Bearing Defect Identification Method, System, Electronic Device and Readable Storage Medium

Abstract

The present invention provides a bearing defect identification method, system, electronic device and readable storage medium, relating to the technical field of bearing defect detection. The bearing defect identification method includes: obtaining an X-ray image containing the internal structure of the bearing to be measured; performing binary segmentation processing on the X-ray image to generate a binary image; converting the binary image into a polar coordinate image, where the origin of the polar coordinate system in the polar coordinate image is the center of the bearing to be measured; performing horizontal integration on the polar coordinate image to generate a line integral signal presented in a first curve graph; based on the first curve graph, performing peak detection on the line integral signal, and when it is detected that the peak value is less than a preset threshold, determining that the bearing to be measured has a defect. In the technical solution of the present invention, defect identification of the bearing to be measured is performed based on the processing of the X-ray image. Compared with the identification method of deep learning, it does not require a large amount of image data as training samples, which is beneficial to reducing costs, improving detection efficiency and the accuracy of detection results.

Description

Technical Field

[0001] The present invention relates to the technical field of bearing defect detection, and in particular, to a bearing defect recognition method, system, electronic device and readable storage medium. Background Art

[0002] Some types of bearings, such as roller bearings, are usually used in heavy-load and low-speed scenarios. The absence of rollers will cause serious vibration and even failure, so high detection accuracy is required. In the related art, when detecting bearing defects, the detection efficiency is low and the accuracy of the detection result is poor.

[0003] How to design a bearing defect recognition method to improve the detection efficiency and the accuracy of the detection result is an urgent problem to be solved at present. Summary of the Invention

[0004] To solve or improve the technical problems of low detection efficiency and poor accuracy of the detection result, an object of the present invention is to provide a bearing defect recognition method.

[0005] Another object of the present invention is to provide a bearing defect recognition system.

[0006] Another object of the present invention is to provide an electronic device.

[0007] Another object of the present invention is to provide a readable storage medium.

[0008] To achieve the above object, a first aspect of the present invention provides a bearing defect recognition method, including: acquiring an X-ray image including the internal structure of the bearing to be tested; performing binary segmentation processing on the X-ray image to generate a binary image; converting the binary image into a polar coordinate image, where the origin of the polar coordinate system in the polar coordinate image is the center of the bearing to be tested; performing horizontal integration on the polar coordinate image to generate a line integral signal presented in a first curve graph; based on the first curve graph, performing peak detection on the line integral signal, and when the detected peak value is less than a preset threshold, determining that the bearing to be tested has a defect.

[0009] The present invention aims to provide a bearing defect recognition method, which performs defect recognition on the bearing to be tested based on the processing of the X-ray image. Compared with the deep learning-based recognition method, it does not require a large amount of image data as training samples, and can improve the detection efficiency and the accuracy of the detection result while reducing costs.

[0010] During the entire defect detection process, no manual intervention is required, only simple peak comparison logic is needed, and no deep learning model or complex pattern matching is required, which is beneficial to reducing the deployment cost and meeting the real-time detection requirements of the production line. This design method has a high degree of automation and is efficient.

[0011] In addition, the above technical solutions provided by the present invention may further have the following additional technical features:

[0012] In some technical solutions, optionally, based on the first curve graph, perform peak detection on the line integral signal. When the detected peak value is less than a preset threshold, it is determined that the bearing to be tested has a defect, including: moving the data within the first range at the head of the abscissa of the first curve graph to the tail of the abscissa to achieve signal alignment, obtaining an optimized curve graph; determining the preset threshold; determining the peak positions in the optimized curve graph and performing peak detection, and when the detected peak value is less than the preset threshold, it is determined that the bearing to be tested has a defect.

[0013] In this technical solution, by moving the data within the first range at the head of the abscissa of the first curve graph to the tail of the abscissa, the starting position of the signal (line integral signal) can be adjusted, enabling signals under different imaging conditions to be compared on the same basis and eliminating the starting angle difference caused by different placement angles of the bearing to be tested. After aligning the signals, the positions of the peaks corresponding to the rolling elements (rollers or balls) are relatively fixed on the optimized curve graph, which can make the subsequent peak detection more accurate and stable, thereby improving the accuracy of defect judgment.

[0014] The purpose of peak detection is to identify the positions of the rolling elements (rollers or balls). Under normal circumstances, each rolling element will correspond to a peak. If the peak value is less than the preset threshold, it indicates that there may be defects or missing rollers or balls.

[0015] In some technical solutions, optionally, the data within the first range is the first 10% of the data at the head of the abscissa.

[0016] In this technical solution, the data from 0 to 10 in the abscissa (the first 10% of the data at the head of the abscissa) is moved to the tail of the abscissa to achieve signal alignment. After aligning the signals, the positions of the peaks corresponding to the rolling elements (rollers or balls) are relatively fixed on the optimized curve graph, which can make the subsequent peak detection more accurate and stable, thereby improving the accuracy of defect judgment.

[0017] In some technical solutions, optionally, determining the peak positions in the optimized curve graph and performing peak detection includes: determining the peak positions in the optimized curve graph based on the local maximum detection algorithm or the derivative method; determining whether the peak value corresponding to the peak position is less than the preset threshold.

[0018] In this technical solution, by adopting the local maximum detection algorithm or the derivative method, it is beneficial to accurately judge the peak positions, thereby improving the accuracy of the detection results.

[0019] The purpose of peak detection is to identify the positions of the rolling elements (rollers or balls). Under normal circumstances, each rolling element corresponds to a peak. If the peak value is less than the preset threshold, it indicates that there may be defects or missing rollers or balls. The purpose of this step is to convert the line integral signal into a defect judgment.

[0020] In some technical solutions, optionally, determining the preset threshold includes: determining the preset threshold based on the statistical average of the peak values of the line integral signals of defect-free bearings.

[0021] In this technical solution, by determining the preset threshold according to the statistical average, it is beneficial to improve the accuracy of peak detection and effectively reduce the possibility of misjudgment.

[0022] In some technical solutions, optionally, determining the preset threshold includes: determining the preset threshold based on the minimum peak value of the line integral signals of defect-free bearings.

