Bearing inner and outer ferrule raceway machining defect detection device and method

Abstract

The invention discloses a bearing inner ring raceway machining defect detection device and method.The bearing inner ring raceway machining defect detection device comprises an image acquisition system and a positioning and supporting system, the positioning and supporting system comprises a supporting frame and a camera fixing support arranged on the supporting frame, and the camera fixing support is used for installing an industrial camera and a monocular microscope; the lower end of the camera fixing support is provided with an object placing disc used for placing a bearing to be detected, the object placing disc is further provided with an optical prism, the optical prism is arranged in the bearing, and the optical prism can reflect an image of an inner raceway of a bearing outer ferrule to the monocular microscope and the industrial camera. The single-tube microscope and the industrial camera are used for shooting the optical prism, full-raceway images of the inner raceway of the bearing outer ferrule are collected, the structure is simple, automatic collection of micro machining defect images of the full-raceway of the bearing inner ferrule and the bearing outer ferrule can be achieved, the single-tube microscope is used in cooperation, and the machining precision of the bearing inner ferrule and the bearing outer ferrule is improved. The problem that small targets are difficult to recognize in the defect detection process is solved.

Description

Technical Field

[0001] The invention belongs to the field of bearing defect detection, and in particular relates to a device and method for detecting machining defects of bearing inner and outer ring raceways. Background Art

[0002] Bearings are basic components of modern industry. Their main function is to support mechanical rotating bodies, fix the rotating shaft and reduce load friction during mechanical operation. The quality of bearings is directly related to the life and reliability of mechanical rotating bodies. Bearing failure will lead to the paralysis of the entire mechanical system. Therefore, bearing defect detection is a key and necessary step in its manufacturing process. How to efficiently realize bearing defect detection is a major problem that the bearing processing industry needs to solve urgently.

[0003] At present, in various industries, most mechanical systems use rolling bearings. The main working surface of rolling bearings is the raceway of the bearing ring. The results of tests and daily applications also show that the main wear part of rolling bearings in actual use is the raceway surface of the bearing ring, and its surface quality has a significant impact on the performance and life of the bearing. Therefore, the bearing processing industry focuses on the surface quality detection of the bearing ring raceway for bearing defect detection. With the improvement of production technology, most of the bearings currently produced by enterprises are high-precision bearings, and the requirements for their surface quality are more stringent. Since the surface of the bearing raceway of high-precision bearings after fine processing is a mirror surface, and the bearing raceway is a curved surface, the surface defects of such bearings are basically invisible to conventional methods such as human eye observation, and they need to be detected with the help of high-precision instruments such as microscopes. In the industry, the existing detection method is manual visual inspection by human eyes with the help of microscopes. Due to the individual differences of the detection personnel, this method is prone to missed detection and false detection. At the same time, the detection speed is slow and the efficiency is low, which greatly affects the overall process of product delivery. Summary of the invention

[0004] The purpose of the present invention is to provide a device and method for detecting machining defects of bearing inner and outer ring raceways, so as to overcome the problems of the prior art that missed detection and false detection are prone to occur, and the detection speed is slow and the efficiency is low.

[0005] To achieve the above purpose, the technical solution adopted by the present invention is as follows: A bearing inner ring raceway machining defect detection device comprises an image acquisition system and a positioning support system, the positioning support system comprising a support frame and a camera fixing support arranged on the support frame; the image acquisition system comprises an industrial camera and a monocular microscope, the camera fixing support is used to install the industrial camera (1) and the monocular microscope, a placement disk for placing a bearing to be detected is arranged at the lower end of the camera fixing support, and also comprises a coaxial light source arranged between the placement disk and the monocular microscope, an optical prism is also arranged on the placement disk, the optical prism is arranged in the bearing, and the optical prism can reflect the image of the inner raceway of the bearing outer ring to the monocular microscope and the industrial camera.

[0006] Preferably, a bottom plate support is provided at the bottom of the support frame, and the bottom plate support serves as the supporting body of the support frame and is also used for placing the storage disc.

[0007] Preferably, a rotating platform is provided on the bottom plate support, and the storage disc is rotatably arranged on the rotating platform.

[0008] Preferably, the support frame adopts a column structure, the camera fixing bracket is fixed to the support frame through a camera transverse bracket, one end of the camera transverse bracket is fixed to the support frame through a first fixing block, the other end of the camera transverse bracket is provided with a guide rail pillar, the guide rail pillar is provided with a guide rail, and the camera fixing bracket is installed on the guide rail.

