Bearing defect modular detection device and method based on machine vision

By designing four independent detection stations and modular optical imaging combination solutions in bearing detection equipment, the problems of poor detection complexity and versatility of existing equipment are solved, and efficient detection of multiple bearing specifications and varieties are achieved, reducing the number of times of bearings Cost and improve device compatibility.

CN113205499BActive Publication Date: 2024-11-26MATRIXTIME ROBOTICS (SHANGHAI) CO LTD
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202110481737.X
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2021-04-30
Publication Date
2024-11-26
Estimated Expiration
2041-04-30

AI Technical Summary

Technical Problem

Existing bearing testing equipment cannot be compatible with the detection of bearings in different process sections on the same production line, resulting in complex testing processes and poor versatility, which cannot meet the testing needs of multiple specifications and varieties.

Method used

A modular detection device for bearing defects based on machine vision is designed, including four independent detection stations, which are used to collect bearing outer ring, inner ring, end face, inner chamfered face, outer circular surface and inner circumferential surface respectively. The image of the bearing is analyzed through a specific image detection algorithm, and the test results are comprehensively used to determine the qualification of the bearing. The device adopts an innovative combination of optical imaging solutions and modular design, suitable for inspection of finished bearings, outer and inner rings.

Benefits of technology

It improves the compatibility and versatility of the equipment, reduces application costs, and can detect bearings of different specifications and varieties under the same modular imaging scheme, realizes customized design, and is suitable for the detection of bearings and similar metal surface defects.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN113205499B_ABST
    Figure CN113205499B_ABST
Patent Text Reader

Abstract

The present invention relates to the field of machine vision technology, and discloses a modular detection device and method for bearing defects based on machine vision, comprising four independent detection stations, which can respectively complete image acquisition of the end face, inner circle chamfered surface, outer circle surface, outer chamfered surface and inner circumferential surface of the bearing outer ring / bearing inner ring, and perform analysis and processing through an image analysis device with a preset specific image detection algorithm, so as to detect possible surface defects of the bearing and classify them, and finally combine the detection results of the four stations to judge whether the detected bearing is qualified, thereby solving the defects of the existing bearing detection method, such as complex detection process and poor versatility.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention relates to the technical field of machine vision, and in particular to a modular detection device and method for bearing defects based on machine vision. Background Art

[0002] As one of the most important basic parts in the industrial field, bearings are widely used and come in many types. In the past, the focus of bearing quality control was more on the processing accuracy and mechanical properties of bearings, because these are related to the various performance and life of mechanical products. However, in recent years, the appearance quality of bearings has received more and more attention. On the one hand, it affects the appearance of the product. On the other hand, surface defects (such as bumps, rust, cracks, black skin, etc.) will also seriously affect the performance and life of mechanical products.

[0003] At present, most small bearing processing plants mainly combine semi-automation with manual inspection. Large bearing plants have installed some imported or domestic automatic defect detection equipment based on machine vision, which has reduced labor and costs to a certain extent, and the detection results are stable. In the entire process of bearing processing, the key components are the outer ring, inner ring and roller. At present, there are inspection equipment for the inner and outer rings, rollers and finished products of bearings.

[0004] However, looking at the bearing visual inspection equipment on the market, as well as related academic papers and patent documents, they all meet the limited inspection needs of a single product, such as bearing end face inspection equipment, bearing ring inspection equipment, finished bearing inspection equipment, etc. Since the position of the optical device in the visual inspection equipment is relatively fixed, and there are many types and specifications of bearings, and the inspection requirements vary greatly, it is impossible to achieve compatibility inspection for finished bearings and bearing rings, or even inner and outer rings. If you want to perform quality control on each process section on a whole bearing production line, you need to customize different products for different workpieces. Summary of the invention

[0005] 1. Technical issues to be resolved

[0006] An embodiment of the present invention provides a modular detection device for bearing defects based on machine vision, including four independent detection stations, which can respectively complete image acquisition of the end face, inner circle chamfered surface, outer circle surface, outer chamfered surface and inner circumferential surface of the bearing outer ring / bearing inner ring, and perform analysis and processing through an image analysis device with a preset specific image detection algorithm, so as to detect possible surface defects of the bearing and classify them. Finally, the detection results of the four stations are combined to judge whether the detected bearing is qualified, thereby solving the defects of the existing bearing detection method, such as complex detection process and poor versatility. (II) Summary of the invention

