Fault diagnosis method and system for ultra-high molecular weight polyethylene crushing device

Abstract

The invention relates to the technical field of cutter damage diagnosis, and discloses a fault diagnosis method and system for an ultra-high molecular weight polyethylene crushing device, and the method comprises the steps: obtaining a cutter impact contact image, recognizing an indentation region, extracting an indentation depth residual value and a rebound recovery height, and calculating an indentation rebound factor; collecting a material particle size distribution map, extracting a particle size main peak drift value, a perturbation particle proportion and a distribution full width at half maximum value, and constructing a particle size disturbance factor; performing difference analysis on the indentation springback factors and the particle size disturbance factors of the current batch and the adjacent batch, and weighting to construct an interference dynamic increment factor; and calculating an impact response deviation value based on the factor change trend, and judging whether the tool has a damage trend or a critical fault state or not. According to the invention, the tool wear trend and the critical fault state are identified and diagnosed.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of tool damage diagnosis, in particular to a fault diagnosis method and system of an ultra-high molecular weight polyethylene crushing device. BACKGROUND

[0002] Ultra-high molecular weight polyethylene is a high-performance thermoplastic engineering plastic, and its molecular weight is usually between 3 million and 10 million g / mol. In practical applications, it is usually crushed into fine powder or particles for subsequent use. However, due to its long molecular chain, it brings a series of significant material properties, but also makes its processing and crushing more difficult.

[0003] Specifically, ultra-high molecular weight polyethylene needs high-frequency shearing or impact to achieve material crushing in the process of frozen crushing due to its high molecular weight, toughness and excellent wear resistance. Current hard alloy cutters are commonly used for crushing tasks, but under long-term operation, the cutter is prone to wear, micro-cracking, dulling or edge damage. If not identified in time, it will lead to abnormal particle size, load fluctuation and even blade fracture.

[0004] Existing crushing systems rely on current, motor torque or particle size abnormalities to determine the state of the cutter, but these indicators usually lag behind the actual development of physical damage to the cutter, especially in the early stages of micro-damage, making it difficult to form an effective early warning.

[0005] In addition, the change in the state of the cutter essentially shows the impact on the shearing method of the material, and this impact mainly reflects the changes in the local deformation behavior of the impact contact area and the disturbance characteristics of the particle size on the material level. The existing system has not established a coupling model between the above physical process and the health status of the cutter, making it difficult to accurately determine whether there is a cutter failure trend in the dynamic process.

[0006] Therefore, it is urgent to establish a cutter fault diagnosis method based on the dual response characteristics of impact deformation behavior and particle size disturbance to realize early identification and multi-level judgment of damage trend. SUMMARY

[0007] In view of this, the present application provides a fault diagnosis method and system of an ultra-high molecular weight polyethylene crushing device to solve the problem of lack of early identification of cutter damage trend in the prior art.

[0008] On the one hand, the present application provides a fault diagnosis method of an ultra-high molecular weight polyethylene crushing device, comprising: obtaining an impact contact image of the cutter, identifying a pressure indentation area image of the impact contact image based on the image, and extracting a pressure indentation depth residual value and a rebound recovery height; calculating a pressure indentation rebound factor based on the pressure indentation depth residual value and the rebound recovery height; Obtain the particle size distribution spectrum of the material, extract the main peak drift value, the proportion of perturbed particles, and the full width at half maximum (FWHM) value, and construct the particle size perturbation factor based on the main peak drift value, the proportion of perturbed particles, and the FWHM value. Obtain the indentation springback factor and particle size perturbation factor of the current batch and the adjacent batches before and after it. Based on the difference between the indentation springback factor and the difference between the particle size perturbation factor of the current batch and the adjacent batches before and after it, construct the interference dynamic increment factor by weighted summation. Based on the changing trend of the interference dynamic increment factor, the impact response deviation value is calculated and compared with the interval threshold to determine whether the current tool has a damage trend or a critical fault state. The residual indentation depth is the difference between the deepest residual indentation formed in the indentation area and the depth of the unimpacted area; the rebound recovery height is the maximum value of the indentation contour edge; and the perturbation particle ratio is the proportion of particles with a particle size smaller than the particle size threshold.

[0009] Furthermore, the indentation springback factor is obtained through the following steps: Obtain an indentation area map after tool impact, identify the lowest point of the indentation contour and the highest point of the contour edge, which correspond to the residual indentation depth and the springback recovery height, respectively; take the sum of the residual indentation depth and the springback recovery height as the total indentation deformation value; take the ratio of the springback recovery height to the total indentation deformation value as the indentation springback factor.

[0010] Furthermore, the particle size perturbation factor is obtained through the following steps: Extract the particle size distribution map within the current sampling period, identify the position of the main peak of particle size and compare it with the position of the main peak in the previous sampling period, and calculate the main peak drift value; The proportion of particles with a diameter smaller than a set particle size threshold to the total number of particles is calculated as the perturbation particle proportion. Calculate the width difference of the main peak at 0.5 times the maximum height in the current sampling period, and use it as the full width at half maximum (FWHM). The particle size perturbation factor is obtained by weighting and summing the main peak drift value, the proportion of perturbed particles, and the half-width at half-maximum (WHM) of the distribution according to preset weights.