[0009] Preferably, the guide rail is fixed to the guide rail support by screws; the camera fixing bracket is slidably connected to the guide rail and fixed to the guide rail support by a locking nut.

[0010] Preferably, a light source bracket is installed on the support frame, the coaxial light source is arranged on the light source bracket, and the coaxial light source is fixed to the light source bracket by bolts and nuts. A light source bracket upright pole is arranged on the light source bracket, the light source bracket is fixed to the light source bracket upright pole by bolts and nuts, and the light source bracket upright pole is fixed to the support frame by a second fixing block.

[0011] Preferably, a bearing centering groove is provided on the placement disc, the outer ring of the bearing to be tested is placed on the bearing centering groove, and the optical prism is located in the middle of the outer ring of the bearing.

[0012] A device for detecting machining defects of a bearing outer ring raceway comprises an image acquisition system and a positioning support system, wherein the positioning support system comprises a camera vertical adjustment device and a camera bracket arranged on the camera vertical adjustment device; the image acquisition system comprises an industrial camera and a monocular microscope, the camera bracket is used to install the industrial camera and the monocular microscope, a placement tray for placing a bearing to be inspected is arranged on one side of the camera bracket, and the camera bracket also comprises a coaxial light source arranged between the placement tray and the monocular microscope, and a bearing centering boss is arranged on the placement tray.

[0013] Preferably, the camera vertical adjustment device is slidably connected to a camera horizontal adjustment device, a base plate is provided at the bottom of the camera horizontal adjustment device, a bearing centering boss is provided on the placement tray, and the inner ring of the bearing to be tested is placed on the bearing centering boss of the placement tray.

[0014] A method for detecting machining defects of bearing inner and outer ring raceways comprises the following steps: S1, using an industrial camera to take pictures of the bearing raceway at different positions to collect the bearing raceway image data; S2, screening out bearing raceway images with defects from the collected bearing raceway images, and performing image processing on the screened images; S3, performs binary classification annotation on the image dataset after image processing in S2; S4, training the improved YOLOv8 network model based on the binary classification annotations obtained above; S5, substitute the trained weights into the model, input the image data to be detected, the detection network processes and recognizes the features of the image, and then automatically generates a rectangular frame for the defect position contained in the original image data, thus completing the entire detection.

[0015] Compared with the prior art, the present invention has the following beneficial technical effects: The present invention discloses a device for detecting machining defects of a bearing inner ring raceway, comprising an image acquisition system and a positioning support system, wherein the positioning support system comprises a support frame and a camera fixing support arranged on the support frame, the camera fixing support being used to install an industrial camera and a monocular microscope, a placement disk for placing a bearing to be detected is arranged at the lower end of the camera fixing support, and the device further comprises a coaxial light source arranged between the placement disk and the monocular microscope, an optical prism is also arranged on the placement disk, the optical prism is arranged in the bearing, the optical prism can reflect the image of the inner raceway of the bearing outer ring to the monocular microscope and the industrial camera, during detection, the monocular microscope and the industrial camera are used to capture the optical prism to realize the acquisition of the entire raceway image of the inner raceway of the bearing outer ring, the present application has a simple structure, can realize the automatic acquisition of tiny machining defect images of the entire raceway of the bearing inner and outer rings, and adopts the monocular microscope in combination to solve the problem that small targets are difficult to identify during defect detection.

[0016] A method for detecting machining defects of bearing inner and outer ring raceways uses an industrial camera and a monocular microscope combined with a suitable light source solution to collect images of tiny defects in the bearing raceways, and then detects and identifies defects in the collected bearing images based on a deep learning method, which has the advantages of fast detection speed and high degree of automation. This application will improve the current status of product quality inspection in the domestic bearing industry and form a new solution for automated detection of tiny machining defects in the inner and outer ring raceways of bearings. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] Figure 1 Schematic diagram of the structure of a bearing outer ring inner raceway defect detection device in an embodiment of the present invention.

[0018] Figure 2 Schematic diagram of the structure of a bearing inner ring outer raceway defect detection device in an embodiment of the present invention.

[0019] Figure 3 It is a simplified diagram of the optical path propagation circuit of the bearing inner and outer ring raceway defect detection device in an embodiment of the present invention.

[0020] Figure 4 This is a diagram of the YOLOv8 network structure in an embodiment of the present invention.