[0023] In this technical solution, by determining the preset threshold according to the minimum peak value, it is beneficial to improve the accuracy of peak detection and effectively reduce the possibility of misjudgment.

[0024] In some technical solutions, optionally, performing a horizontal integration on the polar coordinate image to generate a line integral signal presented as a first curve graph includes: accumulating and summing the pixel values of each row of the polar coordinate image to achieve horizontal integration and generate a line integral signal presented as a first curve graph.

[0025] In this technical solution, the integration process can highlight the periodic characteristics and reduce the influence of noise. Because integration is equivalent to summing the pixels of each row, it may smooth out some random noise. At the same time, reducing the two-dimensional data to one-dimensional simplifies the subsequent signal processing steps, and the peak detection algorithm will be more efficient.

[0026] The second aspect of the present invention provides a bearing defect recognition system, including: an image acquisition module for acquiring an X-ray image containing the internal structure of the bearing to be tested; an image processing module for performing binary segmentation processing on the X-ray image to generate a binary image; an image conversion module for converting the binary image into a polar coordinate image, where the origin of the polar coordinate system in the polar coordinate image is the center of the bearing to be tested; a signal generation module for performing horizontal integration on the polar coordinate image to generate a line integral signal presented as a first curve graph; a defect recognition module for performing peak detection on the line integral signal based on the first curve graph, and determining that the bearing to be tested has a defect when the detected peak value is less than the preset threshold.

[0027] The present invention aims to provide a bearing defect recognition system, which performs defect recognition on a bearing to be measured based on the processing of X-ray images. Compared with the recognition method based on deep learning, it does not require a large amount of image data as training samples, and can improve the detection efficiency and the accuracy of detection results while reducing costs.

[0028] During the entire defect detection process, no manual intervention is required, only simple peak comparison logic is needed, and no deep learning model or complex pattern matching is required, which is beneficial to reducing the deployment cost and meeting the real-time detection requirements of the production line. This design method has a high degree of automation and is efficient.

[0029] The third aspect of the present invention provides an electronic device, including: a memory and a processor. Among them, a program or instruction that can run on the processor is stored on the memory, and when the processor executes the program or instruction, the steps of the bearing defect recognition method in any of the above technical solutions are implemented. The electronic device has the beneficial effects of any of the above technical solutions, which will not be elaborated here.

[0030] The fourth aspect of the present invention provides a readable storage medium, which stores a program or instruction, and when the program or instruction is executed by the processor, the steps of the bearing defect recognition method in any of the above technical solutions are implemented. The readable storage medium has the beneficial effects of any of the above technical solutions, which will not be elaborated here.

[0031] The additional aspects and advantages of the technical solutions of the present invention will become obvious in the following description section, or be understood through the practice of the present invention. BRIEF DESCRIPTION OF THE DRAWINGS

[0032] Figure 1 Shows a flowchart of a bearing defect recognition method according to an embodiment of the present invention;

[0033] Figure 2 Shows a flowchart of a bearing defect recognition method according to another embodiment of the present invention;

[0034] Figure 3 Shows a flowchart of a bearing defect recognition method according to another embodiment of the present invention;

[0035] Figure 4 Shows a flowchart of a bearing defect recognition method according to another embodiment of the present invention;

[0036] Figure 5 Shows a flowchart of a bearing defect recognition method according to another embodiment of the present invention;

[0037] Figure 6 Shows a flowchart of a bearing defect recognition method according to another embodiment of the present invention;

[0038] Figure 7 The flowchart of a bearing defect identification method according to another embodiment of the present invention is shown;

[0039] Figure 8 The flowchart of a bearing defect identification method according to another embodiment of the present invention is shown;

[0040] Figure 9 The flowchart of a bearing defect identification method according to another embodiment of the present invention is shown;

[0041] Figure 10 The structural block diagram of a bearing defect identification system according to an embodiment of the present invention is shown;

[0042] Figure 11 The structural block diagram of an electronic device according to an embodiment of the present invention is shown;

[0043] Figure 12 The schematic diagram of an X-ray image according to an embodiment of the present invention is shown;

[0044] Figure 13 The schematic diagram of a binary image according to an embodiment of the present invention is shown;

[0045] Figure 14 The schematic diagram of a polar coordinate image according to an embodiment of the present invention is shown;

[0046] Figure 15 The schematic diagram of a first curve graph according to an embodiment of the present invention is shown;

[0047] Figure 16 The schematic diagram of an optimized curve graph according to an embodiment of the present invention is shown;

[0048] Figure 17 The schematic diagram of peak detection according to an embodiment of the present invention is shown.

[0049] Among them, Figure 10 and Figure 11 The corresponding relationship between the reference numerals and the component names in the figure is as follows:

[0050] 200: Bearing defect identification system; 210: Image acquisition module; 220: Image processing module; 230: Image conversion module; 240: Signal generation module; 250: Defect identification module; 300: Electronic device; 310: Memory; 320: Processor. Detailed implementation manners

[0051] To better understand the above objects, features, and advantages of the embodiments of the present invention, the embodiments of the present invention will be further described in detail below in conjunction with the accompanying drawings and specific embodiments. It should be noted that, without conflict, the embodiments of the present invention and the features in the embodiments may be combined with each other.

[0052] In the following description, many specific details are set forth to facilitate a thorough understanding of the present invention. However, the embodiments of the present invention may be implemented in other ways different from those described herein. Therefore, the scope of protection of the present invention is not limited to the limitations of the specific embodiments disclosed below.

[0053] Next, refer to Figures 1 to 17 Describe a bearing defect identification method, system, electronic device, and readable storage medium provided according to some embodiments of the present invention.

[0054] In one embodiment of the present invention, as Figure 1 shown, the steps of the bearing defect identification method include:

[0055] S102, obtain an X-ray image including the internal structure of the bearing to be tested.

[0056] Image the bearing to be tested through an X-ray device to obtain an X-ray image including the internal structure of the bearing to be tested (as Figure 12 shown).

[0057] Optionally, the bearing to be tested includes an inner bearing ring, an outer bearing ring, and a plurality of rolling elements. The outer bearing ring is sleeved on the inner bearing ring, and the plurality of rolling elements are arranged between the inner bearing ring and the outer bearing ring. The outer bearing ring and the inner bearing ring can achieve relative rotation through the plurality of rolling elements.

