Fusion decision-based equipment wear stage recognition method and system

Abstract

This invention belongs to the field of mechanical equipment wear condition monitoring technology. It proposes a method and system for identifying equipment wear stages based on fusion decision-making. The method extracts wear rate features, wear degree features, cutting index features, fatigue index features, and oxidation index features from abrasive sample images and clusters them to obtain five sets of discrete stage labels. These five sets of discrete stage labels are then normalized to obtain a normalized data matrix. Based on the data matrix, the weights corresponding to each feature are calculated using multiple methods, and the average of the weights calculated using these methods is the final weight of each feature. Based on the final weights of each feature and the normalized data matrix, a comprehensive score sequence for all abrasive sample images is calculated. Clustering is then performed based on the comprehensive score sequence to obtain multiple optimal cluster centers, and the current wear stage is determined based on the optimal cluster centers. This achieves a more comprehensive and accurate division of wear stages.

Description

Technical Field

[0001] This invention relates to the field of mechanical equipment wear condition monitoring technology, specifically to a method and system for identifying equipment wear stages based on fusion decision-making. Background Technology

[0002] The statements in this section are merely background information related to the present invention and do not necessarily constitute prior art.

[0003] Wear-related failures account for over 75% of mechanical equipment failures. The wear of mechanical components is a dynamic and gradual process, progressing from slight wear to normal wear, and eventually to severe wear and failure. This process is typically divided into several stages based on changes in wear information, such as the break-in period, stabilization period, acceleration period, and failure period. Lubricating oil is a key element ensuring the healthy and stable operation of mechanical equipment, carrying wear particles generated by various components. By analyzing the characteristics of the abrasive particles in the lubricating oil, the changes in the wear stage of components can be determined in real time, thereby enabling the monitoring, analysis, and prediction of the equipment's wear condition.

[0004] Early wear monitoring primarily relies on ferrography or particle counters to extract abrasive particle concentration and size information, reflecting quantitative changes by calculating wear rate and wear severity (such as the abrasive coverage area index IPCA). Subsequently, researchers proposed extended indices based on IPCA, such as the Total Abrasive Concentration Index (TIDC) and the Cumulative Dimensionless Abrasive Concentration Index (ADIDC), to compensate for the limitations of single indices in identifying the pitting stage of gears. Furthermore, the number of large abrasive particles (LPQ) and its size gradient changes have been introduced to enhance sensitivity to later wear stages. However, when these macroscopic quantitative indices, such as wear rate and wear severity, are similar or identical, they still fail to reveal qualitative changes in the wear mechanism, leading to a "simultaneous quantity, heterogeneous quality" phenomenon, making it difficult to accurately distinguish between different wear stages.

[0005] With the development of microscopic imaging and image processing technologies, the microscopic features of abrasive particles, such as shape, texture, and color, can be efficiently acquired and quantitatively analyzed, providing new insights into wear mechanisms. However, while traditional macroscopic quantitative indicators, such as wear rate and wear degree, can reflect the "quantitative change" in wear, they cannot reveal the "qualitative transformation" driven by different mechanisms when the values ​​are similar, easily leading to a "same quantity, different quality" blind spot. Although microscopic imaging technology has enabled the efficient acquisition of microscopic mechanistic features of abrasive particles, such as shape, texture, and color, existing research is mostly limited to abrasive particle classification, failing to systematically apply mechanistic information to the classification of wear stages, resulting in insufficient overall identification accuracy. Summary of the Invention

[0006] To address the shortcomings of existing technologies, this invention provides a method and system for identifying equipment wear stages based on fusion decision-making. This method achieves a more comprehensive and accurate division of wear stages, not only avoiding the limitations of traditional wear indicators being "homogeneous but heterogeneous," but also providing interpretable mechanistic support for online monitoring and fault early warning.

[0007] To achieve the above objectives, the present invention adopts the following technical solution:

[0008] In a first aspect, the present invention provides a method for identifying equipment wear stages based on fusion decision-making.

[0009] A method for identifying equipment wear stages based on fusion decision-making includes the following process:

[0010] Wear rate features, wear degree features, cutting index features, fatigue index features, and oxidation index features are extracted from multiple abrasive sample images;

[0011] Clustering was performed on wear rate characteristics, wear degree characteristics, cutting index characteristics, fatigue index characteristics, and oxidation index characteristics to obtain five groups of discrete stage labels;

[0012] The five sets of discrete stage labels are normalized to obtain a normalized data matrix;

[0013] Based on the data matrix, the weights corresponding to wear rate characteristics, wear degree characteristics, cutting index characteristics, fatigue index characteristics and oxidation index characteristics are calculated in various ways, and the average of the weights calculated in various ways is the final weight of each characteristic.

[0014] Based on the final weights of each feature and the normalized data matrix, calculate the comprehensive score sequence of all abrasive sample images.

[0015] Clustering is performed based on the comprehensive score sequence to obtain multiple optimal cluster centers, and the current wear stage is determined based on the optimal cluster centers.

[0016] In one implementation of the first aspect of the present invention, before extracting wear rate features, wear degree features, cutting index features, fatigue index features, and oxidation index features from multiple abrasive sample images, a preprocessing process is further included, comprising:

[0017] The abrasive sample images were sequentially processed by grayscale, background reduction, binarization, and morphological processing.

[0018] In one implementation of the first aspect of the present invention, the wear rate characteristic is the actual area of ​​the abrasive region, the wear degree characteristic is the number of abrasive grains with an equivalent circle diameter exceeding a set threshold, the cutting index characteristic is the proportion of cutting abrasive grains, the fatigue index characteristic is the proportion of fatigued abrasive grains, and the oxidation index characteristic is the proportion of oxidized abrasive grains.

[0019] As a further limitation of the first aspect of the invention, the area, equivalent circle diameter, aspect ratio, roundness and standard deviation of curvature of each abrasive grain are calculated.

[0020] When the equivalent circle diameter is greater than or equal to the first set threshold and the roundness is greater than or equal to the second set threshold, it is a normal abrasive grain; otherwise, proceed to the next step of judgment.

[0021] When the standard deviation of curvature is less than or equal to the third set threshold and the roundness is greater than or equal to the fourth set threshold, it is an oxidized abrasive particle; otherwise, proceed to the next step of judgment.

[0022] When the aspect ratio is greater than or equal to the fifth set threshold and the roundness is less than or equal to the second set threshold, it is a cutting abrasive; otherwise, it is a fatigue sliding abrasive.

[0023] In one implementation of the first aspect of the present invention, the wear rate characteristics, wear degree characteristics, cutting index characteristics, fatigue index characteristics and oxidation index characteristics are clustered using the Mean-shift algorithm to divide the wear stages.

[0024] In one implementation of the first aspect of the present invention, the weights corresponding to the wear rate characteristic, wear degree characteristic, cutting index characteristic, fatigue index characteristic and oxidation index characteristic are calculated by the entropy weight method, the CRITIC method and the coefficient of variation method, respectively.

[0025] The final weight of the wear rate feature is the average of the entropy weight, CRITIC weight, and coefficient of variation weight of the wear rate feature;

[0026] The final weight of the wear degree feature is the average of the entropy weight, CRITIC weight, and coefficient of variation weight of the wear degree feature;

[0027] The final weight of the cutting index feature is the average of the entropy weight, CRITIC weight, and coefficient of variation weight of the cutting index feature.

[0028] The final weight of the fatigue index feature is the average of the entropy weight, CRITIC weight, and coefficient of variation weight of the fatigue index feature.

[0029] The final weight of the oxidation index feature is the average of the entropy weight, CRITIC weight, and coefficient of variation weight of the oxidation index feature.