[0011] Furthermore, the interference dynamic increment factor is obtained through the following steps: Obtain the indentation springback factor and particle size perturbation factor for the current batch, the previous batch, and the next batch; calculate the difference in indentation springback factor and particle size perturbation factor between the current batch and the two adjacent batches; based on the difference in indentation springback factor and particle size perturbation factor, calculate the average rate of change of the indentation springback difference and the coefficient of variation of the particle size perturbation difference; and sum the average rate of change and the coefficient of variation by weight to obtain the interference dynamic increment factor. Calculate the mean and standard deviation of the particle size perturbation factor difference, where the coefficient of variation is the ratio of the absolute value of the standard deviation to the mean.

[0012] Furthermore, the shock response deviation value is obtained through the following steps: Based on the changing trend of the dynamic increment factor of the interference in three consecutive batches, the slope of the change was obtained by fitting. The change slope is weighted and combined with the latest interference dynamic increment factor to determine the shock response deviation value.

[0013] Furthermore, the impact response deviation value is compared with a preset first threshold and a second threshold. If it is lower than the first threshold, it is determined to be in a normal tool state; if it is higher than or equal to the first threshold and lower than the second threshold, it is determined to be in a damage trend; if it is higher than or equal to the second threshold, it is determined to be in a critical fault state, and an early warning is issued for the current tool.

[0014] Furthermore, when it is determined that there is a damage trend or a critical fault state, the density distribution of the center of the indentation area in the tool impact contact image is obtained, the indentation symmetry offset value is identified, and the degree of impact force eccentricity is determined based on the degree of deformation of the main peak in the current particle size distribution spectrum. If the impact force eccentricity exceeds the eccentricity judgment threshold, it is diagnosed as tool structure imbalance type damage; otherwise, it is diagnosed as cutting edge notch type damage.

[0015] Furthermore, the calculation process for the indentation symmetry offset value includes: The indentation area is divided into left and right symmetrical regions along the main axis, and the pixel gray-scale centroid position in each region is calculated. The distance between the two gray-scale centroid positions is used as the indentation symmetry offset value; If the offset value is greater than the symmetry offset reference value, it is determined that there is an eccentric impact trend.

[0016] Furthermore, after determining that the current tool has a damage trend or a critical fault state, the main peak drift value of particle size, the proportion of perturbation particles, the half width at half maximum value of distribution, the rate of change of indentation springback difference and the coefficient of variation of particle size perturbation factor are stored as historical data. Based on historical data and corresponding diagnostic results, the information gain analysis method is used to calculate the contribution of each weighted term to the diagnostic results, and the information gain value is normalized to adjust the weight coefficient of the particle size perturbation factor. The discriminative ability score of each weighted term is calculated based on the incremental statistical model, and the discriminative ability score is normalized to adjust the weight coefficient of the dynamic incremental factor of the interference.

[0017] Compared with the prior art, the beneficial effects of the present invention are as follows: The fault diagnosis method for an ultra-high molecular weight polyethylene (UHMWPE) pulverizing device provided by this invention enables early identification of tool damage trends. By acquiring tool impact contact images, extracting the residual indentation depth and rebound recovery height, and calculating the resulting indentation rebound factor, microscopic deformation characteristics can be obtained before physical damage significantly affects the equipment load. Simultaneously, a particle size perturbation factor is constructed by combining the material's main particle size drift value, the proportion of perturbed particles, and the half-width at half-maximum (WHM) of the distribution. This reflects changes in the tool's shearing performance from the perspective of the material's discharge state. Furthermore, by analyzing the difference between the indentation rebound factor and the particle size perturbation factor, an interference dynamic increment factor is established, effectively improving the perceptibility and diagnostic sensitivity of fault trends. By calculating the changing trend of the interference dynamic increment factor to obtain the impact response deviation value and setting multi-level interval thresholds, it is possible to determine whether the tool is currently in a normal, damage trend, or critical fault state, achieving hierarchical state diagnosis. Based on this, the impact force eccentricity is determined by combining the indentation symmetry shift and the particle size main peak deformation characteristics, further distinguishing between tool structural imbalance damage and blade notch damage, improving diagnostic accuracy and type identification capabilities. After diagnosis, the system stores key feature parameters and results as historical data, and optimizes the weighting coefficients of each weighting term based on information gain and discriminative ability scores, achieving adaptive model adjustment and dynamic learning, thus possessing robustness and practicality under long-term operation. Compared with existing hysteresis-based diagnostic methods that rely solely on current, voltage, or particle size anomalies, this invention can directly sense changes in impact behavior and supplement them with material disturbance response. The diagnostic process is more physically relevant and has a proactive response, improving the timeliness and accuracy of fault warning.