[0058] The bearing to be tested in the present invention may be a roller bearing or a ball bearing. In the case where the bearing to be tested is a roller bearing, the plurality of rolling elements are a plurality of rollers, the rollers are cylindrical, and are evenly distributed along the circumferential direction between the inner bearing ring and the outer bearing ring. In the case where the bearing to be tested is a ball bearing, the plurality of rolling elements are a plurality of balls, the balls are spherical, and are evenly distributed along the circumferential direction between the inner bearing ring and the outer bearing ring.

[0059] X-rays have strong penetration ability and can penetrate the internal structure of the bearing to be tested (inner bearing ring, outer bearing ring, and rolling elements), thereby obtaining a two-dimensional projection image of the internal structure of the bearing to be tested. The two-dimensional projection image here is the X-ray image. Since the inner bearing ring, outer bearing ring, and rolling elements have different absorption rates (absorption capabilities) for X-rays, clear contrast can be formed in the X-ray image, facilitating the identification of defects in the bearing to be tested.

[0060] Taking the bearing to be tested as a roller bearing as an example: The roller bearing is imaged by an X-ray device to obtain an X-ray image containing the internal structure of the roller bearing. By forming a clear contrast in the X-ray image, it is convenient to identify structural defects such as missing rollers.

[0061] Taking the bearing to be tested as a ball bearing as an example: The ball bearing is imaged by an X-ray device to obtain an X-ray image containing the internal structure of the ball bearing. By forming a clear contrast in the X-ray image, it is convenient to identify structural defects such as missing balls.

[0062] Optionally, the X-ray device includes a transmitting part and a receiving part that can cooperate with each other during the working process. Among them, the transmitting part is used to emit X-rays; the receiving part is used to receive X-rays.

[0063] The imaging parameters of the X-ray device include one or a combination of the following: voltage parameter, current parameter, and integration time parameter. In the case where the imaging parameter is one, the imaging parameter is any one of the voltage parameter, current parameter, and integration time parameter; in the case where the imaging parameter is multiple, the imaging parameter is any combination of the voltage parameter, current parameter, and integration time parameter.

[0064] It should be noted that the voltage parameter determines the energy of the X-rays, affecting the penetration ability and image contrast. The current parameter determines the intensity of the X-rays, and thus affects the brightness and noise of the image. The integration time parameter can be understood as the time for the receiving part to collect data, and is used to determine the signal-to-noise ratio of the image.

[0065] Optionally, the value range of the voltage parameter is from 100 kV to 150 kV. kV is the voltage unit, representing kilovolt.

[0066] In a specific embodiment, the voltage parameter is 130 kV.

[0067] Optionally, the value range of the current parameter is from 1 mA to 3 mA. mA is the current unit, representing milliampere.

[0068] In a specific embodiment, the current parameter is 2 mA.

[0069] Optionally, the value range of the integration time parameter is from 250 ms to 350 ms. ms is the time unit, representing millisecond.

[0070] In a specific embodiment, the integration time parameter is 300 ms.

[0071] S104, perform binary segmentation processing on the X-ray image to generate a binary image.

[0072] A binary image, that is, a 0-1 image, refers to processing each pixel point on the image so that it has only two possible values or gray-level states. For example, a black-and-white image is used to represent a binary image.

[0073] Based on the Otsu algorithm (Inter-class Variance Method), the X-ray image is binarized and segmented to generate a binary image (as Figure 13 shown).

[0074] It should be noted that the Otsu algorithm is an adaptive image binarization method that automatically determines the optimal threshold by maximizing the inter-class variance between the foreground and the background. Its core idea is: assuming that there are two types of pixels (foreground and background) in the image, the optimal threshold should maximize the variance between the two classes, thus ensuring the strongest contrast in the segmented image.

[0075] The advantages of using the Otsu algorithm are as follows: It does not require manual threshold setting and can automatically adapt to the contrast differences of different X-ray images; by maximizing the inter-class variance, it reduces the interference of noise on the segmentation result; it is beneficial to maintaining the integrity of the contour of the bearing to be measured, effectively retaining the boundaries of the inner ring, outer ring, and rolling parts of the bearing, and providing a clear geometric structure for subsequent polar coordinate transformation.

[0076] S106. Convert the binary image into a polar coordinate image, where the origin of the polar coordinate system in the polar coordinate image is the center of the bearing to be measured.

[0077] The purpose of this step is to convert the circular structure into a linear arrangement. The rolling parts of the bearing to be measured are evenly distributed along the circumferential direction and appear as discrete circular arrangements in the rectangular coordinate system (Cartesian coordinate system). Through polar coordinate transformation, the periodic structure on the circumference is mapped into a line in the horizontal direction in the polar coordinate image (the polar angle corresponds to the row coordinate, and the polar radius corresponds to the column coordinate), making the defect characteristics of the rollers or balls more intuitive in the horizontal direction.

[0078] Taking the center of the bearing to be measured as the origin of the polar coordinate system ensures that the image after polar coordinate transformation (polar coordinate image, as Figure 14 shown) is aligned with the actual geometric center of the bearing to be measured, avoiding signal analysis errors caused by the offset of the placement position of the bearing to be measured.

[0079] Polar coordinate transformation converts the two-dimensional image processing problem into one-dimensional signal analysis (horizontal line integral), reducing the computational amount and meeting the requirements of industrial real-time detection. With the bearing center as the origin, regardless of how the bearing to be measured rotates or scales in the image, the periodic arrangement of the rollers or balls in the converted polar coordinate image is maintained, ensuring the consistency of the detection results.

[0080] S108, perform a horizontal integration on the polar coordinate image to generate a line integral signal presented as a first curve graph.

[0081] The horizontal direction (polar angle θ) of the polar coordinate image corresponds to the circumferential distribution of the rolling part, and the vertical direction (radial distance r) corresponds to the radial structure of the bearing to be measured. Through horizontal integration, the pixel values of each row of the two-dimensional image are accumulated into a one-dimensional signal, highlighting the periodic distribution characteristics of the rolling part.

[0082] The purpose of this step is to convert the two-dimensional image into a one-dimensional signal for subsequent peak detection. The integration process can add the pixel values of each row, so that the positions of the rolling parts (rollers or balls) will form peaks in the line integral signal, while the missing rolling parts will cause the peaks to decrease or disappear.