[0030] In one implementation of the first aspect of the present invention, K-means clustering is performed based on the comprehensive score sequence to calculate the optimal number of clusters K, resulting in K cluster centers. The cluster centers are sorted, and adjacent center points are taken as stage boundaries to obtain the current wear stage.

[0031] The current wear stages include: break-in wear stage, stable wear stage, oxidative wear stage, transition stage, accelerated wear stage, and failure stage.

[0032] Secondly, the present invention provides an equipment wear stage identification system based on fusion decision-making.

[0033] A system for identifying equipment wear stages based on fusion decision-making, comprising:

[0034] The feature extraction unit is configured to extract wear rate features, wear degree features, cutting index features, fatigue index features, and oxidation index features from multiple abrasive sample images;

[0035] The feature clustering unit is configured to cluster the wear rate feature, wear degree feature, cutting index feature, fatigue index feature and oxidation index feature respectively to obtain five sets of discrete stage labels;

[0036] The data normalization unit is configured to normalize five sets of discrete stage labels to obtain a normalized data matrix.

[0037] The weight allocation unit is configured to: calculate the weights corresponding to wear rate characteristics, wear degree characteristics, cutting index characteristics, fatigue index characteristics and oxidation index characteristics respectively using multiple methods according to the data matrix, and the average of the weights calculated by various methods is the final weight of each feature;

[0038] The comprehensive score calculation unit is configured to calculate the comprehensive score sequence of all abrasive sample images based on the final weights of each feature and the normalized data matrix.

[0039] The wear stage determination unit is configured to: perform clustering based on the comprehensive score sequence to obtain multiple optimal cluster centers, and determine the current wear stage based on the optimal cluster centers.

[0040] Thirdly, the present invention provides a computer device, comprising: a processor and a computer-readable storage medium;

[0041] A processor, adapted to execute computer programs;

[0042] A computer-readable storage medium storing a computer program, which, when executed by the processor, implements the equipment wear stage identification method based on fusion decision as described in the first aspect of the present invention.

[0043] Fourthly, the present invention provides a computer-readable storage medium storing a computer program adapted to be loaded by a processor and executed as described in the first aspect of the present invention for identifying equipment wear stages based on fusion decision-making.

[0044] Compared with the prior art, the beneficial effects of the present invention are:

[0045] This invention innovatively proposes a method for identifying equipment wear stages based on fusion decision-making. By introducing and comprehensively analyzing three major wear mechanism indicators—cutting, fatigue, and oxidation—it can not only clearly distinguish the dynamic characteristics of each stage—run-in, stable, accelerated, and severe wear—but also deeply reveal the interactive evolution of each mechanism over time: the peak of cutting abrasive particles in the initial stage and their rapid decay reflect the formation of a surface lubricating film during the run-in period; the occasional oscillations in the proportion of oxidation wear and fatigue abrasive particles in the stable stage indicate precursors to instability; the simultaneous fermentation of multiple mechanisms in the accelerated stage reveals the combined effect of crack initiation and synergistic cutting by hard particles; and the multi-indicator explosion in the severe wear stage directly points to the critical state of macroscopic damage. This mechanism-based wear stage segmentation method not only significantly improves the accuracy and robustness of stage segmentation but also provides physical interpretability and decision support for online early warning, life prediction, and fault location.

[0046] This invention organically integrates macroscopic quantitative indicators (wear rate, wear degree) with microscopic mechanistic indicators (cutting index, fatigue index, oxidation index), simultaneously capturing both "quantitative changes" and "qualitative mechanistic transformations" within the same analytical framework. Compared to methods relying solely on single or dual macroscopic indicators, it can reveal the essence of material failure at the microscopic mechanistic level, completely resolving the stage confusion caused by traditional "homogeneous but heterogeneous" methods. It utilizes the Mean-Shift algorithm to cluster each indicator separately and then weights and integrates them within a multi-criteria decision fusion framework of Entropy Weight Method-CRITIC Method-Coefficient of Variation Method (EWM-CRITIC-DWM), significantly improving the identification accuracy of key stages such as break-in, stabilization, acceleration, and failure. Simultaneously, by dynamically analyzing the weight evolution of each feature at different stages, it provides a clear physical mechanism explanation for the stage division results, enhancing the traceability of online early warning and diagnosis.

[0047] Advantages of additional aspects of the invention will be set forth in part in the description which follows, and in part will be obvious from the description, or may be learned by practice of the invention. Attached Figure Description

[0048] The accompanying drawings, which form part of this invention, are used to provide a further understanding of the invention. The illustrative embodiments of the invention and their descriptions are used to explain the invention and do not constitute an improper limitation of the invention.

[0049] Figure 1 A schematic diagram of an equipment wear stage identification method based on fusion decision-making, provided as an exemplary embodiment of the present invention;

[0050] Figure 2 A schematic diagram illustrating the principle of online abrasive particle monitoring is provided as an exemplary embodiment of the present invention.

[0051] Figure 3 A schematic diagram of the abrasive image preprocessing process provided in an exemplary embodiment of the present invention;

[0052] Figure 4 A schematic diagram illustrating the principle of the Mean Shift algorithm provided as an exemplary embodiment of the present invention. Figure 4 In this context, (A) represents the process of step 1. Figure 4 (B) in the text represents step 2. Figure 4 (C) in the text represents step 3. Figure 4 (D) in the text represents step 4. Figure 4 (E) in the text represents the process of step 5. Figure 4 (F) in the text represents the process of step 6;

[0053] Figure 5 This diagram illustrates the changes in five key characteristics during the wear process, as provided in an exemplary embodiment of the present invention. In this diagram, PQ-O represents the proportion of oxidized abrasive particles, PQ-C represents the proportion of cutting abrasive particles, PQ-F represents the proportion of fatigue abrasive particles, and IPCA represents the abrasive particle coverage area index. This represents the proportion of large abrasive particles;

[0054] Figure 6 A schematic diagram of the integrated stage division results provided for an exemplary embodiment of the present invention;

[0055] Figure 7 A schematic diagram illustrating the division of wear stages based on wear rate, provided as an exemplary embodiment of the present invention;

[0056] Figure 8 A schematic diagram illustrating the division of wear stages based on wear degree as an exemplary embodiment of the present invention;

[0057] Figure 9 This is a schematic diagram of abrasive grain images at different times, provided as an exemplary embodiment of the present invention. Figure 9 (a) in the diagram represents the abrasive grain pattern at 20 minutes. Figure 9 (b) in the diagram represents the abrasive grain pattern at 70 minutes. Figure 9 (c) in the diagram represents the abrasive grain pattern at 370 minutes. Figure 9 (d) in the diagram represents the abrasive grain pattern at 480 minutes.

[0058] Figure 10 A schematic diagram of an equipment wear stage identification system based on fusion decision-making, provided as an exemplary embodiment of the present invention;

[0059] Figure 11 A schematic diagram of a computer device provided as an exemplary embodiment of the present invention;

[0060] Among them, 1. rolling element; 2. abrasive grains; 3. peristaltic pump; 4. first image sensor; 5. second image sensor; 6. abrasive grain annular image; 7. dispersed abrasive grain image. Detailed Implementation

[0061] The present invention will be further described below with reference to the accompanying drawings and embodiments.

[0062] It should be noted that the following detailed descriptions are exemplary and intended to provide further illustration of the invention. Unless otherwise specified, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains.