[0008] The embodiment of the present invention provides a modular bearing defect detection device based on machine vision, comprising four groups of detection stations, namely a first detection station, a second detection station, a third detection station and a fourth detection station;

[0009] The first inspection station and the second inspection station include an inspection platform, a low-angle annular light source arranged above the inspection platform, and a high-angle annular light source arranged above the low-angle annular light source, and an area array camera with a full-frame lens is also arranged above the high-angle annular light source;

[0010] The third inspection station includes an inspection platform, an arc light source disposed above the inspection platform, and a linear array camera disposed above the arc light source, wherein the linear array camera has a telecentric lens;

[0011] The fourth inspection station includes an inspection platform, an arc light source disposed above the inspection platform, and a linear array camera disposed above the arc light source, wherein the linear array camera has a telecentric lens, and a prism is further disposed between the inspection platform and the arc light source, wherein the prism is used to be placed inside the inner ring / outer ring of the bearing to be inspected;

[0012] It also includes an image storage device for receiving images collected by the area array camera and the line array camera, and an image analysis device for processing the above images according to a set algorithm.

[0013] Preferably, the third inspection station further includes an encoder for triggering the linear array camera.

[0014] The modular detection method for bearing defects based on machine vision uses the above-mentioned modular detection device for bearing defects based on machine vision, and includes the following steps:

[0015] Step 1: Place the bearing workpiece to be inspected on the inspection table of the first inspection station, collect images of its first end face and first inner chamfered surface and transmit them to an image storage device;

[0016] Step 2: flip the bearing workpiece and transport it to the inspection table of the second inspection station, collect images of its second end face and second inner chamfered surface and transmit them to the image storage device;

[0017] Step 3: transport the bearing workpiece to the inspection table of the third inspection station, adjust the arc light source so that it illuminates the outer cylindrical surface and the outer chamfered surface of the bearing workpiece at the same time, and while rotating the bearing workpiece, collect a complete image of the outer cylindrical surface and the outer chamfered surface of the bearing workpiece and transmit them to an image storage device;

[0018] Step 4: transport the bearing workpiece to the inspection table of the fourth inspection station, place the prism inside the bearing workpiece, and collect the image of the inner circumference of the bearing workpiece while rotating the bearing workpiece and transmit it to the image storage device;

[0019] Step 5: The image analysis device analyzes whether the inspected bearing workpiece is qualified according to the image in the image storage device and a preset algorithm.

[0020] Preferably, in step three, a trigger is used to control the photographing frequency of the linear array camera.

[0021] Preferably, in step three, the line frequency of the camera is determined according to the diameter of the bearing to ensure the integrity of the image of the outer surface of the bearing collected.

[0022] Preferably, in step five, any one of Faster RCNN, SSD, YOLO vn algorithm models or improved versions thereof is preset in the image analysis device.

[0023] (III) Beneficial effects

[0024] The modular detection device and method for bearing defects based on machine vision provided in the embodiment of the present invention include four independent detection stations, which can respectively complete the image acquisition of the end face, inner circle chamfered surface, outer circle surface, outer chamfered surface and inner circumferential surface of the bearing outer ring / bearing inner ring, and perform analysis and processing through an image analysis device with a preset specific image detection algorithm to detect possible surface defects of the bearing and classify them. Finally, the detection results of the four stations are combined to judge whether the detected bearing is qualified.

[0025] Firstly, the innovative optical imaging combination scheme is adopted. On the one hand, it ensures that each measured surface of the bearing can be clearly imaged. On the other hand, the finished bearing, bearing outer ring and bearing inner ring can be inspected under the same modular imaging scheme, without the need to customize special equipment separately. This improves the compatibility and versatility of the equipment and reduces the application cost of the equipment.

[0026] Secondly, by adopting the modular design of optical imaging hardware and the modular design concept of corresponding software, customized design of visual defect detection of finished bearings and inner and outer rings is realized according to user needs to meet the detection needs of bearings of different specifications and varieties.

[0027] Finally, the modular design concept of software and hardware can be easily extended to visual defect detection of metal surfaces similar to bearings. Systems can be quickly designed to separately detect the annular surface, line scanning outer cylindrical surface, and inner cylindrical surface of workpieces, providing a new solution idea for parts inspection processes based on machine vision technology. BRIEF DESCRIPTION OF THE DRAWINGS

[0028] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0029] Figure 1 is a structural schematic diagram of a modular bearing defect detection device based on machine vision in an embodiment of the present invention;

[0030] Figure 2 It is the correspondence between different types of detection bearing workpieces and corresponding detection surfaces in the embodiment of the present invention;

[0031] Figure 3 It is a design and optimization method of a deep learning method preset in a modular bearing defect detection device based on machine vision in an embodiment of the present invention.