[0083] The integration process can highlight the periodic characteristics and reduce the influence of noise. Because integration is equivalent to summing the pixels of each row, it may smooth out some random noise. At the same time, reducing the two-dimensional data to one-dimensional simplifies the subsequent signal processing steps, and the peak detection algorithm will be more efficient.

[0084] It should be noted that the horizontal axis of the first curve graph (as Figure 15 shown) is the row coordinate, and the vertical axis of the first curve graph is the integral value.

[0085] S110, based on the first curve graph, perform peak detection on the line integral signal. When the detected peak value is less than the preset threshold, it is determined that the bearing to be measured has a defect.

[0086] The purpose of peak detection is to identify the positions of the rolling parts (rollers or balls). Normally, each rolling part will correspond to a peak. If the peak value is less than the preset threshold, it indicates that there may be defects or missing rollers or balls. The purpose of this step is to convert the line integral signal into a defect judgment.

[0087] Judge whether the peak value corresponding to each peak is less than the preset threshold and generate a first judgment result; if the first judgment result is yes, indicating that at least one peak value is less than the preset threshold, it is determined that the bearing to be measured has a defect (there are defects or missing rollers or balls); if the first judgment result is no, indicating that each peak value is greater than or equal to the preset threshold, it is determined that the bearing to be measured has no defect.

[0088] It should be noted that the preset threshold can be an empirical threshold.

[0089] During the entire defect detection process, no manual intervention is required, only simple peak comparison logic is needed, without a deep learning model or complex pattern matching, which is beneficial to reducing the deployment cost and meeting the real-time detection requirements of the assembly line. This design method has a high degree of automation and is efficient.

[0090] The present invention aims to provide a method for identifying bearing defects, which identifies the bearing to be measured based on the processing of X-ray images. Compared with the identification method based on deep learning, it does not require a large amount of image data as training samples, and can improve the detection efficiency and the accuracy of detection results while reducing costs.

[0091] The method provided by the present invention (bearing defect identification method) is applicable to roller bearings, ball bearings, etc.

[0092] Taking a roller bearing as an example: A roller bearing includes an inner bearing ring, a plurality of rollers, and an outer bearing ring. The outer bearing ring is sleeved on the inner bearing ring, and a plurality of rollers are arranged between the inner bearing ring and the outer bearing ring. The outer bearing ring and the inner bearing ring can achieve relative rotation through a plurality of rollers. The rollers are cylindrical and are evenly distributed along the circumferential direction between the inner bearing ring and the outer bearing ring.

[0093] X-rays have strong penetration ability and can penetrate metal materials (such as the inner bearing ring, the outer bearing ring, and the rollers), so as to obtain a two-dimensional projection image of the internal structure of the roller bearing. Since substances with different densities have different absorption rates (absorption capabilities) for X-rays, clear contrast can be formed in the two-dimensional projection image, which is convenient for identifying structural defects such as missing rollers. X-ray detection does not require disassembling the roller bearing and can achieve non-destructive detection of the assembled roller bearing, which is suitable for industrial on-line detection scenarios.

[0094] Since the cylindrical rollers of the roller bearing are evenly distributed along the circumferential direction, periodic signal characteristics (such as line integral signals) will be formed after polar coordinate conversion, which is convenient for identifying defects through peak detection. The missing of a roller will directly cause a significant decrease in the signal intensity in a certain angular region of the polar coordinate image (for example, the peak value is less than a preset threshold).

[0095] It should be noted that the technical solution of the present invention identifies the defects of the roller bearing based on X-ray images. Compared with the method of industrial CT (industrial computerized tomography), it is beneficial to improve the detection speed, is suitable for large-scale on-line detection, and can also reduce the equipment cost; compared with the method of ultrasonic detection, it can penetrate metal materials and is beneficial to improve the accuracy of detection results.

[0096] In some embodiments, optionally, as Figure 2 shown, before S104 (performing binary segmentation processing on the X-ray image to generate a binary image), the steps of the bearing defect identification method further include:

[0097] S103, preprocessing the X-ray image.

[0098] Gray-scale process the X-ray image. When the X-ray image is a color image, convert it to a grayscale image. Specifically, convert each pixel of the color image to a grayscale value to obtain the grayscale image.

[0099] Filter and denoise the X-ray image. To reduce the possible noise interference in the X-ray image, use Gaussian filtering to smooth the grayscale image.

[0100] In some embodiments, optionally, as Figure 3 shown, based on the Otsu algorithm, perform binary segmentation on the X-ray image to generate a binary image. The steps include:

[0101] S1042, Calculate the grayscale histogram.

[0102] Count the number of pixels at each grayscale level in the X-ray image to generate a grayscale histogram. Use the grayscale level as the abscissa and the corresponding number of pixels as the ordinate to visually display the distribution of grayscale values in the image.

[0103] S1044, Traverse all possible thresholds.

[0104] Based on each candidate threshold T, divide the image into foreground (grayscale ≤ T) and background (grayscale > T).

[0105] S1046, Calculate the between-class variance.

[0106] The pixel proportion of the foreground is ω0, and the average grayscale value is μ0. The pixel proportion of the background is ω1, and the average grayscale value is μ1.

[0107] The between-class variance σ 2 = ω0 × ω1 × (μ0 - μ1) 2 .

[0108] S1048, Select the optimal threshold.

[0109] Traverse all candidate thresholds T, and take the threshold with the maximum value in the between-class variance σ 2 as the segmentation threshold.

[0110] In some embodiments, optionally, as Figure 4 shown, S106 (Convert the binary image to a polar coordinate image, where the origin of the polar coordinate system in the polar coordinate image is the center of the bearing to be measured) includes:

[0111] S1062, Establish a polar coordinate system with the center of the bearing to be measured as the origin.

[0112] Take the center of the bearing to be measured as the origin of the polar coordinate system to ensure that the image after polar coordinate conversion (polar coordinate image) is aligned with the actual geometric center of the bearing to be measured, and avoid signal analysis errors caused by the offset of the placement position of the bearing to be measured.