[0021] Figure 5 Schematic diagram of the structure of each module of the YOLOv8 network in an embodiment of the present invention.

[0022] Figure 6 Schematic diagram of the feature fusion method of the four-detection head PANet network and the BiFPN network in an embodiment of the present invention.

[0023] Figure 7 The figure shows the visual detection results of bearing raceway defect detection in an embodiment of the present invention, where (a) and (c) are images of the inner raceway of the outer ring of the bearing, (b) and (d) are image detection images of the inner raceway of the outer ring of the bearing; (e) and (g) are images of the outer raceway of the inner ring of the bearing, and (f) and (h) are image detection images of the outer raceway of the inner ring of the bearing.

[0024] Among them, 1. industrial camera, 2. monocular microscope, 3. camera fixing bracket, 4. guide rail, 5. camera horizontal bracket, 6. coaxial light source, 7. light source bracket, 8. optical prism, 9. bearing centering groove, 10. storage disc, 11. bottom plate support, 12. rotating platform, 13. bearing outer ring, 14. light source bracket upright, 15. second fixed block, 16. first fixed block, 17. support frame, 20. camera bracket, 22. storage disc, 23. bearing inner ring, 24. bearing storage boss, 25. hollow rotating platform, 26. light source storage table, 27. bottom plate, 28. camera vertical adjustment device, 29. camera horizontal adjustment device. DETAILED DESCRIPTION

[0025] In order to enable those skilled in the art to better understand the scheme of the present invention, the technical scheme 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 only 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 should fall within the scope of protection of the present invention.

[0026] It should be noted that the terms "first", "second", etc. in the specification and claims of the present invention and the above-mentioned drawings are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequence. It should be understood that the data used in this way can be interchanged where appropriate, so that the embodiments of the present invention described herein can be implemented in an order other than those illustrated or described herein. In addition, the terms "including" and "having" and any variations thereof are intended to cover non-exclusive inclusions, for example, a process, method, system, product or device that includes a series of steps or units is not necessarily limited to those steps or units that are clearly listed, but may include other steps or units that are not clearly listed or inherent to these processes, methods, products or devices.

[0027] like Figure 1 As shown, the present invention proposes a bearing inner ring raceway machining defect detection device, including an image acquisition system and a positioning support system, the positioning support system includes a support frame 17 and a camera fixing bracket 3 arranged on the support frame, the camera fixing bracket 3 is used to install an industrial camera 1 and a monocular microscope 2, a placement disk 10 for placing a bearing to be detected is arranged at the lower end of the camera fixing bracket 3, and also includes a coaxial light source 6 arranged between the placement disk 10 and the monocular microscope 2, an optical prism 8 is also arranged on the placement disk 10, the optical prism 8 is arranged in the bearing, the optical prism 8 can reflect the image of the inner raceway of the bearing outer ring to the monocular microscope 2 and the industrial camera 1, during detection, the monocular microscope 2 and the industrial camera 1 are used to capture the full raceway image of the inner raceway of the bearing outer ring by shooting the optical prism 8, the present application has a simple structure, can realize the automatic acquisition of tiny machining defect images of the full raceway of the inner and outer rings of the bearing, and adopts the monocular microscope 2 to solve the problem of small targets being difficult to identify during defect detection.

[0028] In a specific implementation of the present application, a bottom plate support 11 is provided at the bottom of the support frame 17 , and the bottom plate support 11 serves as a supporting body of the support frame 17 and is also used for placing the storage disc 10 .

[0029] A rotating platform 12 is provided on the base support 11, and the storage disc 10 can be rotatably provided on the rotating platform 12. Specifically, a driving motor can be provided in the rotating platform 12, and the output shaft of the driving motor is rotatably connected to the storage disc 10. The driving motor can be used to automatically control the rotation of the storage disc 10, thereby realizing the collection of the full raceway image of the inner raceway of the bearing outer ring.

[0030] In a specific embodiment of the present application, the support frame 17 adopts a column structure, the camera fixing bracket 3 is fixed to the support frame 17 through the camera transverse bracket 5, one end of the camera transverse bracket 5 is fixed to the support frame 17 through the first fixing block 16, and the other end of the camera transverse bracket 5 is provided with a guide rail pillar, and the guide rail pillar is provided with a guide rail 4, and the camera fixing bracket 3 is installed on the guide rail 4; the camera transverse bracket 5 is used in conjunction with the support frame 17 to adjust the industrial camera 1 and the monocular microscope 2 on the camera fixing bracket 3, to achieve a wide range of adjustment, to facilitate the removal and placement of the bearings, and to quickly adjust the positioning.