[0018] On the other hand, the present invention proposes a fault diagnosis system for an ultra-high molecular weight polyethylene pulverizing device, comprising: The image recognition module is configured to acquire the impact contact image of the tool, obtain the indentation area map of the impact contact image based on the image recognition, and extract the residual value of the indentation depth and the springback recovery height. The springback factor module is configured to calculate the indentation springback factor based on the residual value of the indentation depth and the springback recovery height. The particle size feature module is configured to acquire the particle size distribution map of the material, extract the main peak drift value, the proportion of perturbed particles, and the full width at half maximum (FWHM) value, and construct the particle size perturbation factor based on the main peak drift value, the proportion of perturbed particles, and the FWHM value. The difference analysis module is configured to obtain the indentation rebound factor and particle size perturbation factor of the current batch and the adjacent batches before and after it. Based on the difference between the indentation rebound factor and the difference between the particle size perturbation factor of the current batch and the adjacent batches before and after it, the interference dynamic increment factor is constructed by weighted summation. The status judgment module is configured to calculate the impact response deviation value based on the changing trend of the interference dynamic increment factor, and compare it with the interval threshold to determine whether the current tool has a damage trend or a critical fault state.

[0019] It should be noted that the fault diagnosis method and system of the ultra-high molecular weight polyethylene pulverizing device of the present invention have the same beneficial effects, and will not be described in detail here. Attached Figure Description

[0020] Various other advantages and benefits will become apparent to those skilled in the art upon reading the following detailed description of preferred embodiments. The accompanying drawings are for illustrative purposes only and are not intended to limit the invention. Furthermore, the same reference numerals denote the same parts throughout the drawings. In the drawings: Figure 1 A flowchart of a fault diagnosis method for an ultra-high molecular weight polyethylene pulverizing device provided in an embodiment of the present invention.

[0021] Figure 2 A functional block diagram of the fault diagnosis method and system for the ultra-high molecular weight polyethylene pulverizing device provided in the embodiments of the present invention. Detailed Implementation

[0022] Exemplary embodiments of the present application will now be described in more detail with reference to the accompanying drawings. While exemplary embodiments of the present disclosure are shown in the drawings, it should be understood that the present disclosure may be implemented in various forms and should not be limited to the embodiments set forth herein. Rather, these embodiments are provided to enable a more thorough understanding of the present disclosure and to fully convey the scope of the disclosure to those skilled in the art. It should be noted that, unless otherwise specified, the embodiments and features described herein can be combined with each other. The present invention will now be described in detail with reference to the accompanying drawings and embodiments.

[0023] See Figure 1 As shown, this embodiment of the invention provides a fault diagnosis method for an ultra-high molecular weight polyethylene (UHMWPE) pulverizing device, including: Obtain the impact contact image of the tool, obtain the indentation area map of the impact contact image based on image recognition, and extract the residual value of indentation depth and springback recovery height; The indentation rebound factor is calculated based on the residual value of indentation depth and the rebound recovery height. Obtain the particle size distribution map of the material, extract the main peak drift value, the proportion of perturbed particles, and the full width at half maximum (FWHM) value, and construct the particle size perturbation factor based on the main peak drift value, the proportion of perturbed particles, and the FWHM value. Obtain the indentation springback factor and particle size perturbation factor of the current batch and the adjacent batches before and after it. Based on the difference between the indentation springback factor and the difference between the particle size perturbation factor of the current batch and the adjacent batches before and after it, construct the interference dynamic increment factor by weighted summation. Based on the changing trend of the interference dynamic increment factor, the impact response deviation value is calculated and compared with the interval threshold to determine whether the current tool has a damage trend or a critical fault state. The residual indentation depth is the difference between the deepest residual indentation formed in the indentation area and the depth of the unimpacted area; the rebound recovery height is the maximum value at the edge of the indentation profile; the perturbation particle ratio is the proportion of particles with a particle size smaller than the particle size threshold.

[0024] It should be noted that the residual indentation depth is defined as the vertical depth from the lowest point of the indentation area to the reference surface; the springback recovery height is defined as the height difference between the highest point of the indentation edge and the reference surface; the reference surface refers to the plane when the tool is not in contact with the material.

[0025] The fault diagnosis method in this embodiment is based on the coupling mechanism between the mechanical deformation behavior during physical impact and the material particle size disturbance response. It constructs a technical path for judging the damage state of the blades in an ultra-high molecular weight polyethylene crushing device. Its principle can be elaborated from the following aspects: First, the impact process between a cutting tool and material is essentially a high-speed, transient contact deformation process. When a cutting tool impacts ultra-high molecular weight polyethylene (UHMWPE) after it has become embrittled at low temperatures, a local indentation zone is formed on the material surface. Under normal conditions, the cutting tool has good sharpness and a balanced structure, and the indentation formed by its impact is regular, exhibiting deep indentation and strong edge springback recovery. However, when the cutting tool suffers physical damage such as wear, microcracks, or notches, the impact force transmission path changes, resulting in insufficient indentation at the center of the indentation or reduced edge springback, leading to deformation of the indentation contour. Therefore, by extracting the residual indentation depth and springback recovery height through image recognition, the degradation trend of the cutting tool's impact capability can be effectively characterized.