[0113] The polar coordinate transformation takes the bearing center as the origin. Regardless of how the bearing to be measured rotates or scales in the image, the converted polar coordinate image maintains the periodic arrangement of the rollers or balls, ensuring the consistency of the detection results.

[0114] S1064, convert the coordinates of the pixel points in the binary image in the rectangular coordinate system into the coordinates in the polar coordinate system to determine the polar coordinate image.

[0115] Convert the pixel points of the image in the Cartesian coordinate system (rectangular coordinate system) into polar angle and polar radius parameters; use the bilinear interpolation method for pixel resampling.

[0116] In some embodiments, optionally, as Figure 5 shown, S108 (perform horizontal integration on the polar coordinate image to generate a line integral signal presented as a first curve graph) includes:

[0117] S1082, accumulate and sum the pixel values of each row of the polar coordinate image to achieve horizontal integration and generate a line integral signal presented as a first curve graph.

[0118] The integration process can add the pixel values of each row. In this way, the positions of the rolling parts (rollers or balls) will form peaks in the line integral signal, while the missing rolling parts will cause the peaks to decrease or disappear.

[0119] The integration process can highlight the periodic characteristics and reduce the influence of noise. Because integration is equivalent to summing the pixels of each row, it may smooth out some random noise. At the same time, reducing the two-dimensional data to one-dimensional simplifies the subsequent signal processing steps, and the peak detection algorithm will be more efficient.

[0120] In some embodiments, optionally, as Figure 6 shown, S110 (perform peak detection on the line integral signal based on the first curve graph, and when the detected peak value is less than the preset threshold, determine that the bearing to be measured has a defect) includes:

[0121] S1102, move the data in the first range in the head of the abscissa of the first curve graph to the tail of the abscissa to achieve signal alignment and obtain an optimized curve graph.

[0122] In actual detection, each time an X-ray image is taken of the bearing to be tested, there may be a certain deviation in the angle at which the bearing to be tested is placed. This will result in inconsistent starting positions of the wave peaks corresponding to the rolling parts in the generated polar coordinate image and the subsequent line integral signal (the first curve graph). By moving the data within the first range at the head of the abscissa of the first curve graph to the tail of the abscissa, the starting position of the signal (line integral signal) can be adjusted, enabling signals under different imaging conditions to be compared on the same basis and eliminating the starting angle differences caused by different placement angles of the bearing to be tested.

[0123] When performing wave peak detection, if the signals are not aligned, it may be difficult to accurately identify the positions and quantities of the wave peaks, increasing the possibility of misjudgment. After aligning the signals, the positions of the wave peaks corresponding to the rolling parts (rollers or balls) are relatively fixed on the optimized curve graph, enabling more accurate and stable subsequent wave peak detection, thereby improving the accuracy of defect judgment.

[0124] Determining the optimized curve graph based on the first curve graph (as Figure 16 shown) is conducive to improving the accuracy of the detection results.

[0125] S1104, determine the preset threshold.

[0126] The preset threshold can be an empirical threshold. Throughout the defect detection process, no manual intervention is required, only simple wave peak comparison logic is needed, without a deep learning model or complex pattern matching, which is conducive to reducing the deployment cost and meeting the real-time detection requirements of the assembly line.

[0127] S1106, determine the wave peak positions in the optimized curve graph and perform wave peak detection. When it is detected that the wave peak value is less than the preset threshold, it is determined that the bearing to be tested has a defect.

[0128] The purpose of wave peak detection is to identify the positions of the rolling parts (rollers or balls). Normally, each rolling part corresponds to a wave peak. If the wave peak value is less than the preset threshold, it indicates that there may be defects or missing rollers or balls. The purpose of this step is to convert the line integral signal into a defect judgment.

[0129] It should be noted that in Figure 17 , wave peak detection is performed based on the optimized curve graph. The preset threshold appears as "the dotted line with a vertical coordinate of 10" in Figure 17 . Figure 17 "num = 22" in

[0130] In some embodiments, optionally, the data within the first range is the first 10% of the data at the head of the abscissa.

[0131] It should be noted that in the first curve graph of the online integral signal, the endpoints of the abscissa are 0, 20, 40, 60, 80, and 100 in sequence. Move the data from 0 to 10 in the abscissa (the first 10% of the data at the head of the abscissa) to the tail of the abscissa, so as to achieve signal alignment.

[0132] After aligning the signal, the position of the wave peak corresponding to the rolling part (roller or ball) is relatively fixed on the optimized curve graph, which can make the subsequent wave peak detection more accurate and stable, thereby improving the accuracy of defect judgment.

[0133] In some embodiments, optionally, as Figure 7 shown, to determine the wave peak position in the optimized curve graph and perform wave peak detection, the steps include:

[0134] S1112, determine the wave peak position in the optimized curve graph based on the local maximum detection algorithm or the derivative method.

[0135] This step aims to solve the technical problem of how to accurately judge the wave peak position. By adopting the local maximum detection algorithm or the derivative method, it is beneficial to accurately judge the wave peak position, thereby improving the accuracy of the detection result.

[0136] S1114, determine whether the wave peak value corresponding to the wave peak position is less than the preset threshold.

[0137] Judge whether the wave peak value corresponding to each wave peak is less than the preset threshold, and generate a first judgment result; if the first judgment result is yes, it means that at least one wave peak value is less than the preset threshold, and it is determined that the bearing to be tested has a defect (the roller or ball has a defect or is missing); if the first judgment result is no, it means that each wave peak value is greater than or equal to the preset threshold, and it is determined that the bearing to be tested has no defect.

[0138] In some embodiments, optionally, as Figure 8 shown, S1104 (determine the preset threshold) includes:

[0139] S1122, determine the preset threshold based on the statistical average value of the wave peak values of the line integral signal of the defect-free bearing.

[0140] Optionally, the value range of the preset threshold is 75% to 90% of the statistical average value.

[0141] By determining the preset threshold according to the statistical average value, it is beneficial to improve the accuracy of wave peak detection and effectively reduce the possibility of misjudgment.

[0142] In some embodiments, optionally, as Figure 9 shown, S1104 (determine the preset threshold) includes:

[0143] S1124. Determine a preset threshold based on the minimum peak value of the line integral signal of a defect-free bearing.

[0144] Optionally, the value range of the preset threshold is 75% to 90% of the minimum peak value.