[0031] In the specific implementation of the present application, the guide rail 4 is fixed to the guide rail pillar by screws. The camera fixing bracket 3 is slidably connected to the guide rail 4 and fixed to the guide rail pillar by a locking nut. The guide rail pillar 5 is fixed to the first fixing block 16 by a locking nut, and the first fixing block 16 is fixed to the support frame 17 by a locking nut.

[0032] In a specific embodiment of the present application, a light source bracket 7 is installed on the support frame 17, and a coaxial light source 6 is arranged on the light source bracket 7, and the coaxial light source 6 is fixed to the light source bracket 7 by bolts and nuts. A light source bracket upright pole 14 is arranged on the light source bracket 7, and the light source bracket 7 is fixed to the light source bracket upright pole 14 by bolts and nuts. The light source bracket upright pole 14 is fixed to the support frame 17 by a second fixing block 15, and the second fixing block 15 is fixed to the support frame 17 by a locking nut.

[0033] The placement disc 10 is provided with a bearing centering groove 9, and the bearing outer ring 13 to be inspected is placed on the bearing centering groove 9. The optical prism 8 is located in the middle of the bearing outer ring 13. During inspection, a motor controller is used to control the motor to drive the bearing outer ring 12 to rotate, and the monocular microscope 2 and the industrial camera 1 capture the full raceway image of the inner raceway of the bearing outer ring by shooting the optical prism 8.

[0034] like Figure 2As shown, a bearing outer ring raceway machining defect detection device proposed by the present invention can detect the outer ring of the bearing outer ring 13, the industrial camera 1 and the monocular microscope 2 are connected by threads, the industrial camera 1 and the monocular microscope 2 are fixed on the camera bracket 20 by a locking nut, the camera bracket 20 is fixed by a camera vertical adjustment device 28, the camera bracket 20 is slidably connected with the camera vertical adjustment device 28, the camera vertical adjustment device 28 is horizontally slidably connected with the camera horizontal adjustment device 29, the camera horizontal adjustment device 29 is fixed on the bottom plate 27, the coaxial light source 6 is placed on the light source placement table 26, the light source placement table 26 is fixed on the bottom plate 29, a bearing centering boss 24 is provided on the placement plate 22, the bearing inner ring 23 to be detected is placed on the bearing centering boss 24 of the placement plate 22, the bearing centering boss 24 is placed on the placement plate 22, and the placement plate 22 is bolted to the hollow rotating platform 25 and then placed on the bottom plate 27. During automated testing, a motor controller is used to control a stepper motor to drive the bearing inner ring 33 to rotate, and an industrial camera captures a full raceway image of the bearing inner ring outer raceway by photographing the bearing inner ring outer raceway.

[0035] like Figure 3 As shown, the present invention proposes a bearing inner and outer ring raceway processing defect detection device, and the detection process is as follows: the bearing outer ring inner raceway system, the LED lamp group of the coaxial light source emits light, and the light is reflected to the optical prism through the coaxial light source. After the light is reflected twice inside the optical prism, it is irradiated to the bearing inner raceway, and then the raceway reflects the light through the right-angle prism along the original path, and enters the lens through the beam splitter; the bearing inner ring outer raceway system, the LED lamp group of the coaxial light source emits light, and the light is reflected to the bearing outer raceway through the coaxial light source. Then the raceway reflects the light and enters the lens through the beam splitter.

[0036] In another embodiment of the present invention, a method for detecting machining defects of bearing inner and outer ring raceways is provided, comprising the following steps: S1, image acquisition, turn on the industrial camera and coaxial light source, start the stepper motor to make the bearing ring rotate evenly, use the industrial camera to take pictures of the bearing raceway at different positions, and collect the bearing raceway image data; S2, data preprocessing: from the collected bearing raceway images, the bearing raceway images with defects are screened out, and the screened images are processed, including horizontal up and down flipping, scaling, rotation, adding noise, and filtering operations, so as to expand the data set and improve the robustness of the subsequent training model; S3, data annotation: Use Labelimg tool to perform binary classification annotation on the image dataset expanded from S2; S4, model training: training is performed based on the improved YOLOv8 network model, the image input size is 1280*1280 pixels, the initial learning rate is 0.001, and the optimizer selects the Adam optimizer; S5, defect detection: Substitute the trained weights into the model, input the image data to be detected, the detection network performs feature processing and recognition on the image, and then automatically generates a rectangular frame for the defect position contained in the original image data, thus completing the entire detection. After the detection, the detected image data will be automatically saved to facilitate further analysis of the bearing raceway defects.