[0026] The residual indentation depth is defined as the vertical distance between the lowest point of the indentation area and the unimpacted reference surface, reflecting the residual depth of plastic deformation under impact load. The springback recovery height is defined as the vertical height of the highest point of the indentation edge relative to the reference surface, reflecting the elastic recovery capability of the local material. Both together constitute the total indentation deformation value. The ratio of the springback recovery height to the total deformation value is defined as the indentation springback factor. A higher factor indicates stronger elastic recovery capability and more concentrated impact transmission. If this factor continuously decreases in adjacent batches, it indicates a decrease in tool impact efficiency and a damage trend.

[0027] Secondly, changes in the condition of the cutting tools affect the overall particle size distribution characteristics after material crushing. Sharp, intact cutting tools can produce a uniform particle size distribution, while worn tools can lead to abnormal particle size, increased fine powder, or main peak drift. Therefore, three key indicators were identified through particle size distribution analysis: the main peak drift value reflects the systematic shift within the main particle size range; the proportion of perturbed particles reflects the abnormal increase in small-diameter powder; and the full width at half maximum (FWHM) characterizes the extent of particle size distribution expansion. These three indicators are weighted and synthesized to form a particle size perturbation factor, which serves as an output response signal characterizing changes in the cutting tool's crushing performance.

[0028] Furthermore, to enhance the monitoring capability of the dynamic evolution process, the indentation springback factor and grain size perturbation factor of the current batch are compared with those of its preceding and following batches. A difference sequence is calculated, and dynamic change rate and fluctuation index are constructed. The average change rate of the indentation springback factor difference reflects the rate of change in the tool impact state; the coefficient of variation of the grain size perturbation factor difference characterizes the instability of the grain size response fluctuation. These two indices are weighted to form an interference dynamic increment factor; the larger the value, the more drastic the change in tool state.

[0029] Finally, the evolution trajectory of multiple batch interference dynamic incremental factors is fitted to obtain the slope of the changing trend, which, together with the current batch factor value, forms the impact response deviation value, used to determine whether the current tool is showing a damage trend or a critical failure. By setting two threshold intervals, such as a safety threshold and a risk threshold, a three-stage determination of the tool's health status can be achieved: normal, warning, and failure.

[0030] In some embodiments of this application, the indentation springback factor is obtained through the following steps: Obtain an image of the indentation area after tool impact, identify the lowest point of the indentation contour and the highest point of the contour edge, which correspond to the residual indentation depth and the springback recovery height, respectively; take the sum of the residual indentation depth and the springback recovery height as the total indentation deformation value; take the ratio of the springback recovery height to the total indentation deformation value as the indentation springback factor.

[0031] It should be noted that the indentation springback factor is obtained through the following steps: An indentation area image after tool impact is obtained; preferably, the impact contact image is grayscaled and edge extracted using image preprocessing methods; the indentation contour is located using a region contour recognition algorithm; the center point of the indentation located at the lowest grayscale value in the image within the contour is identified as the measurement benchmark for the residual indentation depth; the highest grayscale point on the contour edge is identified as the measurement benchmark for the springback recovery height; both the residual indentation depth and the springback recovery height are calculated using a reference plane fitted to the unimpacted area outside the indentation region as the reference plane; the sum of the residual indentation depth and the springback recovery height is taken as the total indentation deformation value, and the ratio of the springback recovery height to the total indentation deformation value is taken as the indentation springback factor, reflecting the relative proportion of residual springback characteristics after impact; when the total indentation deformation value is less than a preset minimum deformation threshold, to avoid interference caused by image misidentification or edge abnormalities, the current sample is not included in the springback factor calculation process.

[0032] In some embodiments of this application, the particle size perturbation factor is obtained through the following steps: Extract the particle size distribution map within the current sampling period, identify the position of the main peak of particle size and compare it with the position of the main peak in the previous sampling period, and calculate the main peak drift value; The proportion of particles with a diameter smaller than a set particle size threshold to the total number of particles is calculated as the perturbation particle proportion. Calculate the width difference of the main peak at 0.5 times the maximum height in the current sampling period, and use it as the full width at half maximum (FWHM). The particle size perturbation factor is obtained by weighting and summing the main peak drift value, the proportion of perturbed particles, and the half-width at half-maximum (WHM) of the distribution according to preset weights.

[0033] It should be noted that the particle size perturbation factor is obtained through the following steps: First, extract the particle size distribution map within the current sampling period, preferably using laser particle size analysis or image-based particle size statistics to obtain the frequency distribution curve of the material discharge particle size within the set particle size range; identify the position of the main peak in the distribution curve and calculate the position difference with the main peak position in the previous sampling period to obtain the main peak drift value, reflecting the overall shift trend of the particle size distribution; further, calculate the ratio of the number of particles with a particle size smaller than the set particle size threshold to the total number of particles in the current period, as the perturbation particle ratio. For example, the particle size distribution curve is used to reflect the proportion of abnormally fine powder or broken particles. Then, the width range corresponding to 0.5 times the height of the main peak in the current period is measured, and the difference is calculated to obtain the full width at half maximum (FWHM) of the main peak, which is used to assess the concentration and diffusion trend of the particle size distribution. Finally, the main peak drift value, the proportion of perturbed particles, and the FWHM of the distribution are multiplied by their respective preset weighting coefficients and summed. The resulting weighted sum is the particle size perturbation factor corresponding to the current sampling period, used to characterize the sensitivity and interference degree of the material particle size response to changes in the tool state. In this embodiment, the initial weighting coefficient for the particle size main peak drift value is 0.5; the initial weighting coefficient for the proportion of perturbed particles is 0.3; and the initial weighting coefficient for the FWHM of the distribution is 0.2.