[0145] Determining the preset threshold according to the minimum peak value is beneficial to improving the accuracy of peak detection and effectively reducing the possibility of misjudgment.

[0146] In an embodiment of the present invention, as Figure 10 shown, the bearing defect recognition system 200 includes an image acquisition module 210, an image processing module 220, an image conversion module 230, a signal generation module 240, and a defect recognition module 250.

[0147] The image acquisition module 210 is used to acquire an X-ray image including the internal structure of the bearing to be measured.

[0148] Optionally, the image acquisition module 210 performs imaging on the bearing to be measured through an X-ray device to acquire an X-ray image including the internal structure of the bearing to be measured.

[0149] The bearing to be measured in the present invention can be a roller bearing or a ball bearing. In the case where the bearing to be measured is a roller bearing, the plurality of rolling parts are a plurality of rollers, the rollers are cylinders, and are evenly distributed along the circumferential direction between the bearing inner ring and the bearing outer ring. In the case where the bearing to be measured is a ball bearing, the plurality of rolling parts are a plurality of balls, the balls are spheres, and are evenly distributed along the circumferential direction between the bearing inner ring and the bearing outer ring.

[0150] X-rays have strong penetration ability and can penetrate the internal structure of the bearing to be measured (bearing inner ring, bearing outer ring, and rolling parts), so as to obtain a two-dimensional projection image of the internal structure of the bearing to be measured. The two-dimensional projection image here is an X-ray image. Since the absorption rates (absorption capabilities) of the bearing inner ring, bearing outer ring, and rolling parts for X-rays are different, a clear contrast can be formed in the X-ray image, which is convenient for identifying the defects of the bearing to be measured.

[0151] Optionally, the X-ray device includes a transmitting part and a receiving part that can cooperate with each other during the working process. Among them, the transmitting part is used to emit X-rays; the receiving part is used to receive X-rays.

[0152] The imaging parameters of the X-ray device include one or a combination of the following: voltage parameter, current parameter, and integration time parameter. In the case where the imaging parameter is one, the imaging parameter is any one of the voltage parameter, current parameter, and integration time parameter; in the case where the imaging parameter is multiple, the imaging parameter is any combination of the voltage parameter, current parameter, and integration time parameter.

[0153] It should be noted that the voltage parameter determines the energy of the X-ray, affecting the penetration ability and image contrast. The current parameter determines the intensity of the X-ray, and thus affects the brightness and noise of the image. The integration time parameter can be understood as the time for the receiving part to collect data, which is used to determine the signal-to-noise ratio of the image.

[0154] Optionally, the value range of the voltage parameter is from 100 kV to 150 kV. kV is the voltage unit, representing kilovolt.

[0155] In a specific embodiment, the voltage parameter is 130 kV.

[0156] Optionally, the value range of the current parameter is from 1 mA to 3 mA. mA is the current unit, representing milliampere.

[0157] In a specific embodiment, the current parameter is 2 mA.

[0158] Optionally, the value range of the integration time parameter is from 250 ms to 350 ms. ms is the time unit, representing millisecond.

[0159] In a specific embodiment, the integration time parameter is 300 ms.

[0160] The image processing module 220 is used to perform binary segmentation processing on the X-ray image to generate a binary image.

[0161] The binary image, that is, a 0-1 image, refers to processing each pixel point on the image so that it has only two possible values or gray level states. For example, a black-and-white image is used to represent the binary image.

[0162] Optionally, the image processing module 220 performs binary segmentation processing on the X-ray image based on the Otsu algorithm to generate a binary image.

[0163] It should be noted that the Otsu algorithm is an adaptive image binarization method that automatically determines the optimal threshold by maximizing the between-class variance of the foreground and the background. Its core idea is: assuming that there are two types of pixels (foreground and background) in the image, the best threshold should maximize the variance between the two classes, so as to ensure the strongest contrast of the segmented image.

[0164] The advantages of using the Otsu algorithm are as follows: it does not require manual setting of the threshold, and automatically adapts to the contrast differences of different X-ray images; by maximizing the between-class variance, it reduces the interference of noise on the segmentation result; it is beneficial to maintaining the integrity of the contour of the bearing to be measured, effectively retaining the boundaries of the inner ring, outer ring and rolling part of the bearing, and providing a clear geometric structure for subsequent polar coordinate transformation.

[0165] The image conversion module 230 is used to convert the binary image into a polar coordinate image, and the origin of the polar coordinate system in the polar coordinate image is the center of the bearing to be measured.

[0166] The rolling parts of the bearing to be tested are evenly distributed along the circumferential direction and appear as a discrete circular arrangement in a rectangular coordinate system (Cartesian coordinate system). Through polar coordinate transformation, the periodic structure on the circumference is mapped to a line in the horizontal direction in the polar coordinate image (the polar angle corresponds to the row coordinate, and the polar radius corresponds to the column coordinate), making the defect characteristics of the rollers or balls more intuitive in the horizontal direction.

[0167] Take the center of the bearing to be tested as the origin of the polar coordinate system to ensure that the image after polar coordinate transformation (polar coordinate image) is aligned with the actual geometric center of the bearing to be tested, avoiding signal analysis errors caused by the offset of the placement position of the bearing to be tested.

[0168] Polar coordinate transformation converts the two-dimensional image processing problem into one-dimensional signal analysis (horizontal line integral), reducing the amount of calculation and meeting the requirements of industrial real-time detection. With the bearing center as the origin, regardless of how the bearing to be tested rotates or scales in the image, the periodic arrangement of the rollers or balls in the transformed polar coordinate image is maintained, ensuring the consistency of the detection results.

[0169] The signal generation module 240 is used to perform horizontal integration on the polar coordinate image to generate a line integral signal presented as a first curve graph.

[0170] The horizontal direction (polar angle θ) of the polar coordinate image corresponds to the circumferential distribution of the rolling parts, and the vertical direction (polar radius r) corresponds to the radial structure of the bearing to be tested. Through horizontal integration, the pixel values of each row in the two-dimensional image are accumulated into a one-dimensional signal, highlighting the periodic distribution characteristics of the rolling parts.

[0171] The integration process can add up the pixel values of each row. In this way, the positions of the rolling parts (rollers or balls) will form peaks in the line integral signal, while the missing rolling parts will cause the peaks to decrease or disappear.