[0032] Description of reference numerals:

[0033] 1: low-angle ring light source; 2: high-angle ring light source; 3: area array camera;

[0034] 4: Linear array camera; 5: Telecentric lens; 6: Arc light source;

[0035] 7: Prism. DETAILED DESCRIPTION

[0036] In order to make the purpose, technical solution and advantages of the embodiments of the present invention clearer, the technical solution in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0037] In the description of the embodiments of the present invention, it should be noted that, unless otherwise clearly specified and limited, the terms "first", "second" and "third" are for the purpose of clearly describing the numbering of product components and do not represent any substantial difference. The directions of "up", "down", "left" and "right" are all based on the directions shown in the accompanying drawings. For ordinary technicians in this field, the specific meanings of the above terms in the embodiments of the present invention can be understood according to specific circumstances.

[0038] It should be noted that, unless otherwise clearly specified and limited, the term "connection" should be understood in a broad sense, for example, it can be directly connected or indirectly connected through an intermediate medium. For ordinary technicians in this field, the specific meanings of the above terms in the embodiments of the invention can be understood in specific circumstances.

[0039] Figure 1 It is a modular bearing defect detection device based on machine vision in an embodiment of the present invention, comprising four groups of detection stations, namely a first detection station, a second detection station, a third detection station and a fourth detection station; the first detection station and the second detection station are respectively used to detect two end faces and two inner chamfers, each detection station detects the end face and the inner chamfer on the same side at one time, and since the basic size and structure of the two end faces are the same, the structures of the first detection station and the second detection station can be completely the same, both comprising a detection platform, a low-angle annular light source 1 arranged above the low-angle annular light source 1, and a high-angle annular light source 2 arranged above the low-angle annular light source 1, and a surface array camera 3 with a full-frame lens is also arranged above the high-angle annular light source 2;

[0040] The third inspection station is used to inspect the outer surface and the outer chamfered surface, which includes an inspection table, an arc light source 6 arranged above the inspection table, and a linear array camera 4 arranged above the arc light source 6, and the linear array camera 4 has a telecentric lens 5; the third inspection station also includes an encoder for triggering the linear array camera 4, and the arc light source 6 is used to illuminate the outer cylindrical surface and the outer chamfered surface. The rotating workpiece uses a camera line to scan the outer circumferential surface, and the encoder is used to trigger the camera to collect images. For different diameter sizes of the bearing, the camera line frequency can be accurately set to collect the complete outer surface image; due to the difference in the depth direction between the outer cylindrical surface and the outer chamfered surface, the telecentric lens 5 is used to eliminate the projection imaging error.

[0041] The fourth inspection station includes an inspection table, an arc light source 6 arranged above it, and a linear array camera 4 arranged above the arc light source 6. The linear array camera 4 has a telecentric lens 5. A prism 7 is also arranged between the inspection table and the arc light source 6. The prism 7 is used to be placed inside the inner ring / outer ring of the bearing to be inspected. The inspection method is: rotate the bearing workpiece to be inspected and place a prism 7 inside the workpiece to scan the inner cylindrical surface of the workpiece by reflection. The light source is reflected by the prism 7, and the light is refracted to illuminate the inner wall. After being reflected on the inner wall surface, it is reflected by the prism 7 to the lens and collected by the camera. Since the inner cylindrical surface may be a curved surface with a certain depth, a telecentric lens 5 is used to eliminate the projection imaging error. It is worth mentioning that this solution is not for miniature bearings, because it is impossible to place a prism 7 inside them. Too small bearings require other methods for inner hole imaging.

[0042] It also includes an image storage device for receiving images collected by the area array camera 3 and the line array camera 4, and an image analysis device for processing the above images according to a set algorithm.

[0043] This modular design method uses different light sources for different bearing types, so that different tested surfaces can be imaged clearly while ensuring good contrast for defects. Appropriate light sources and imaging devices are used to unify the inspection requirements of these bearing outer rings, bearing inner rings and finished bearings into a set of inspection solutions. While realizing the inspection of each bearing surface, it provides a complete solution with strong compatibility that meets different inspection objects.