[0063] As described in the background section, existing wear evaluation indicators, such as wear rate and wear degree, can only quantitatively reflect the "quantitative change" in wear degree, and are unable to reveal the "mechanistic qualitative change" in the wear process, thus failing to meet the requirements for high-precision wear stage identification. In view of this, this implementation proposes a method for equipment wear stage identification based on fusion decision-making, including the following process:

[0064] S101: Extract wear rate features, wear degree features, cutting index features, fatigue index features, and oxidation index features from multiple abrasive sample images;

[0065] S102: Clustering is performed on wear rate characteristics, wear degree characteristics, cutting index characteristics, fatigue index characteristics and oxidation index characteristics to obtain five sets of discrete stage labels;

[0066] S103: Normalize the five sets of discrete stage labels to obtain the normalized data matrix;

[0067] S104: Based on the data matrix, the weights corresponding to the wear rate feature, wear degree feature, cutting index feature, fatigue index feature and oxidation index feature are calculated in various ways respectively, and the average of the weights calculated in various ways is the final weight of each feature.

[0068] S105: Calculate the comprehensive score sequence of all abrasive sample images based on the final weights of each feature and the normalized data matrix.

[0069] S106: Cluster the data according to the comprehensive score sequence to obtain multiple optimal cluster centers, and determine the current wear stage based on the optimal cluster centers.

[0070] In S101 of this implementation, specifically, it includes:

[0071] Abrasive images can provide rich abrasive features, such as quantity, size, shape, and texture, and can thus characterize different wear indicators. To obtain abrasive images during the wear process of components, an abrasive image sensor is used to acquire abrasive images, which can simultaneously acquire abrasive features that reflect the quantity and morphology of abrasive particles.

[0072] like Figure 2 The diagram shows the specific equipment used, including a rolling element 1, abrasive particles 2, a peristaltic pump 3, a first image sensor 4, and a second image sensor 5. Ultimately, it can obtain annular abrasive particle images 6 and dispersed abrasive particle images 7, respectively. Abrasive particles 2, generated by bearing wear, are carried by lubricating oil and pumped into the sensor's flow channel by the peristaltic pump 3 at a flow rate of 120 ml / min. They flow through the deposition zone, which consists of an oil pool and a glass observation window. A high-gradient magnetic field is formed within the tiny gap between the inner and outer annular electromagnets, attracting the passing ferromagnetic abrasive particles to form an abrasive particle ring. CMOS camera 1 (i.e., the first image sensor 4) captures the image of the abrasive particle ring. Then, the electromagnets are disconnected, and the lubricating oil disperses the abrasive particle ring. At this time, CMOS camera 2 (i.e., the second image sensor 5) captures a dynamic image of dispersed abrasive particles. By connecting the sensors to the equipment's lubrication system, real-time acquisition of abrasive particle images during operation can be achieved. The acquired abrasive particle images (including annular abrasive particle image 6 and dispersed abrasive particle image 7) are then transmitted to a computer for subsequent processing and analysis.

[0073] To extract the features of abrasive grains from the image, it is necessary to eliminate interference from other factors in the image and preprocess the abrasive grain image. The processing flow is as follows: Figure 3 As shown, after the abrasive particles are imaged, the RGB three channels of the color abrasive particle image contain too much redundant information, which is not conducive to subsequent analysis. In particular, the morphology and concentration characteristics of the abrasive particles have a low correlation with color information. Therefore, grayscale conversion can reduce data dimensionality, highlight the brightness differences of the abrasive particles, and improve processing efficiency. A weighted average method is used to convert the color image to a grayscale image:

[0074] (1);

[0075] in, , , These represent the red, green, and blue channel intensities, respectively, with weights reflecting the human eye's sensitivity to different colors. After grayscale processing, each pixel in the image is represented by only one grayscale value (0-255), making brighter areas easier to distinguish.

[0076] Background elements in abrasive particle images (such as illumination, equipment reflection, and lubricating oil flow) can interfere with abrasive particle detection, easily leading to noise and misjudgments. Background subtraction helps enhance the saliency of the abrasive particle region, thereby extracting features more accurately. By performing a difference operation between the abrasive particle image and the background image, only the abrasive particle region is retained, eliminating interference from background illumination, reflection, and other factors.

[0077] Because the abrasive grains themselves are dark gray, and their grayscale is similar to that of the abrasive-free areas after subtracting the background, the boundaries of the abrasive grains remain indistinct, making direct morphological feature analysis difficult. Therefore, global thresholding segmentation is performed on the image, and the Otsu algorithm is used to automatically calculate the optimal global threshold, dividing the image into foreground (abrasive grains) and background:

[0078] (2);

[0079] in, Grayscale value To calculate the grayscale threshold, threshold segmentation was used to achieve precise segmentation of the abrasive region. The background was completely removed, improving the clarity of the abrasive outline and facilitating further extraction and statistical analysis of features such as particle size and shape.

[0080] To improve the accuracy of feature extraction, further morphological operations are needed on the pre-processed dispersed abrasive grain image 7. Due to illumination or binarization, there are holes or discontinuities inside the abrasive grains, which affects the calculation of the abrasive grain area and the extraction of shape features. Dilation and erosion operations are used to fill the holes and restore the integrity of the pixels in the abrasive grain area. Then, based on the difference between the gray values ​​of the abrasive grain boundary and the background, the contour of each abrasive grain is extracted and numbered to provide a basis for subsequent shape feature analysis.

[0081] Specifically, S102 of this implementation includes:

[0082] Wear rate, wear degree, and wear mechanism together constitute a three-dimensional wear stage identification system encompassing "process-state-essence," providing comprehensive information support for accurate stage division. Simultaneously, these three indicators also need to be described through specific abrasive grain characteristics, as follows:

[0083] The interphase coverage index (IPCA) characterizes the wear rate. IPCA can be viewed as an optical projection of the wear rate. It transforms the abstract "material loss rate" into an observable "spatial coverage change" by quantifying the area of ​​abrasive grain deposition. After image preprocessing, all connected regions in the abrasive ring image can be identified, i.e., areas where foreground pixels are connected. By counting all foreground pixels, the foreground pixel coverage area can be obtained.

[0084] (3);

[0085] Wherein, FPCA represents the coverage area of ​​the foreground pixels in the abrasive ring image. The number of foreground pixels in the abrasive ring image, when the first... When the grayscale value of each pixel is 255, When the grayscale value is 1; when the grayscale value is 0, It is 0.

[0086] Converting pixel area to actual abrasive grain area, if the image pixel resolution is... The actual area of ​​the abrasive region can then be expressed as:

[0087] (4).

[0088] A significant increase in the number of large abrasive particles indicates the onset of accelerated wear. Therefore, the proportion of large abrasive particles is used as an indicator to assess the degree of wear. Abrasive particles with an equivalent circular diameter exceeding 30 μm are considered large abrasive particles, and their proportion is as follows:

[0089] (5);

[0090] in, The proportion of large abrasive particles For a large number of abrasive grains, The calculation method for the equivalent circle diameter, which represents the total number of abrasive grains, will be introduced below.

[0091] Besides normal wear, the wear generation mechanisms can be mainly divided into four types: adhesive wear, fatigue wear, cutting wear, and oxidative wear.

[0092] Adhesive wear often results from contaminated lubricating oil deterioration, or from insufficient lubrication supply and extreme conditions such as high load, high speed, and high temperature, leading to the breakdown of the lubricating film and direct metal-to-metal contact. When the friction pairs slide relative to each other, an adhesive effect occurs, and the resulting adhesive nodes undergo shear fracture. The sheared material either detaches as wear debris or migrates to another surface. Severe adhesion can tear the surface, producing flat, coarse fragments, often with scratches in the direction of sliding.