[0037] In the present invention, in step S1, the defects of the bearing ring are defects of different types and sizes generated in the actual production process, and the bearing raceway defect detection device is used to collect defect images of the bearing ring raceway. Different image acquisition methods are used for the inner and outer raceway of the bearing, wherein the inner raceway image acquisition of the outer raceway of the bearing: the camera indirectly acquires the inner raceway image of the bearing by shooting the inner raceway image reflected by the optical prism; and the outer raceway image acquisition of the inner raceway of the bearing: the camera directly captures the outer raceway image of the bearing to realize the acquisition of the outer raceway image of the bearing.

[0038] In the present invention, in step S2, more than 200 defective photos taken from defective bearings provided by the enterprise are collected. After data set expansion, the data set reaches more than 1,800 photos, which are randomly divided into training set and test set, with a ratio of 8:2.

[0039] In the present invention, in step S3, the data labeling method is to create a rectangular box at the defect position of the original image, fill in the label as defect, and perform binary classification labeling. The labeling tool will store the labeled type code and the position information of the rectangular box in the image. First, the original image taken in step S1 is labeled using the labeling tool; then, for the data set expanded in step S2, the algorithm is used to automatically label it in combination with the labeling result of the original image; finally, the automatically labeled data set is manually checked, and the incorrectly labeled data set is manually corrected.

[0040] In the present invention, in the step S4, the data set used for training is the data set constructed in the steps S2 and S3. The improved YOLOv8 network model adds a P2 detection head on the basis of the original YOLOv8 model to prevent the loss of targets with tiny scale defects during the detection process. The initial weight of YOLOv8 is selected as the yolov8s weight. The YOLOv8 network includes three parts: Backbone, Neck and Head. Among them, Backbone is the backbone network, and its function is to extract features from the input image; Head is the head network, and its function is to generate detection results; Neck is the neck network, which is located between the backbone network and the head network, and its function is to perform feature fusion and feature enhancement on the image features extracted by the Backbone module, so as to facilitate the Head module to perform target recognition. Figure 4 As shown in the figure, the improved YOLOv8 network mainly adds a small-scale detection head P2 in the Head module to increase the scale range of detection. At the same time, the Neck module uses the feature fusion method of the bidirectional feature pyramid network (BiFPN) to replace the feature fusion method of the original PANet network, as shown in the figure. Figure 6 As shown in Figure 2, multi-scale feature fusion is achieved.

[0041] In the present invention, in step S5, the detection algorithm will be connected to the industrial camera to realize real-time detection of the bearing raceway. New defects detected during the detection process will be added to the data set of step S2 to continuously enrich the database. Figure 7 As shown, (a) and (c) are images of the inner raceway of the outer ring of the bearing, (b) and (d) are image detection images of the inner raceway of the outer ring of the bearing; (e) and (g) are images of the outer raceway of the inner ring of the bearing, (f) and (h) are image detection images of the outer raceway of the inner ring of the bearing. The detection algorithm will mark the defect position of the captured image and indicate the detection confidence of the defect.

[0042] Working principle: First, place the outer ring 13 or inner ring 23 of the bearing to be tested on the bearing placement groove 9 or the bearing placement boss 24; then use the adjustment device to adjust the distance between the industrial camera 1, the coaxial light source 6 and the outer ring 13 or the inner ring 23 of the bearing, and use the motor controller to set the rotation speed of the hollow rotating platform 12 or the hollow rotating platform 25; then start the stepper motor, turn on the industrial camera 1 and the coaxial light source 6, and use the pre-trained detection network on the computer - the real-time detection function of the improved YOLOv8 network, and the computer will perform defect detection on the bearing ring raceway obtained by the industrial camera system; finally, operate the computer to stop the detection function, and the computer will save the defect detection results. After turning off the power of each device, the automated defect detection of the bearing ring raceway is completed.