[0172] The integration process can highlight the periodic characteristics and reduce the influence of noise. Because integration is equivalent to summing the pixels of each row, it may smooth out some random noise. At the same time, reducing the two-dimensional data to one-dimensional simplifies the subsequent signal processing steps, and the peak detection algorithm will be more efficient.

[0173] It should be noted that the horizontal axis of the first curve graph is the row coordinate, and the vertical axis of the first curve graph is the integral value.

[0174] The defect identification module 250 is used to perform peak detection on the line integral signal based on the first curve graph. When the detected peak value is less than the preset threshold, it is determined that the bearing to be tested has a defect.

[0175] The purpose of peak detection is to identify the positions of the rolling elements (rollers or balls). Under normal circumstances, each rolling element corresponds to a peak. If the peak value is less than the preset threshold, it indicates that there may be defects or missing rollers or balls. The purpose of this step is to convert the line integral signal into a defect judgment.

[0176] Judge whether the peak value corresponding to each peak is less than the preset threshold, and generate a first judgment result; if the first judgment result is yes, indicating that at least one peak value is less than the preset threshold, it is determined that the bearing under test has a defect (the roller or ball has a defect or is missing); if the first judgment result is no, indicating that each peak value is greater than or equal to the preset threshold, it is determined that the bearing under test has no defect.

[0177] It should be noted that the preset threshold can be an empirical threshold.

[0178] During the entire defect detection process, no manual intervention is required. Only simple peak comparison logic is needed, without a deep learning model or complex pattern matching, which is beneficial to reducing the deployment cost and meeting the real-time detection requirements of the production line. This design method has a high degree of automation and is efficient.

[0179] The present invention aims to provide a bearing defect identification system 200, which identifies the defects of the bearing under test based on the processing of X-ray images. Compared with the identification method of deep learning, it does not require a large amount of image data as training samples, and can improve the detection efficiency and the accuracy of the detection results while reducing the cost.

[0180] The system provided by the present invention (bearing defect identification system 200) is applicable to roller bearings, ball bearings, etc.

[0181] Taking a roller bearing as an example: A roller bearing includes an inner bearing ring, a plurality of rollers, and an outer bearing ring. The outer bearing ring is sleeved on the inner bearing ring, and a plurality of rollers are arranged between the inner bearing ring and the outer bearing ring. The outer bearing ring and the inner bearing ring can achieve relative rotation through a plurality of rollers. The rollers are cylindrical and are evenly distributed along the circumferential direction between the inner bearing ring and the outer bearing ring.

[0182] X-rays have strong penetration ability and can penetrate metal materials (such as the inner bearing ring, the outer bearing ring, and the rollers), so as to obtain a two-dimensional projection image of the internal structure of the roller bearing. Since substances with different densities have different absorption rates (absorption abilities) for X-rays, clear contrasts can be formed in the two-dimensional projection image, which is convenient for identifying structural defects such as missing rollers. X-ray detection does not require disassembling the roller bearing and can achieve non-destructive detection of the assembled roller bearing, which is suitable for industrial on-line detection scenarios.

[0183] Since the cylindrical rollers of a roller bearing are evenly distributed in the circumferential direction, periodic signal features (such as line integral signals) will be formed after polar coordinate conversion, which is convenient for defect identification through peak detection. The absence of a roller will directly lead to a significant reduction in the signal intensity in a certain angular region of the polar coordinate image (for example, the peak value is less than a preset threshold).

[0184] It should be noted that the technical solution of the present invention is based on the X-ray image method to identify defects in roller bearings. Compared with the industrial CT method, it is beneficial to improve the detection speed, suitable for large-scale online detection, and can also reduce the equipment cost; compared with the ultrasonic detection method, it can penetrate metal materials and is beneficial to improve the accuracy of detection results.

[0185] In some embodiments, optionally, the bearing defect identification system 200 further includes a preprocessing module. The preprocessing module is used to preprocess the X-ray image.

[0186] The preprocessing module is used to grayscale the X-ray image. When the X-ray image is a color image, it is converted into a grayscale image. Specifically, each pixel point of the color image is converted into a grayscale value to obtain the grayscale image.

[0187] The preprocessing module is used to filter and denoise the X-ray image. In order to reduce the possible noise interference in the X-ray image, Gaussian filtering is used to smooth the grayscale image.

[0188] In some embodiments, optionally, the image processing module 220 performs binary segmentation processing on the X-ray image based on the Otsu algorithm to generate a binary image, including: calculating the grayscale histogram; traversing all possible thresholds; calculating the between-class variance; and selecting the optimal threshold.

[0189] In some embodiments, optionally, the image conversion module 230 is used to convert the binary image into a polar coordinate image, and the origin of the polar coordinate system in the polar coordinate image is the center of the bearing to be measured, including: establishing a polar coordinate system with the center of the bearing to be measured as the origin; converting the coordinates of the pixel points in the binary image in the rectangular coordinate system into the coordinates in the polar coordinate system to determine the polar coordinate image.

[0190] In some embodiments, optionally, the signal generation module 240 is used to perform horizontal integration on the polar coordinate image to generate a line integral signal presented as a first curve graph, including: accumulating and summing the pixel values of each row of the polar coordinate image to achieve horizontal integration and generate a line integral signal presented as a first curve graph.

[0191] In some embodiments, optionally, the defect recognition module 250 is configured to perform peak detection on the line integral signal based on the first curve graph. When the detected peak value is less than a preset threshold, it is determined that the bearing under test has a defect, including: moving the data within the first range at the head of the abscissa of the first curve graph to the tail of the abscissa to achieve signal alignment, obtaining an optimized curve graph; determining the preset threshold; determining the peak positions in the optimized curve graph and performing peak detection. When the detected peak value is less than the preset threshold, it is determined that the bearing under test has a defect.

[0192] In one embodiment of the present invention, as Figure 11 shown, the electronic device 300 includes a memory 310 and a processor 320. Among them, a program or instruction that can run on the processor 320 is stored on the memory 310. When the processor 320 executes the program or instruction, the steps of the bearing defect recognition method in any of the above embodiments are implemented. The electronic device 300 has the beneficial effects of any of the above embodiments, which will not be elaborated here.