[0093] When two contact surfaces undergo pure rolling or a combination of rolling and sliding friction, the small contact area will deform slightly due to the load. After multiple stress cycles, the surface will break down at a certain point, forming fatigue fracture. Small pieces of material will peel off from the deformation area. As the cracks gradually extend, large fatigue metal particles will be generated, which can reach hundreds of micrometers in length, resulting in fatigue wear. Once it starts, it will continue to grow until the surface completely fails and ends in catastrophic damage.

[0094] Cutting wear is generally caused by machine part fracture or misalignment, or by abrasive particles being trapped in the lubricating oil. When the machine part surfaces are misaligned, the harder friction surface will penetrate the coupled softer surface, causing cutting, similar to the chips produced during machine tool processing. These chips are generally 2-5 μm wide and about 25-100 μm long. When hard particles are embedded in a soft surface, they will cut into the coupled soft surface, producing chips. These chips are often small, generally only a few micrometers long, but still slender.

[0095] Oxidative wear is a serious problem for mechanical equipment. It is caused by the presence of acidic oxidation products and moisture in the oil, or by the formation of oxides on metal surfaces upon contact with air. These oxides chemically react to damage the surface of machine parts, producing very fine spherical particles that lead to welding and pitting. It is highly destructive and often results in secondary wear, such as adhesive and fatigue wear.

[0096] Wear mechanism is the fundamental cause of abrasive grain formation, directly determining the type, shape, size, and other characteristics of abrasive grains. Conversely, differences in abrasive grain characteristics allow for their classification, thus characterizing the wear mechanism and enabling the inverse deduction of the generation mechanism of abnormal abrasive grains from their characteristics. Therefore, extracting the current abrasive grain morphological characteristics and calculating the proportion of each abrasive grain type (normal abrasive grains, fatigue abrasive grains, cutting abrasive grains, and oxidizing abrasive grains) can quantify the wear mechanism at a given moment.

[0097] Abrasive grain area A: Similar to obtaining IPCA, the abrasive grain area is obtained by calculating the number of pixels occupied by the abrasive grain.

[0098] (6);

[0099] in, The area of ​​a single abrasive grain. To disperse the number of individual abrasive grain pixels in the abrasive grain image, when the first... When the grayscale value of each pixel is 255, When the grayscale value is 1; when the grayscale value is 0, It is 0.

[0100] Equivalent circle diameter D: The diameter of a circle with the same area as the abrasive grain.

[0101] (7);

[0102] in, This is the equivalent diameter of a single abrasive grain.

[0103] Aspect Ratio (AR): The ratio of the abrasive grain's length (L) to its width (W). The minimum bounding rectangle is calculated based on the abrasive grain's profile, yielding its height and width. The larger value is the abrasive grain's length (L), and the smaller value is the width (W). When AR ≈ 1, the abrasive grain is approximately square; when AR > 1, the abrasive grain is elongated or flat.

[0104] (8);

[0105] Roundness R is an important indicator describing how close the shape of an abrasive grain is to a perfect circle. An ideal circle has a roundness of 1, meaning the abrasive grain is almost perfectly round. A roundness less than 1 indicates that the abrasive grain shape deviates from a circle, and the closer it is to 0, the more irregular the shape.

[0106] (9);

[0107] Where P is the perimeter of the abrasive grain, which can be obtained from the profile of the abrasive grain.

[0108] Standard deviation of curvature C: The curvature of a pixel in an abrasive grain is the angle between it and the two consecutive pixels before and after it. The edge contour of a round abrasive grain is relatively smooth, so the curvature change is relatively gentle. However, due to the presence of concave or convex corners, the curvature distribution of other shaped abrasive grains is uneven, and there may even be obvious peaks. By calculating the standard deviation of the curvature of each abrasive grain, round abrasive grains can be distinguished from other abrasive grains.

[0109] To determine different abrasive types based on their characteristics, a three-level discriminant tree model was established to classify four types of wear particles, categorizing all abrasive particles into four types: normal abrasive particles, oxidizing abrasive particles, cutting abrasive particles, and fatigue sliding abrasive particles. Specifically, these include:

[0110] When the equivalent circle diameter Greater than or equal to the first set threshold and roundness Greater than or equal to the second set threshold If the abrasive grains are normal, proceed to the next step of judgment; otherwise, proceed to the next step.

[0111] When the standard deviation of curvature Less than or equal to the third set threshold and roundness Greater than or equal to the fourth set threshold If the condition is met, the particles are oxidized abrasive grains; otherwise, proceed to the next step of the judgment.

[0112] When aspect ratio Greater than or equal to the fifth set threshold and roundness Less than or equal to the second set threshold When the abrasive grains are in the cutting phase, they are cutting abrasive grains; otherwise, they are fatigue sliding abrasive grains.

[0113] In this implementation, since the diameter of normal abrasive grains is generally small and their roundness differs greatly from that of oxide abrasive grains with the same small diameter, the accuracy of the judgment of oxide abrasive grains can be improved by combining the standard deviation of curvature and roundness with the equivalent circular diameter and roundness of normal abrasive grains and other abrasive grains. Compared with fatigue abrasive grains, cutting abrasive grains have a larger aspect ratio and a smaller roundness. Through this classification rule, the four types of abrasive grains can be well identified.

[0114] To accurately describe the wear trend of components, the Mean-Shift algorithm is used to divide the wear stages. Mean-Shift is a non-parametric clustering algorithm used to identify high-density regions in a data distribution, especially suitable for non-linearly distributed data. Its core idea is to iteratively adjust data points to move in the direction of the gradient of the probability density function, eventually converging to the peak of the local density, thus achieving data clustering. The algorithm principle is as follows: Figure 4 As shown.

[0115] Step 1: As Figure 4 As shown in (A), the first sampling point is selected as the center point. A region defined by radius r is called a neighborhood, and points in the neighborhood belong to the same cluster. Calculate the distance from all points in the neighborhood to the center point. The vectors are superimposed to obtain the vector shift, which represents the gradient direction of the data distribution. Red dots represent unlabeled data, and gray dots represent labeled data.

[0116] Step 2: As Figure 4 As shown in (B), the center point Move along the vector shift to obtain the second center point. (Right now Continue repeating steps 1 and 2 until |shift| is less than the convergence value. Then proceed to step 3, where red dots indicate unmarked items and gray dots indicate marked items;

[0117] Step 3: As Figure 4 As shown in (C), check the current cluster. If the distance to other existing clusters is less than the threshold d, the two clusters are merged; if there are no clusters that meet the merging conditions, a new cluster is created, where red dots represent unmarked clusters and gray dots represent marked clusters.

[0118] Step 4: As Figure 4 As shown in (D), steps 1 through 3 are repeated for all unlabeled data points until all points are labeled. The figure shows clusters. That is, the corresponding center point Red dots indicate unmarked areas, while gray dots indicate marked areas.

[0119] Step 5: As Figure 4 As shown in (E), the frequency of each point being visited by different clusters is calculated, and the cluster with the highest frequency of visit is taken as its final cluster. The clusters are shown in the figure. and cluster Red dots indicate unmarked areas, while gray dots indicate marked areas.

[0120] Step 6: As Figure 4 As shown in (F), the partitioning results of all data points (C1, C2, C3, and C4) are obtained, where orange points represent points in cluster C1, blue points represent points in cluster C2, green points represent points in cluster C3, and red points represent points in cluster C4.

[0121] Mean-shift clustering was performed on five quantitative features of the three types of wear indicators to obtain five sets of discrete stage labels:

[0122] (10);

[0123] in, , , , and These represent the first group of discrete stage labels, the second group of discrete stage labels, the third group of discrete stage labels, the fourth group of discrete stage labels, and the fifth group of discrete stage labels, respectively. Indicates the first The sample at the th Features The stage number to which it belongs (may differ from the total number of stages for other features); K is the total number of stages. This invention focuses on how to achieve a comprehensive stage division of five features based on five existing single-feature stage labels, thus obtaining a global stage label set.