Claims

1. A bearing inner ring raceway machining defect detection device, characterized in that: The invention comprises an image acquisition system and a positioning support system, wherein the positioning support system comprises a support frame (17) and a camera fixing frame (3) arranged on the support frame; the image acquisition system comprises an industrial camera (1) and a monocular microscope (2); the camera fixing frame (3) is used to install the industrial camera (1) and the monocular microscope (2); a placement disk (10) for placing a bearing to be inspected is arranged at the lower end of the camera fixing frame (3); and the camera fixing frame (3) further comprises a coaxial light source (6) arranged between the placement disk (10) and the monocular microscope (2); an optical prism (8) is also arranged on the placement disk (10); the optical prism (8) is arranged in the bearing; and the optical prism (8) can reflect an image of a raceway in an outer ring of the bearing to the monocular microscope (2) and the industrial camera (1).

2. A bearing inner ring raceway machining defect detection device according to claim 1, characterized in that: A bottom plate support (11) is provided at the bottom of the support frame (17), and the bottom plate support (11) serves as a supporting body of the support frame (17) and is also used to place the storage disc (10).

3. A bearing inner ring raceway machining defect detection device according to claim 2, characterized in that: A rotating platform (12) is arranged on the bottom plate support (11), and the object placement disc (10) is rotatably arranged on the rotating platform (12).

4. A bearing inner ring raceway machining defect detection device according to claim 1, characterized in that: The support frame (17) adopts a column structure, the camera fixing bracket (3) is fixed to the support frame (17) through the camera transverse bracket (5), one end of the camera transverse bracket (5) is fixed to the support frame (17) through a first fixing block (16), the other end of the camera transverse bracket (5) is provided with a guide rail pillar, the guide rail pillar is provided with a guide rail (4), and the camera fixing bracket (3) is installed on the guide rail (4).

5. A bearing inner ring raceway machining defect detection device according to claim 1, characterized in that: The guide rail (4) is fixed to the guide rail support by screws; the camera fixing bracket (3) is slidably connected to the guide rail (4) and is fixed to the guide rail support by a locking nut.

6. A bearing inner ring raceway machining defect detection device according to claim 1, characterized in that: A light source bracket (7) is mounted on the support frame (17); a coaxial light source (6) is arranged on the light source bracket (7); the coaxial light source (6) is fixed to the light source bracket (7) by means of bolts and nuts; a light source bracket upright pole (14) is arranged on the light source bracket (7); the light source bracket (7) is fixed to the light source bracket upright pole (14) by means of bolts and nuts; the light source bracket upright pole (14) is fixed to the support frame (17) by means of a second fixing block (15).

7. A bearing inner ring raceway machining defect detection device according to claim 1, characterized in that: The placement disc (10) is provided with a bearing centering groove (9), the outer bearing ring (13) to be tested is placed on the bearing centering groove (9), and the optical prism (8) is located in the middle of the outer bearing ring (13).

8. A bearing outer ring raceway machining defect detection device, characterized in that: The invention comprises an image acquisition system and a positioning support system, wherein the positioning support system comprises a camera vertical adjustment device (28) and a camera support (20) arranged on the camera vertical adjustment device (28); the image acquisition system comprises an industrial camera (1) and a monocular microscope (2); the camera support (20) is used to install the industrial camera (1) and the monocular microscope (2); a placement tray (22) for placing a bearing to be tested is arranged on one side of the camera support (20); the camera support (20) also comprises a coaxial light source (6) arranged between the placement tray (22) and the monocular microscope (2); and a bearing centering boss (24) is arranged on the placement tray (22).

9. A bearing outer ring raceway machining defect detection device according to claim 8, characterized in that: The camera vertical adjustment device (28) is slidably connected to a camera horizontal adjustment device (29); a bottom plate (27) is disposed at the bottom of the camera horizontal adjustment device (29); a bearing centering boss (24) is disposed on the storage tray (22); and the inner ring (23) of the bearing to be tested is placed on the bearing centering boss (24) of the storage tray (22).

10. A method for detecting machining defects of inner and outer race raceways of bearings, comprising the following steps: S1, using an industrial camera to take pictures of the bearing raceway at different positions to collect the bearing raceway image data; S2, screening out bearing raceway images with defects from the collected bearing raceway images, and performing image processing on the screened images; S3, performs binary classification annotation on the image dataset after image processing in S2; S4, training the improved YOLOv8 network model based on the binary classification annotations obtained above; S5, substitute the trained weights into the model, input the image data to be detected, the detection network processes and recognizes the features of the image, and then automatically generates a rectangular frame for the defect position contained in the original image data, thus completing the entire detection.

Images