[0034] In some embodiments of this application, the interference dynamic increment factor is obtained through the following steps: Obtain the indentation springback factor and particle size perturbation factor for the current batch, the previous batch, and the next batch. Calculate the difference in indentation springback factor and particle size perturbation factor between the current batch and the two adjacent batches. Based on the difference in indentation springback factor and particle size perturbation factor, calculate the average rate of change of the indentation springback difference and the coefficient of variation of the particle size perturbation difference. Sum the average rate of change and the coefficient of variation using weighted averages to obtain the interference dynamic increment factor. Calculate the mean and standard deviation of the particle size perturbation factor difference. The coefficient of variation is the ratio of the absolute value of the standard deviation to the mean.

[0035] In this embodiment, the acquisition of the interference dynamic increment factor not only includes the extraction of the amplitude changes of the difference between the indentation springback factor and the difference between the particle size disturbance factor, but also further introduces the coupling relationship between the trend consistency between batches and the relative disturbance intensity, so as to achieve sensitive response and dynamic evolution evaluation of the current tool state disturbance.

[0036] Specifically, after obtaining the indentation springback factor and particle size disturbance factor of the current batch, the previous batch, and the next batch, the difference in the indentation springback factor of the current batch compared to the previous batch and the next batch is calculated. The arithmetic mean of the changes in the two sets of differences is divided by the normalized rate of change formed by the previous difference to obtain the average rate of change of the indentation springback difference, which represents the dynamic offset speed of the tool contact deformation capability. At the same time, the difference sequence between the particle size disturbance factor of the current batch and two adjacent batches is obtained, and the ratio of the absolute value of the standard deviation to the absolute value of the mean of the difference sequence is calculated as the coefficient of variation of the particle size disturbance difference, which is used to characterize the dispersion and fluctuation risk of the output particle size response.

[0037] The average rate of change of the indentation rebound difference and the coefficient of variation of the particle size perturbation difference are weighted and summed according to a preset proportion to obtain the interference dynamic increment factor, which characterizes the degree of deviation and trend intensity of the current state compared with the states of the previous and subsequent batches. The initial weighting coefficient for both the average rate of change and the coefficient of variation is 0.5.

[0038] It should be noted that if the indentation springback factor and grain size perturbation factor of the current batch show little change and consistent trends compared to adjacent batches, the interference dynamic increment factor remains at a low level; if there are abrupt, jump, or inconsistent changes, the interference dynamic increment factor increases significantly, indicating that the current batch has significant deviations or potential anomalies. This factor is the core input for subsequent impact response deviation values, which is beneficial for sensitively capturing and early identifying tool damage trends. In some embodiments of this application, the shock response deviation value is obtained through the following steps: Based on the changing trend of the dynamic increment factor of the interference in three consecutive batches, the slope of the change was obtained by fitting. The deviation value of the impact response is determined by weighting the change slope with the latest interference dynamic increment factor.

[0039] In this embodiment, the acquisition of the shock response deviation value not only focuses on the current absolute value of the interference dynamic increment factor, but also introduces its dynamic change trend in multiple consecutive batches, and constructs a comprehensive evaluation index that integrates trend guidance and instantaneous offset.

[0040] Specifically, the interference dynamic increment factors of the current batch and the previous two batches are first obtained as a sequence. The slope of this sequence is then calculated using a linear fitting method to characterize the evolution rate of the tool's physical state disturbance. A larger slope indicates that the interference factors are shifting at an accelerated rate, which may correspond to a more severe or sudden damage evolution process.

[0041] Subsequently, the slope obtained from the fitting is weighted and combined with the interference dynamic increment factor of the current batch, that is, the current deviation degree and the intensity of the change trend are considered simultaneously in a certain proportion to construct the impact response deviation value. This deviation value comprehensively reflects the risk level of the current tool state in both spatial (current batch) and temporal (evolution trend) dimensions.

[0042] The initial weighting coefficient of the aforementioned slope change is 0.3, and the initial weighting coefficient of the latest interferometric dynamic increment factor is 0.7.

[0043] The resulting impact response deviation value will serve as the core indicator for judging whether the tool is in a normal state, a damage trend state, or a critical fault state. It features timely response, strong adaptability, and high discrimination accuracy, providing a reliable basis for real-time monitoring of tool status and fault early warning.

[0044] In some embodiments of this application, the impact response deviation value is compared with a preset first threshold and a second threshold. If it is lower than the first threshold, it is determined to be in a normal state of the tool; if it is higher than or equal to the first threshold and lower than the second threshold, it is determined to be in a damage trend; if it is higher than or equal to the second threshold, it is determined to be in a critical fault state, and an early warning is given to the current tool.