[0124] To avoid the influence of subjective experience-based scoring on the final stage segmentation, this paper employs three classic objective weighting methods (entropy weighting, CRITIC method, and coefficient of variation method) to automatically calculate weights for the five features. The entropy weighting method emphasizes information entropy differences, the CRITIC method balances contrast and redundancy, and the coefficient of variation method emphasizes dispersion. By averaging the three weights, the core advantages of each method are considered while mitigating their individual limitations (such as the entropy weighting method neglecting correlation, the CRITIC method being sensitive to extreme values, and the coefficient of variation method overweighting indicators with low mean). This results in a more robust, comprehensive weight allocation scheme that requires no subjective intervention, ensuring the objectivity and interpretability of the final wear stage segmentation results.

[0125] The following process is based on a data matrix normalized by five features. , It is the normalized value of the i-th sample on the j-th feature, which is in the range [0,1].

[0126] (1) Entropy weight method.

[0127] First, calculate the weight of the j-th feature in the i-th sample:

[0128] (11);

[0129] in, This represents the normalized value of the k-th sample on the j-th feature. .

[0130] Then calculate the information entropy of each feature. :

[0131] (12);

[0132] Among them, when Time definition , This represents the number of samples.

[0133] Calculate information redundancy :

[0134] (13).

[0135] Calculate entropy weights after normalization :

[0136] (14);

[0137] in, Indicates the first Information redundancy of each feature .

[0138] (2) CRITIC method.

[0139] Based on the normalized matrix Z, calculate the standard deviation of the j-th feature:

[0140] (15);

[0141] in, For the first The standard deviation of each feature For the first The average value of each feature.

[0142] Calculate the Pearson correlation coefficient between feature pairs. :

[0143] (16);

[0144] in, For the first The average value of each feature.

[0145] calculate Information content of each feature :

[0146] (17).

[0147] Calculate CRITIC weights after normalization :

[0148] (18).

[0149] in, Representing the The amount of information in each feature.

[0150] (3) Coefficient of variation method.

[0151] Based on equation (15) The mean of each feature with standard deviation The coefficient of variation can be calculated. :

[0152] (19);

[0153] Calculate DWM weights after normalization :

[0154] (20);

[0155] in, For the first The coefficient of variation of each feature.

[0156] To take into account the advantages of various weighting methods, a simple averaging method is adopted for weight fusion, resulting in the final comprehensive weight vector as follows:

[0157] (twenty one);

[0158] Wherein, equation (21) satisfies Given a comprehensive weight And the normalized data matrix Z, calculate the first The overall score of each sample:

[0159] (twenty two).

[0160] The comprehensive score sequence obtained based on formula (22) The comprehensive score sequence Perform K-means clustering, calculate the optimal number of clusters K, and obtain K cluster centers. Sort the cluster centers and use adjacent center points as stage boundaries to complete the discretization.

[0161] To verify the practicality of the proposed method, a series of wear tests were conducted on the bearings on an accelerated life testing platform. Wear particles were sampled using an abrasive image sensor to acquire abrasive images. After image processing and feature extraction, three types of indicators—wear rate, wear degree, and wear mechanism—were obtained and fused to evaluate the bearing's wear stage. Furthermore, the combined information from these three indicators was used to reveal the key wear mechanisms at different wear stages.

[0162] This study is as follows Figure 2 Bearing roller wear experiments were conducted on the bearing accelerated life test platform shown. The test bearing was an XTAB-A7204AC angular contact ball bearing, and its specific parameters are shown in Table 1. The test platform was driven by a variable frequency speed-regulating motor, and a stable unidirectional force load was applied to the test bearing in the radial direction via a hydraulic cylinder. To provide necessary lubrication and monitor the wear condition in real time, an oil bath lubrication system was installed at the test bearing, with the lubricating oil flow rate stably controlled at 120 ml / min. Simultaneously, an abrasive image sensor was integrated into this lubrication circuit to capture abrasive images in the lubricating oil in real time at 10-minute sampling intervals.

[0163] Table 1: Bearing Accelerated Life Test Parameters

[0164]

[0165] In order to obtain the wear stage evolution law of the bearing in a short period of time, accelerated full life test of bearing is carried out to simulate the long-term service process of bearing under actual working conditions. By applying higher loads and speeds, the failure process of bearing is accelerated.

[0166] Before starting the experiment, the test bearing was cleaned and installed on the test platform, ensuring proper alignment of the force loading module. The variable frequency speed control motor was started, and the speed was gradually increased to the target speed in stages. Then, the radial load was gradually applied to the set value via the hydraulic cylinder. After the speed and load were maintained stably for 5 minutes, the start time of the experiment was officially recorded (t=0), and the abrasive image sensor and vibration sensor were simultaneously activated. Throughout the experiment, the speed and radial load remained constant, and the lubricating oil bath system continuously provided stable lubrication. When the vibration signal amplitude measured by the vibration acceleration sensor mounted on the bearing housing continuously exceeded... hour( If the maximum amplitude of the bearing during stable operation is reached, the bearing can be considered to have failed, and the test should be terminated.

[0167] The experiment lasted for 482 minutes, and 48 sets of abrasive images were collected. After image processing, the abrasive features were extracted as shown in Table 2.

[0168] Table 2: Abrasive characteristics

[0169]

[0170] Based on the three-level discrimination tree model established above, abrasive particles are classified into normal abrasive particles, cutting abrasive particles, oxidizing abrasive particles, and fatigue abrasive particles according to their shape characteristics. The proportions of cutting abrasive particles, oxidizing abrasive particles, and fatigue abrasive particles are used as indicators of wear mechanism, IPCA as the wear rate indicator, and the proportion of large abrasive particles as the wear degree indicator. The changes in the five types of characteristics during the wear process are as follows: Figure 5 As shown, the wear rate and wear degree generally follow the same trend, but in the later stages, the wear degree increases in a stepwise manner, better reflecting the changes in the severity of wear. Furthermore, the three types of wear mechanisms exhibit different trends, effectively demonstrating the mechanistic changes during the wear process.

[0171] Based on the Mean-shift clustering algorithm, the five features are divided into stages, and an objective weighting method is applied to fuse the division results of the five features, resulting in the final stage division result as follows: Figure 5 As shown.