[0193] In one embodiment of the present invention, a readable storage medium stores a program or instruction. When the program or instruction is executed by a processor, the steps of the bearing defect recognition method in any of the above embodiments are implemented. The readable storage medium has the beneficial effects of any of the above embodiments, which will not be elaborated here.

[0194] In the present invention, the terms "first", "second", and "third" are only used for descriptive purposes and cannot be construed as indicating or implying relative importance; the term "plurality" refers to two or more, unless otherwise clearly defined. Terms such as "installed", "connected", "connected to", and "fixed" should be understood in a broad sense. For example, "connected" can be a fixed connection, a detachable connection, or an integral connection; "connected" can be a direct connection or an indirect connection through an intermediate medium. For those of ordinary skill in the art, the specific meanings of the above terms in the present invention can be understood according to specific circumstances.

[0195] In the description of the present invention, it should be understood that the orientation or positional relationship indicated by terms such as "upper", "lower", "left", "right", "front", "rear", etc. is based on the orientation or positional relationship shown in the drawings. It is only for the convenience of describing the present invention and simplifying the description, rather than indicating or implying that the device or unit referred to must have a specific direction, be constructed and operated in a specific orientation. Therefore, it cannot be construed as a limitation to the present invention.

[0196] In the description of this specification, the descriptions of the terms "one embodiment", "some embodiments", "specific embodiments", etc. mean that the specific features, structures, materials or characteristics described in connection with the embodiment or example are included in at least one embodiment or example of the present invention. In this specification, the schematic representations of the above terms do not necessarily refer to the same embodiment or instance. Moreover, the specific features, structures, materials or characteristics described may be combined in any one or more embodiments or examples in a suitable manner.

[0197] The above are only the preferred embodiments of the present invention and are not intended to limit the present invention. For those skilled in the art, the present invention may have various modifications and variations. Any modifications, equivalent replacements, improvements, etc. made within the spirit and principle of the present invention shall be included within the protection scope of the present invention.

Claims

1. A bearing defect identification method, characterized in that: include: Acquire an X-ray image containing the internal structure of the bearing to be tested; Performing binary segmentation processing on the X-ray image to generate a binary image; Converting the binary image into a polar coordinate image, wherein the origin of the polar coordinate system in the polar coordinate image is the center of the bearing to be tested; Integrate the polar coordinate image in a horizontal direction to generate a line integral signal presented as a first curve graph; Based on the first curve graph, performing peak detection on the line integral signal, and determining that the bearing to be tested has a defect when the peak value is detected to be less than a preset threshold; The step of integrating the polar coordinate image in a horizontal direction to generate a line integral signal presented as a first curve graph includes: By integrating horizontally, the pixel values ​​of each row of the two-dimensional image are accumulated into a one-dimensional signal, highlighting the periodic distribution characteristics of the rolling part. The integration processing can add the pixel values ​​of each row. The position of the rolling part will form a peak in the line integration signal, and the missing rolling part will cause the peak to decrease or disappear.

2. The bearing defect identification method according to claim 1, characterized in that: The step of performing peak detection on the line integral signal based on the first curve graph, and determining that the bearing to be tested has a defect when the peak value is detected to be less than a preset threshold, includes: Move the data in the first range in the head of the abscissa of the first curve graph to the tail of the abscissa to achieve signal alignment and obtain an optimized curve graph; Determining the preset threshold; The peak position in the optimization curve diagram is determined and peak detection is performed. When it is detected that the peak value is less than the preset threshold, it is determined that the bearing to be tested has a defect.

3. The bearing defect identification method according to claim 2, characterized in that: The data in the first range is the first 10% of the data at the head of the horizontal axis.

4. The bearing defect identification method according to claim 2, characterized in that: The step of determining the peak position in the optimization curve graph and performing peak detection includes: Determining the peak position in the optimization curve graph based on a local maximum detection algorithm or a derivative method; Determine whether the peak value corresponding to the peak position is less than the preset threshold.

5. The bearing defect identification method according to claim 2, characterized in that: The determining the preset threshold comprises: The preset threshold is determined based on the statistical average of the peak values ​​of the line integral signal of the defect-free bearing.

6. The bearing defect identification method according to claim 2, characterized in that: The determining the preset threshold comprises: The preset threshold is determined based on a minimum peak value of the line integral signal of a defect-free bearing.

7. The bearing defect identification method according to any one of claims 1 to 6, characterized in that: The step of integrating the polar coordinate image in a horizontal direction to generate a line integral signal presented as a first curve graph includes: The pixel values ​​of each row of the polar coordinate image are cumulatively summed to achieve horizontal integration and generate the line integral signal presented in the first curve graph.

8. A bearing defect identification system, characterized in that: include: An image acquisition module (210) is used to acquire an X-ray image containing the internal structure of the bearing to be tested; An image processing module (220) is used to perform binary segmentation processing on the X-ray image to generate a binary image; An image conversion module (230) is used to convert the binary image into a polar coordinate image, wherein the origin of the polar coordinate system in the polar coordinate image is the center of the bearing to be tested; A signal generating module (240), configured to perform horizontal integration on the polar coordinate image to generate a line integral signal presented as a first curve graph; A defect identification module (250) is used to perform peak detection on the line integral signal based on the first curve graph, and determine that the bearing to be tested has a defect when the peak value is detected to be less than a preset threshold; The signal generating module (240) is specifically used for: By integrating horizontally, the pixel values ​​of each row of the two-dimensional image are accumulated into a one-dimensional signal, highlighting the periodic distribution characteristics of the rolling part. The integration processing can add the pixel values ​​of each row. The position of the rolling part will form a peak in the line integration signal, and the missing rolling part will cause the peak to decrease or disappear.

9. An electronic device, characterized in that: include: A memory (310) and a processor (320), wherein the memory (310) stores a program or instruction that can be run on the processor (320), and when the processor (320) executes the program or the instruction, the steps of the bearing defect identification method according to any one of claims 1 to 7 are implemented.

10. A readable storage medium, characterized in that: The readable storage medium stores a program or an instruction, and when the program or the instruction is executed by a processor, the steps of the bearing defect identification method according to any one of claims 1 to 7 are implemented.

Images