[0045] In this embodiment, the methods for obtaining the first threshold and the second threshold are as follows: I. Collect sample data: Extract historical data from ≥300 batches of known tool conditions. The data should include the following for each batch: impact response deviation value, corresponding tool diagnostic result label, such as normal, damage trend and critical damage.

[0046] Grouping by diagnostic results: The above batch data are divided into three categories according to the diagnostic labels: Category A: Tool is normal, marked as 0; Category B: There is a trend of damage, marked as 1; Category C: Critical failure, marked as 2.

[0047] II. Threshold Calculation: The first threshold T1 = μ1 + 1.2σ1; where μ1 is the mean of the A-class samples and σ1 is the standard deviation of the A-class samples.

[0048] The second threshold T2 = μ2 - 0.8σ2; where μ2 is the mean of the C-class samples and σ2 is the standard deviation of the C-class samples.

[0049] If, in historical annotations, the deviation values ​​of some damage trend samples overlap too much in the normal or fault range, the median of their deviation value distribution can be calculated, and T1 and T2 can be fine-tuned accordingly, so that T1 < damage trend < T2, ensuring that trend samples are distributed between the two boundaries as much as possible.

[0050] In some embodiments of this application, when it is determined that there is a damage trend or a critical fault state, the density distribution of the center of the indentation area in the tool impact contact image is obtained, the indentation symmetry offset value is identified, and the degree of impact force eccentricity is determined based on the degree of deformation of the main peak in the current particle size distribution spectrum. If the impact force eccentricity exceeds the eccentricity judgment threshold, it is diagnosed as tool structure imbalance type damage; otherwise, it is diagnosed as cutting edge notch type damage.

[0051] It should be noted that, firstly, the impact contact images of the cutting tools in the current batch are acquired, and the indentation areas in the images are extracted based on image recognition algorithms. The edge contours of the indentation areas are identified using image grayscale analysis, and the image grayscale value distribution at the center of the indentation area is extracted. Based on this, the indentation area is divided into two symmetrical regions along the principal axis, and the center position of the image grayscale distribution within each region is statistically analyzed, specifically by calculating the weighted grayscale centroid coordinates of each pixel within the region. Subsequently, the distance between the grayscale centroids of the two symmetrical regions is defined as the indentation symmetry offset value, used to measure the symmetry change in the indentation morphology.

[0052] Next, by combining the particle size distribution maps of the same batch of materials, the deformation characteristics corresponding to the main peak position of the particle size were extracted. This process includes symmetry analysis and contour fitting of the main peak shape, identifying the differences in width, slope, and height distribution on both sides of the main peak to determine whether the main peak exhibits asymmetric distortion. Based on the degree of distortion, the deformation coefficient of the main peak was calculated to characterize the sensitivity of the material particle size response to tool impact eccentricity.

[0053] Finally, the indentation symmetry shift value and the deformation coefficient of the main particle size peak are comprehensively judged. When the indentation symmetry shift value exceeds the eccentricity judgment threshold and the deformation coefficient of the main particle size peak also shows significant asymmetry, it is determined that the current impact behavior has obvious eccentricity characteristics, and the tool may have structural imbalance or axial displacement problems, and is diagnosed as "structural imbalance type damage"; if the indentation symmetry shift value is shifted but within an acceptable range, and the main particle size peak still maintains a symmetrical shape, only showing slight widening or low value drift, it is determined that the tool has unilateral edge defects or wear, and is diagnosed as "edge notch type damage".

[0054] In some embodiments of this application, the calculation process for the indentation symmetry offset value includes: The indentation area is divided into left and right symmetrical regions along the main axis, and the pixel gray-scale centroid position in each region is calculated. The distance between the two gray-scale centroid positions is used as the indentation symmetry offset value; If the offset value is greater than the symmetry offset reference value, it is determined that there is an eccentric impact trend.

[0055] It should be noted that in this embodiment, the indentation symmetry offset value is used to quantify whether the force on the tool is symmetrical during the impact contact process. Its core lies in reflecting the spatial offset of the tool's impact force through the difference in the center of gravity of the image grayscale distribution. The larger the offset value, the more unbalanced the force generated by the tool during the impact, which is highly likely due to problems such as eccentric wear, rotation axis misalignment, or edge imbalance in the tool structure. The symmetry offset reference value is set based on the statistical range of indentation symmetry offset values ​​of the tool under normal working conditions in historical image data. By analyzing a large number of image samples under normal conditions, the mean and standard deviation of the offset value distribution are extracted, and then an acceptable upper limit for the symmetry reference is defined. If the currently calculated indentation symmetry offset value exceeds the upper limit of this reference range, it indicates a significant eccentric impact trend, which should attract the system's attention and trigger further fault type identification procedures. This method uses structural symmetry features to assist in judging the tool's operating state, enabling early identification of eccentric impact behavior of the tool without relying on traditional mechanical sensors.