[0044] According to the homogeneity of imaging and algorithm processing, the above detection planes can be placed in four stations. The specific detection methods are as follows:

[0045] Step 1: The first inspection station inspects the single-side end face and inner chamfered surface of the workpiece: the light source adopts a combination of high-angle ring light and low-angle ring light, which not only takes into account the simultaneous imaging of the end face and the inner chamfered surface, but also weakens the texture of the metal material to a certain extent through the combined light, which is conducive to the subsequent algorithm to better detect defects;

[0046] Step 2: After the workpiece is turned over, enter the second inspection station to inspect the other side end face and the inner chamfered surface. The scheme is the same as the first inspection station;

[0047] Step 3: The third inspection station inspects the outer cylindrical surface and the outer chamfered surface: an arc light source 6 is used to illuminate the outer cylindrical surface and the outer chamfered surface, the rotating workpiece uses a camera line scan of the outer circumference, and an encoder is used to trigger the camera to collect images. For different diameters of the bearing, the camera line frequency can be accurately set to collect the complete outer surface image; since the outer cylindrical surface and the outer chamfered surface are different in the depth direction, a telecentric lens 5 is used to eliminate the projection imaging error;

[0048] Step 4: The fourth inspection station inspects the inner circumferential surface: rotate the workpiece and place a prism 7 inside the workpiece to scan the inner circumferential surface of the workpiece by reflection. The light source is reflected by the prism 7, and the light is folded 90 degrees and illuminates the inner wall. After being reflected on the inner wall surface, it is reflected by the prism 7 to the lens and collected by the camera (this solution is not for miniature bearings, because it is impossible to place a prism 7 inside them, and bearings that are too small require other methods to image the inner hole); because the inner circumferential surface may be a curved surface with a certain depth, a telecentric lens 5 is used to eliminate projection imaging errors.

[0049] Each workpiece passes through four stations in sequence, and four images are obtained respectively. After algorithm processing, defects in the image can be detected and classified. After the four stations are completed, the software integrates the inspection results of the four stations and conducts a comprehensive assessment according to the quantitative standards for bearing quality assessment. If the requirements are exceeded, the workpiece will be rejected as a defective product in the subsequent station. If the requirements are met, it is a good product. The preset algorithm in the image analysis device uses any of the mature open source algorithm models such as Faster RCNN, SSD, YOLO vn, or their improved versions.

[0050] In order to illustrate the combined detection method, deep groove ball bearings are taken as an example. Figure 2 For illustration, as shown in the figure, the finished bearing uses the first inspection station to inspect the upper end face, the second inspection station to inspect the lower end face, the third inspection station to inspect the outer cylindrical surface, and the fourth inspection station to inspect the inner cylindrical surface; the combination relationship between the inner ring and the outer ring of the bearing is shown in the figure, which will not be repeated here. Relying on this combination method, the design difficulty and application cost of this type of testing equipment are greatly reduced, while improving its versatility.

[0051] Regarding versatility, the compatibility analysis of the workpiece types being measured is as follows: Generally, the diameter size range of industrial bearings is relatively large, ranging from a dozen millimeters to several hundred millimeters. Under the condition of ensuring the same image resolution, the bearing specifications can be divided into several large ranges, and cameras with the same target surface size and the same pixel resolution can be used to cover the imaging of bearings in multiple size ranges; for the end face inspection station, a combination of high-angle ring light and low-angle ring light is used. High-angle ring light is not sensitive to size; for low-angle ring light, when illuminating the inner chamfered surface, if the reflection angle changes due to the size, the corresponding height of the low-angle light source can be adjusted according to the diameter change when the bearing specifications are changed, which also adjusts the irradiation angle of the low-angle ring light and ensures the consistency of lighting; for the outer cylindrical surface station, the camera lens light source can be adjusted as a whole. For different bearing diameters, it is only necessary to ensure that the camera lens light source and the outer circle of the bearing maintain the same working distance; at the same time, the arc light illumination of the outer surface of the bearing is multi-angle illumination, which is insensitive to changes in the thickness direction of the bearing; for the inner cylindrical surface station, similar to the outer cylindrical surface station, the camera lens light source is assembled as a whole, and it is moved as a whole when the bearing diameter changes to ensure the same working distance; arc light is also insensitive to changes in the thickness direction of the bearing. In summary, there is no problem with compatibility.