[0172] As can be seen, the entire wear process can be divided into six stages, corresponding to the constantly changing wear mechanism throughout the operation. Stage 1 (0-30 min) is the break-in wear stage, where the values ​​of all five characteristics are relatively high. High-speed meshing and the cutting action of hard inclusions dominate, generating a large number of abrasive grains. Simultaneously, the surface lubricating film has not yet fully formed, leading to a rapid increase in abrasive grain concentration followed by a gradual plateauing. This typically reflects the combined process of surface micro-polishing and lubrication layer construction during the break-in period. Stage 2 (40-280 min) is the stable wear stage, where the five characteristics decrease overall and tend to stabilize. This indicates that the surface lubricating film has been basically established, and the friction pair at the contact interface is mainly characterized by slight abrasive wear and micro-particle scratches. Stage 3 (280-290 min) is the oxidative wear stage, where the proportion of oxidized abrasive grains shows a significant peak, and the other four characteristics also show a slight increase. This phenomenon can be attributed to the accumulation of temperature at the friction interface or local lubrication failure, leading to the formation and peeling of the surface oxide film, thus resulting in oxidative wear as the dominant process, marking a sudden change in mechanism from the stable wear period to the accelerated wear period; the fourth stage (300-370 min) is a transitional stage or the initial stage of acceleration, in which the proportion of oxidized abrasive particles and other mechanistic characteristics temporarily drop to a low level, possibly due to the removal and cleaning of surface oxidation products and local repair of the lubricating film, but at the same time, it indicates that the accumulation of secondary surface damage caused by fatigue and cutting action begins to appear, laying the foundation for subsequent accelerated wear; the fifth stage The 380–450 min stage is the accelerated wear stage, where the wear rate and wear degree rise rapidly. The proportion of fatigue and oxide abrasive particles increases significantly, reflecting the coupled effect of multiple mechanisms of fatigue crack propagation and repeated oxide film peeling. At this stage, the microscopic pits and crack networks on the material surface intensify, and the wear rate of the equipment further increases. The 6th stage (470–480 min) is the failure stage, where various characteristics rise sharply: the proportion of fatigue and oxide abrasive particles reaches its highest point, and the proportion of cutting abrasive particles also increases significantly but is relatively small, reflecting the complete failure of the lubrication system and large-scale surface delamination. At this point, the wear enters an uncontrolled state, indicating that the equipment or component is approaching its limit of life and must be stopped immediately for maintenance or replacement to avoid more serious system failures.

[0173] If only wear rate or wear degree is used for stage division, the Mean-shift clustering result is as follows: Figure 6 and Figure 7 As shown.

[0174] pass Figure 7As can be seen, the wear rate is divided into three stages: 0-40 min and 370-460 min are stage 1, 50-360 min is stage 2, and 470-480 min is stage 3. According to the bathtub curve, the first 40 min should be the break-in period, accompanied by the formation of a surface lubricating layer and accelerated polishing of the material surface micro-interlocking, during which the wear rate will have a brief peak. However, since the wear rate in the accelerated period (370-450 min) is comparable to that in the break-in period, data from different periods are classified into the same stage. This is because in the accelerated period, due to sudden fracture or lubrication failure, the wear rate will gradually increase, reaching a similar magnitude to that in the break-in period. The wear rate index only focuses on the wear rate result; simply looking at the numerical value of the wear rate makes it difficult to distinguish the differences in mechanism, failing to reflect the different reasons for the higher wear rate in the break-in and accelerated periods, resulting in overlapping characteristics of the wear rate stages.

[0175] pass Figure 8 As can be seen, the wear level is divided into three stages: 0-30 min and 420-450 min are stage 1, 40-420 min is stage 2, and 460-480 min is stage 3. Similar to the wear rate index, the 420-450 min period, which should be the acceleration stage, is classified as part of the break-in period. This is because the wear level is also a result of the wear mechanism. During the break-in period, during the initial surface engagement and lubrication film establishment, occasional microscopic surface chipping or local hard point detachment can generate a large number of large abrasive particles in a short period. During the acceleration period, large abrasive particles are continuously generated due to lubrication failure, material fracture, etc. This results in the instantaneous proportion of large abrasive particles in the break-in and acceleration periods being close, thus placing them in the same stage.

[0176] Since both wear rate and wear degree are results mapped by wear mechanism, the stage division based on these two indicators may also lead to misjudgment due to overlapping stage characteristics.

[0177] As can be seen, the introduction of wear mechanism is necessary in the classification of wear stages. It can not only reveal the essential differences in material failure at different stages from a microscopic level, but also effectively make up for the "collision value" blind spot brought about by a single macroscopic indicator. By dynamically monitoring and quantitatively analyzing the main wear mechanism indicators such as cutting, fatigue, and oxidation, a multidimensional feature space can be constructed, thereby fully reflecting the intrinsic mechanism attributes of each stage in the clustering model and improving the accuracy and stability of stage identification.

[0178] During the break-in period (0–30 min), the weights of the five characteristic categories were equal, indicating that multiple mechanisms, including cutting, fatigue, and oxidation, contributed equally to surface damage during the initial polishing and lubrication film formation. After entering the stable period (40–280 min), the weight of PQ-O (the proportion of oxidized abrasive particles) rapidly increased to 60%–73%, while the weights of the other four characteristics remained at a low level of 7%–10%, indicating that the oxidation mechanism dominated the stable wear stage. During the transition period (280–300 min), the PQ-O weight still played a dominant role, indicating that although the effects of brief cutting and fatigue began to appear, the oxidation process still occupied a major position, providing a precursor to the subsequent mechanism abrupt change. Subsequently, during the accelerated wear period (300–450 min), the PQ-O weight gradually decreased, while the weights of IPCA and PQ-F simultaneously rebounded to the range of 20%–50%. It also reached 34%, corresponding to the combined aggravation of fatigue crack propagation effect; by the failure period (after 450 min), IPCA, The PQ-F and PQ-C values ​​all showed secondary peaks, while the PQ-O weight dropped back to 16%–26%, indicating the synergistic outbreak of multiple mechanisms during the macroscopic damage stage. This evolution fully demonstrates the crucial role of wear mechanism indicators in different stage classifications, providing an indispensable physical basis for high-precision identification of wear stages.

[0179] The introduction of wear mechanism indices not only improves the accuracy of wear stage classification but also reveals the microscopic failure mechanisms and evolution patterns in the wear process. For example, by real-time monitoring of the instantaneous distribution and changing trends of three mechanism indices—cutting wear, fatigue wear, and oxidation wear—the dominant wear type can be accurately determined.

[0180] like Figure 9 As shown, in the first 30 minutes of the experiment, the proportion of cutting abrasive particles was very high and gradually decreased, indicating that the material surface was undergoing high-speed meshing or cutting by hard inclusions. Furthermore, at this time, the proportion of fatigue abrasive particles and oxide abrasive particles was low, and the wear debris concentration increased rapidly in the early stage and then leveled off. This indicates that the material was in a typical break-in period. Figure 9 The presence of cutting abrasive grains is clearly visible in Figure (a).

[0181] Within 40–360 minutes, the wear rate remained at a low level. The degree of wear and the proportion of cutting abrasive particles fluctuated slightly but were also low. It can be considered that this was in a stable wear stage. Figure 9In (b), the abrasive coverage area is significantly reduced, and the particles are primarily small. However, between 280 and 290 min, the proportion of oxidized abrasive particles suddenly increases, indicating that oxidative wear may have occurred due to rapid formation and peeling of the surface oxide film caused by the accumulation of temperature at the friction interface or local lubrication failure. Around this time period, the wear degree, the proportion of cutting abrasive particles, and the proportion of fatigue abrasive particles all show a slight upward trend, suggesting that the cutting and fatigue wear mechanisms also begin to work synergistically, and the initiation of microcracks on the material surface and the cutting effect of hard particles gradually emerge. The combined effect of this complex mechanism indicates a transition from the stable wear stage to the accelerated wear stage, serving as an early signal of a sudden change in the wear mechanism.

[0182] Within 370–460 minutes, the wear rate increased significantly, and the proportion of fatigue abrasive particles fluctuated periodically and accumulated continuously. The degree of wear, the proportion of cutting abrasive particles, and the proportion of oxidized abrasive particles all increased to varying degrees, indicating that multiple wear mechanisms had simultaneously spiraled out of control: fatigue cracks continuously initiated and propagated under cyclic stress, hard particles continued to cut the surface and generate a larger amount of large abrasive debris, while the peeling of the high-temperature oxide film exacerbated the brittle failure of the material. At this point, obvious microscopic pits and crack networks appeared on the bearing surface, and adhesive wear and scuffing began to occur in the friction pair, marking the system entering an accelerated phase characterized by a continuous increase in wear rate and rapid accumulation of surface damage. This can be seen from… Figure 9 In (c), we see a significant increase in the abrasive coverage area and the appearance of large abrasive particles.