[0056] In some embodiments of this application, after determining that the current tool has a damage trend or a critical fault state, the particle size main peak drift value, the proportion of perturbation particles, the half width at half maximum value of the distribution, the rate of change of indentation springback difference and the coefficient of variation of particle size perturbation factor are stored as historical data. Based on historical data and corresponding diagnostic results, the information gain analysis method is used to calculate the contribution of each weighted term to the diagnostic results, and the information gain value is normalized to adjust the weight coefficient of the particle size perturbation factor. The discriminative ability score of each weighted term is calculated based on the incremental statistical model, and the discriminative ability score is normalized to adjust the weight coefficient of the dynamic incremental factor of the interference.

[0057] It should be noted that, in order to dynamically optimize the weighted structure of the particle size perturbation factor and the interference dynamic increment factor, and further improve the accuracy and sensitivity of fault identification, when the current tool is determined to have a damage trend or a critical fault state, the system will automatically store the key indicator data of the current batch as historical data. The key indicator data includes the particle size main peak drift value, the proportion of perturbation particles, the half-width at half-maximum (WHM) of the distribution, the rate of change of indentation springback difference, and the coefficient of variation of the particle size perturbation factor, and together with the diagnostic results of the current batch (including whether there is a damage trend, whether it is a critical fault state, and its specific fault type), a sample set is established.

[0058] In the subsequent optimization process, based on the accumulated sample set, information gain analysis was first performed on the three weighting terms (main peak drift value, perturbation particle ratio, and distribution half-width at half-maximum) constituting the particle size perturbation factor. Specifically, all samples were classified and statistically analyzed according to diagnostic results. The conditional entropy of each weighting term under different diagnostic states was calculated, and the difference between this entropy and the information entropy of the entire sample was used as the contribution of that indicator to the diagnostic results, i.e., the information gain value. The information gain values ​​of all weighting terms were normalized so that their sum was 1, which served as the dynamic weight of each term in the particle size perturbation factor.

[0059] Subsequently, regarding the construction process of the interferometric dynamic incremental factor, the system further analyzed the discriminative power of the average rate of change of indentation rebound difference and the coefficient of variation of particle size perturbation factor in historical data. By establishing an incremental statistical model, the discriminative power score of each weighted term under different diagnostic results was calculated. For example, methods such as analysis of variance and the ratio of between-class differences to within-class differences were used to evaluate its distinguishability of fault states. After normalizing the discriminative power score, it was used as the update weight of each term in the interferometric dynamic incremental factor.

[0060] Through the above mechanism, the weight distribution of the weighted terms in the construction of diagnostic indicators can be adjusted autonomously based on the structural patterns of existing historical data. This makes the particle size perturbation factor and interference dynamic increment factor constructed in the subsequent fault identification process closer to the actual physical response characteristics, improves the sensitivity to abnormal trends and the clarity of the judgment boundary, and ensures that stable and accurate diagnosis and judgment can be achieved under the conditions of multi-state and multi-feature fusion.

[0061] See Figure 2 As shown, this embodiment of the invention provides a fault diagnosis system for an ultra-high molecular weight polyethylene (UHMWPE) pulverizing device, comprising: The image recognition module is configured to acquire the impact contact image of the tool, obtain the indentation area map of the impact contact image based on the image recognition, and extract the residual value of the indentation depth and the springback recovery height. The springback factor module is configured to calculate the indentation springback factor based on the residual value of the indentation depth and the springback recovery height. The particle size feature module is configured to acquire the particle size distribution map of the material, extract the main peak drift value, the proportion of perturbed particles, and the full width at half maximum (FWHM) value, and construct the particle size perturbation factor based on the main peak drift value, the proportion of perturbed particles, and the FWHM value. The difference analysis module is configured to obtain the indentation rebound factor and particle size perturbation factor of the current batch and the adjacent batches before and after it. Based on the difference between the indentation rebound factor and the difference between the particle size perturbation factor of the current batch and the adjacent batches before and after it, the interference dynamic increment factor is constructed by weighted summation. The status judgment module is configured to calculate the impact response deviation value based on the changing trend of the interference dynamic increment factor, and compare it with the interval threshold to determine whether the current tool has a damage trend or a critical fault state.

[0062] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and not to limit it. Although the present invention has been described in detail with reference to the above embodiments, those skilled in the art should understand that modifications or equivalent substitutions can still be made to the specific implementation of the present invention. Any modifications or equivalent substitutions that do not depart from the spirit and scope of the present invention should be covered within the scope of protection of the claims of the present invention.

Claims

1. A fault diagnosis method for an ultra-high molecular weight polyethylene pulverizing device, characterized in that, include: Obtain the impact contact image of the tool, obtain the indentation area map of the impact contact image based on image recognition, and extract the residual value of indentation depth and springback recovery height; The indentation rebound factor is calculated based on the residual value of the indentation depth and the rebound recovery height. Obtain the particle size distribution spectrum of the material, extract the main peak drift value, the proportion of perturbed particles, and the full width at half maximum (FWHM) value, and construct the particle size perturbation factor based on the main peak drift value, the proportion of perturbed particles, and the FWHM value. Obtain the indentation springback factor and particle size perturbation factor of the current batch and the adjacent batches before and after it. Based on the difference between the indentation springback factor and the difference between the particle size perturbation factor of the current batch and the adjacent batches before and after it, construct the interference dynamic increment factor by weighted summation. Based on the changing trend of the interference dynamic increment factor, the impact response deviation value is calculated and compared with the interval threshold to determine whether the current tool has a damage trend or a critical fault state. The residual indentation depth is the difference between the deepest residual indentation formed in the indentation area and the depth of the unimpacted area; the rebound recovery height is the maximum value of the indentation contour edge; and the perturbation particle ratio is the proportion of particles with a particle size smaller than the particle size threshold.