[0052] The algorithm is not the core part of the modular design method of the present invention, because no matter whether the device itself adopts modular design or not, the detection algorithm needs to be modularly designed to cope with different detection scenarios.

[0053] According to Figure 2As shown in the classification of different surfaces, for different types of finished bearings, the images are not much different and can be processed with similar algorithm modules; but for the inner and outer rings of different bearings, the images will obviously be quite different, and corresponding algorithm modules need to be designed for different types, regardless of whether traditional image processing algorithms or deep learning algorithms are used.

[0054] On the other hand, for the same surface, the imaging of the metal surface of the bearing will be different due to changes in the material, processing method and process. Usually the image will have background texture and noise, but the difference will not be too large. For traditional image algorithms, it is not easy to adapt to changes in different complex backgrounds. In order to cooperate with the idea of ​​modular design of this patent, consider using a deep learning algorithm to classify materials with similar image backgrounds into one category, collect the corresponding defect image sample data set, and after the defects are labeled, input the designed algorithm model, and optimize the model output results by adjusting the model hyperparameters and training; by updating the data set, adjusting the labeling accuracy, adjusting the algorithm model combination, and optimizing the model parameters, the learning process of the model is continuously iterated to gradually form an algorithm that can be applied online. The specific algorithm design and optimization process are as follows: Figure 3 shown.

[0055] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A modular bearing defect detection device based on machine vision. It is characterized in that It includes four groups of inspection stations, namely the first inspection station, the second inspection station, the third inspection station and the fourth inspection station; The first inspection station and the second inspection station include an inspection platform, a low-angle annular light source disposed above the inspection platform, and a high-angle annular light source disposed above the low-angle annular light source, and an area array camera with a full-frame lens is also disposed above the high-angle annular light source; The third inspection station includes an inspection platform, an arc light source disposed above the inspection platform, and a linear array camera disposed above the arc light source, wherein the linear array camera has a telecentric lens; The fourth inspection station includes an inspection platform, an arc light source arranged above the inspection platform, and a linear array camera arranged above the arc light source, wherein the linear array camera has a telecentric lens, and a prism is further arranged between the inspection platform and the arc light source, wherein the prism is used to be placed inside the inner ring / outer ring of the bearing to be inspected; It also includes an image storage device for receiving images collected by the area array camera and the line array camera, and an image analysis device for processing the above images according to a set algorithm.

2. The modular bearing defect detection device based on machine vision according to claim 1, It is characterized in that The third inspection station also includes an encoder for triggering the linear array camera.

3. A modular detection method for bearing defects based on machine vision. It is characterized in that The bearing defect modular detection device based on machine vision as claimed in any one of claims 1 or 2 is used, comprising the following steps: Step 1: placing the bearing workpiece to be inspected on the inspection table of the first inspection station, collecting images of the first end face and the first inner chamfered surface thereof and transmitting them to the image storage device; Step 2: flip the bearing workpiece and transport it to the inspection table of the second inspection station, collect images of its second end face and second inner chamfered surface and transmit them to the image storage device; Step 3: transport the bearing workpiece to the inspection table of the third inspection station, adjust the arc light source so that it illuminates the outer cylindrical surface and the outer chamfered surface of the bearing workpiece at the same time, and while rotating the bearing workpiece, collect a complete image of the outer cylindrical surface and the outer chamfered surface of the bearing workpiece and transmit them to the image storage device; Step 4: transport the bearing workpiece to the inspection table of the fourth inspection station, place the prism inside the bearing workpiece, and collect an image of the inner circumference of the bearing workpiece while rotating the bearing workpiece and transmit it to the image storage device; Step 5: The image analyzing device analyzes whether the inspected bearing workpiece is qualified according to the image in the image storage device and a preset algorithm.

4. The bearing defect modular detection method based on machine vision according to claim 3, It is characterized in that In the step three, a trigger is used to control the photographing frequency of the linear array camera.

5. The bearing defect modular detection method based on machine vision according to claim 3, It is characterized in that In the step three, the line frequency of the camera is determined according to the diameter of the bearing to ensure the integrity of the image of the outer surface of the bearing collected.

6. The bearing defect modular detection method based on machine vision according to claim 3, It is characterized in that In the step 5, any one of Faster RCNN, SSD, YOLO vn algorithm models or improved versions thereof is preset in the image analysis device.