[0183] In the final stage, from 470 to 480 minutes, all indicators rise sharply: the wear rate reaches its peak, the proportions of cutting, fatigue, and oxidizing abrasive particles erupt simultaneously, and the wear degree breaks historical highs. It can be considered that the bearing has undergone macroscopic damage and is approaching its failure limit at this point. In this critical failure stage, the wear mechanisms further overlap, forming a chain reaction—crack propagation leads to material block detachment, large particle cutting exacerbates surface roughness, and repeated peeling of the oxide film induces new cracks. These cracks propagate rapidly under high load and high temperature environments, ultimately leading to the collapse of the structural integrity of the load-bearing component. This can be seen from... Figure 9 In (d), we can see that the abrasive coverage area increases sharply, and there are obvious fatigue-induced large abrasive particles.

[0184] In summary, by introducing and comprehensively analyzing the three major wear mechanism indicators of cutting, fatigue, and oxidation, we can not only clearly distinguish the dynamic characteristics of each stage—run-in, stabilization, accelerated wear, and severe wear—but also deeply reveal the interactive evolution of each mechanism over time: the peak of cutting abrasive grains in the initial stage and their rapid decay reflect the formation of the surface lubricating film during the run-in period; the occasional oscillations in the proportion of oxidative wear and fatigue abrasive grains in the stabilization stage indicate precursors to instability; the simultaneous fermentation of multiple mechanisms in the accelerated wear stage reveals the combined effect of crack initiation and the synergistic cutting of hard particles; and the multi-indicator explosion in the severe wear stage directly points to the critical state of macroscopic failure. This mechanism-based wear stage segmentation method not only significantly improves the accuracy and robustness of stage segmentation but also provides physical interpretability and decision support for online early warning, life prediction, and fault location.

[0185] Figure 10 A system for identifying equipment wear stages based on fusion decision-making is shown, comprising:

[0186] The feature extraction unit 1001 is configured to extract wear rate features, wear degree features, cutting index features, fatigue index features and oxidation index features from multiple abrasive sample images;

[0187] Feature clustering unit 1002 is configured to cluster wear rate features, wear degree features, cutting index features, fatigue index features and oxidation index features respectively to obtain five sets of discrete stage labels;

[0188] Data normalization unit 1003 is configured to: normalize five sets of discrete stage labels to obtain a normalized data matrix;

[0189] The weight allocation unit 1004 is configured to: calculate the weights corresponding to wear rate characteristics, wear degree characteristics, cutting index characteristics, fatigue index characteristics and oxidation index characteristics respectively using multiple methods according to the data matrix, and the average of the weights calculated by various methods is the final weight of each feature;

[0190] The comprehensive score calculation unit 1005 is configured to calculate the comprehensive score sequence of all abrasive sample images based on the final weights of each feature and the normalized data matrix.

[0191] Wear stage determination unit 1006 is configured to: perform clustering based on the comprehensive score sequence to obtain multiple optimal cluster centers, and determine the current wear stage based on the optimal cluster centers.

[0192] It is understood that the aforementioned units can be individually or entirely merged into one or more other units, or some of the units can be further divided into multiple functionally smaller units. This achieves the same operation without affecting the technical effects of the embodiments of this application. The aforementioned units are based on logical functional division. In practical applications, the function of one unit can be implemented by multiple units, or the function of multiple units can be implemented by one unit. In other embodiments of this application, the system may also include other units. In practical applications, these functions can also be implemented with the assistance of other units, and can be implemented collaboratively by multiple units.

[0193] According to another embodiment of this application, the system described in this embodiment can be constructed by running a computer program (including program code) capable of performing the steps involved in the corresponding method of the present invention on a general-purpose computing device, such as a computer, which includes processing elements and storage elements such as a central processing unit (CPU), random access memory (RAM), and read-only memory (ROM). The computer program can be recorded on, for example, a computer-readable recording medium, loaded into the aforementioned computing device through the computer-readable recording medium, and run therein.

[0194] Figure 11 A computer device is shown, which includes a processor 1101, a communication interface 1102, and a computer-readable storage medium 1103. The processor 1101, communication interface 1102, and computer-readable storage medium 1103 can be connected via a bus or other means.

[0195] The communication interface 1102 is used to receive and send data. The computer-readable storage medium 1103 can be stored in the memory of the electronic device. The computer-readable storage medium 1103 is used to store computer programs, which include program instructions. The processor 1101 is used to execute the program instructions stored in the computer-readable storage medium 1103.

[0196] The processor 1101 is the computing and control core of the electronic device. It is suitable for implementing one or more instructions, specifically for loading and executing one or more instructions to achieve the corresponding method flow or corresponding function.

[0197] The processor 1101 is configured to perform the following process:

[0198] Wear rate features, wear degree features, cutting index features, fatigue index features, and oxidation index features are extracted from multiple abrasive sample images;

[0199] Clustering was performed on wear rate characteristics, wear degree characteristics, cutting index characteristics, fatigue index characteristics, and oxidation index characteristics to obtain five groups of discrete stage labels;

[0200] The five sets of discrete stage labels are normalized to obtain a normalized data matrix;

[0201] Based on the data matrix, the weights corresponding to wear rate characteristics, wear degree characteristics, cutting index characteristics, fatigue index characteristics and oxidation index characteristics are calculated in various ways, and the average of the weights calculated in various ways is the final weight of each characteristic.

[0202] Based on the final weights of each feature and the normalized data matrix, calculate the comprehensive score sequence of all abrasive sample images.

[0203] Clustering is performed based on the comprehensive score sequence to obtain multiple optimal cluster centers, and the current wear stage is determined based on the optimal cluster centers.

[0204] This invention also provides a computer-readable storage medium, which is a memory device in an electronic device for storing programs and data. It is understood that the computer-readable storage medium here may include both built-in storage media in the electronic device and extended storage media supported by the electronic device. The computer-readable storage medium provides storage space for storing the processing system of the electronic device.

[0205] Furthermore, this storage space also contains one or more instructions suitable for loading and execution by the processor. These instructions can be one or more computer programs (including program code). It should be noted that the computer-readable storage medium here can be high-speed RAM memory or unstable memory, such as at least one disk storage device; optionally, it can also be at least one computer-readable storage medium located remotely from the aforementioned processor.

[0206] In one embodiment, the computer-readable storage medium stores one or more instructions; the processor loads and executes the one or more instructions stored in the computer-readable storage medium to perform the following process:

[0207] Wear rate features, wear degree features, cutting index features, fatigue index features, and oxidation index features are extracted from multiple abrasive sample images;

[0208] Clustering was performed on wear rate characteristics, wear degree characteristics, cutting index characteristics, fatigue index characteristics, and oxidation index characteristics to obtain five groups of discrete stage labels;

[0209] The five sets of discrete stage labels are normalized to obtain a normalized data matrix;

[0210] Based on the data matrix, the weights corresponding to wear rate characteristics, wear degree characteristics, cutting index characteristics, fatigue index characteristics and oxidation index characteristics are calculated in various ways, and the average of the weights calculated in various ways is the final weight of each characteristic.

[0211] Based on the final weights of each feature and the normalized data matrix, calculate the comprehensive score sequence of all abrasive sample images.

[0212] Clustering is performed based on the comprehensive score sequence to obtain multiple optimal cluster centers, and the current wear stage is determined based on the optimal cluster centers.

