing inspection as claimed in claim 1, wherein, The process of generating the detection quality report comprises: According to the matching of the sorting instruction set and the defect types in a historical sorting period, a plurality of defect type occurrence frequencies are obtained, and a defect distribution thermodynamic map is constructed according to the plurality of defect type occurrence frequencies; According to the time domain correlation identification of the defect distribution thermodynamic map and the production process parameters of the micro bearing, a parameter abnormal correlation mode is determined; Based on the parameter abnormal correlation mode, a defect prediction is performed, a defect prediction trend graph is drawn, and a process optimization suggestion is generated according to the defect prediction trend graph; The parameter abnormal correlation mode, the defect prediction trend graph and the process optimization suggestion are associated and integrated to generate the detection quality report.

6. A rapid batch on-line detection system for micro-bearings, characterized by, The system is used to implement the rapid batch online detection method for micro bearings according to any one of claims 1 to 5, and the system comprises: The cooperative acquisition module is configured to cooperatively acquire the micro bearing to generate a multi-dimensional detection data set; The dynamic space registration module is configured to perform dynamic space registration on the multi-dimensional detection data set to locate a key detection area of the micro bearing; The dynamic segmentation module is configured to traverse the key detection area to extract defect features, generate a defect probability distribution map, perform dynamic segmentation on the defect probability distribution map, and generate a defect type label group; The online sorting module is configured to sort the micro bearing based on the defect type label group to obtain a sorting instruction set and generate a detection quality report for feedback to a production control system.

Images