[0213] The present invention also provides a computer program product or computer program comprising computer instructions stored in a computer-readable storage medium. A processor of an electronic device reads the computer instructions from the computer-readable storage medium and executes the computer instructions, causing the electronic device to perform the following process:

[0214] Wear rate features, wear degree features, cutting index features, fatigue index features, and oxidation index features are extracted from multiple abrasive sample images;

[0215] Clustering was performed on wear rate characteristics, wear degree characteristics, cutting index characteristics, fatigue index characteristics, and oxidation index characteristics to obtain five groups of discrete stage labels;

[0216] The five sets of discrete stage labels are normalized to obtain a normalized data matrix;

[0217] Based on the data matrix, the weights corresponding to wear rate characteristics, wear degree characteristics, cutting index characteristics, fatigue index characteristics and oxidation index characteristics are calculated in various ways, and the average of the weights calculated in various ways is the final weight of each characteristic.

[0218] Based on the final weights of each feature and the normalized data matrix, calculate the comprehensive score sequence of all abrasive sample images.

[0219] Clustering is performed based on the comprehensive score sequence to obtain multiple optimal cluster centers, and the current wear stage is determined based on the optimal cluster centers.

[0220] Those skilled in the art will recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed in this application can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this application.

[0221] In the above embodiments, implementation can be achieved, in whole or in part, through software, hardware, firmware, or any combination thereof. When implemented in software, it can be implemented, in whole or in part, as a computer program product. A computer program product includes one or more computer instructions. When the computer program instructions are loaded and executed on a computer, all or part of the flow or function according to the embodiments of this application is generated. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The computer instructions can be stored in or transmitted through a computer-readable storage medium. The computer instructions can be transmitted from one website, computer, server, or data center to another website, computer, server, or data center via wired (e.g., coaxial cable, fiber optic cable, digital cable) or wireless (e.g., infrared, wireless, microwave, etc.). The computer-readable storage medium can be any available medium that a computer can access or a data processing device such as a server or data center that integrates one or more available media. The available medium can be a magnetic medium (e.g., floppy disk, hard disk, magnetic tape), an optical medium (e.g., DVD), or a semiconductor medium (e.g., solid-state drive), etc.

[0222] The above description is merely a preferred embodiment of the present invention and is not intended to limit the invention. Various modifications and variations can be made to the present invention by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. A method for identifying equipment wear stages based on fusion decision-making, characterized in that, Includes the following processes: Wear rate features, wear degree features, cutting index features, fatigue index features, and oxidation index features are extracted from multiple abrasive sample images; Clustering was performed on wear rate characteristics, wear degree characteristics, cutting index characteristics, fatigue index characteristics, and oxidation index characteristics to obtain five groups of discrete stage labels; The five sets of discrete stage labels are normalized to obtain a normalized data matrix; Based on the data matrix, the weights corresponding to wear rate characteristics, wear degree characteristics, cutting index characteristics, fatigue index characteristics and oxidation index characteristics are calculated in various ways, and the average of the weights calculated in various ways is the final weight of each characteristic. Based on the final weights of each feature and the normalized data matrix, calculate the comprehensive score sequence of all abrasive sample images. Clustering is performed based on the comprehensive score sequence to obtain multiple optimal cluster centers, and the current wear stage is determined based on the optimal cluster centers; The wear rate characteristic is the actual area of ​​the abrasive region; the wear degree characteristic is the number of abrasive grains whose equivalent circle diameter exceeds a set threshold; the cutting index characteristic is the proportion of cutting abrasive grains; the fatigue index characteristic is the proportion of fatigue abrasive grains; and the oxidation index characteristic is the proportion of oxidized abrasive grains. The weights corresponding to wear rate characteristics, wear degree characteristics, cutting index characteristics, fatigue index characteristics, and oxidation index characteristics were calculated using the entropy weight method, CRITIC method, and coefficient of variation method, respectively. The final weight of the wear rate feature is the average of the entropy weight, CRITIC weight, and coefficient of variation weight of the wear rate feature; The final weight of the wear degree feature is the average of the entropy weight, CRITIC weight, and coefficient of variation weight of the wear degree feature; The final weight of the cutting index feature is the average of the entropy weight, CRITIC weight, and coefficient of variation weight of the cutting index feature; The final weight of the fatigue index feature is the average of the entropy weight, CRITIC weight, and coefficient of variation weight of the fatigue index feature. The final weight of the oxidation index feature is the average of the entropy weight, CRITIC weight, and coefficient of variation weight of the oxidation index feature. K-means clustering is performed based on the comprehensive score sequence to calculate the optimal number of clusters K, resulting in K cluster centers. The cluster centers are sorted, and adjacent center points are taken as stage boundaries to obtain the current wear stage. The current wear stages include: break-in wear stage, stable wear stage, oxidative wear stage, transition stage, accelerated wear stage, and failure stage.

2. The equipment wear stage identification method based on fusion decision-making as described in claim 1, characterized in that, Before extracting wear rate features, wear degree features, cutting index features, fatigue index features, and oxidation index features from multiple abrasive sample images, a preprocessing process is also included, including: The abrasive sample images were sequentially processed by grayscale, background reduction, binarization, and morphological processing.

3. The equipment wear stage identification method based on fusion decision-making as described in claim 1, characterized in that, Calculate the area, equivalent circle diameter, aspect ratio, roundness, and standard deviation of curvature for each abrasive grain; When the equivalent circle diameter is greater than or equal to the first set threshold and the roundness is greater than or equal to the second set threshold, it is a normal abrasive grain; otherwise, proceed to the next step of judgment. When the standard deviation of curvature is less than or equal to the third set threshold and the roundness is greater than or equal to the fourth set threshold, it is an oxidized abrasive particle; otherwise, proceed to the next step of judgment. When the aspect ratio is greater than or equal to the fifth set threshold and the roundness is less than or equal to the second set threshold, it is a cutting abrasive; otherwise, it is a fatigue sliding abrasive.

4. The equipment wear stage identification method based on fusion decision-making as described in claim 1, characterized in that, The wear rate characteristics, wear degree characteristics, cutting index characteristics, fatigue index characteristics, and oxidation index characteristics were clustered using the Mean-shift algorithm to classify the wear stages.

5. A system for identifying equipment wear stages based on fusion decision-making, characterized in that, The equipment wear stage identification method based on fusion decision-making as described in any one of claims 1-4 includes: The feature extraction unit is configured to extract wear rate features, wear degree features, cutting index features, fatigue index features, and oxidation index features from multiple abrasive sample images; The feature clustering unit is configured to cluster the wear rate feature, wear degree feature, cutting index feature, fatigue index feature and oxidation index feature respectively to obtain five sets of discrete stage labels; The data normalization unit is configured to normalize five sets of discrete stage labels to obtain a normalized data matrix. The weight allocation unit is configured to: calculate the weights corresponding to wear rate characteristics, wear degree characteristics, cutting index characteristics, fatigue index characteristics and oxidation index characteristics respectively using multiple methods according to the data matrix, and the average of the weights calculated by various methods is the final weight of each feature; The comprehensive score calculation unit is configured to calculate the comprehensive score sequence of all abrasive sample images based on the final weights of each feature and the normalized data matrix. The wear stage determination unit is configured to: perform clustering based on the comprehensive score sequence to obtain multiple optimal cluster centers, and determine the current wear stage based on the optimal cluster centers.

6. A computer device, characterized in that, include: Processor and computer-readable storage media; A processor, adapted to execute computer programs; A computer-readable storage medium storing a computer program, which, when executed by the processor, implements the equipment wear stage identification method based on fusion decision as described in any one of claims 1 to 4.

7. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a computer program adapted to be loaded by a processor and executed as described in any one of claims 1 to 4, for the equipment wear stage identification method based on fusion decision-making.

Images