2. The fault diagnosis method for the ultra-high molecular weight polyethylene pulverizing device according to claim 1, characterized in that, The indentation resilience factor is obtained through the following steps: Obtain an indentation area map after tool impact, identify the lowest point of the indentation contour and the highest point of the contour edge, which correspond to the residual indentation depth and the springback recovery height, respectively; take the sum of the residual indentation depth and the springback recovery height as the total indentation deformation value; take the ratio of the springback recovery height to the total indentation deformation value as the indentation springback factor.

3. The fault diagnosis method for the ultra-high molecular weight polyethylene pulverizing device according to claim 2, characterized in that, The particle size perturbation factor is obtained through the following steps: Extract the particle size distribution map within the current sampling period, identify the position of the main peak of particle size and compare it with the position of the main peak in the previous sampling period, and calculate the main peak drift value; The proportion of particles with a diameter smaller than a set particle size threshold to the total number of particles is calculated as the perturbation particle proportion. Calculate the width difference of the main peak at 0.5 times the maximum height in the current sampling period, and use it as the full width at half maximum (FWHM). The particle size perturbation factor is obtained by weighting and summing the main peak drift value, the proportion of perturbed particles, and the half-width at half-maximum (WHM) of the distribution according to preset weights.

4. The fault diagnosis method for the ultra-high molecular weight polyethylene pulverizing device according to claim 3, characterized in that, The interference dynamic increment factor is obtained through the following steps: Obtain the indentation rebound factor and particle size perturbation factor of the current batch, the previous batch, and the next batch, and calculate the difference in indentation rebound factor and particle size perturbation factor between the current batch and the two adjacent batches respectively. Based on the difference between the indentation springback factor and the difference between the particle size perturbation factor, the average rate of change of the indentation springback difference and the coefficient of variation of the particle size perturbation difference are calculated; the average rate of change and the coefficient of variation are weighted and summed to obtain the interference dynamic increment factor. Calculate the mean and standard deviation of the particle size perturbation factor difference, where the coefficient of variation is the ratio of the absolute value of the standard deviation to the mean.

5. The fault diagnosis method for the ultra-high molecular weight polyethylene pulverizing device according to claim 4, characterized in that, The shock response deviation value is obtained through the following steps: Based on the changing trend of the dynamic increment factor of the interference in three consecutive batches, the slope of the change was obtained by fitting. The change slope is weighted and combined with the latest interference dynamic increment factor to determine the shock response deviation value.

6. The fault diagnosis method for the ultra-high molecular weight polyethylene pulverizing device according to claim 5, characterized in that, The impact response deviation value is compared with a preset first threshold and a second threshold. If it is lower than the first threshold, the tool is judged to be in a normal state. If it is higher than or equal to the first threshold and lower than the second threshold, it is judged to have a damage trend. If it is higher than or equal to the second threshold, it is judged to be in a critical fault state, and an early warning is issued for the current tool.

7. The fault diagnosis method for the ultra-high molecular weight polyethylene pulverizing device according to claim 6, characterized in that, When it is determined that there is a damage trend or a critical fault state, the density distribution of the center of the indentation area in the tool impact contact image is obtained, the indentation symmetry offset value is identified, and the degree of impact force eccentricity is determined based on the degree of deformation of the main peak in the current particle size distribution spectrum. If the impact force eccentricity exceeds the eccentricity judgment threshold, it is diagnosed as tool structure imbalance type damage; otherwise, it is diagnosed as cutting edge notch type damage.

8. The fault diagnosis method for the ultra-high molecular weight polyethylene pulverizing device according to claim 7, characterized in that, The calculation process for the indentation symmetry offset value includes: The indentation area is divided into left and right symmetrical regions along the main axis, and the pixel gray-scale centroid position in each region is calculated. The distance between the two gray-scale centroid positions is used as the indentation symmetry offset value; If the offset value is greater than the symmetry offset reference value, it is determined that there is an eccentric impact trend.

9. The fault diagnosis method for the ultra-high molecular weight polyethylene pulverizing device according to claim 8, characterized in that, After determining that the current tool has a damage trend or a critical fault state, the particle size main peak drift value, the proportion of perturbation particles, the half width at half maximum value of the distribution, the change rate of indentation springback difference and the coefficient of variation of particle size perturbation factor are stored as historical data. Based on historical data and corresponding diagnostic results, the information gain analysis method is used to calculate the contribution of each weighted term to the diagnostic results, and the information gain value is normalized to adjust the weight coefficient of the particle size perturbation factor. The discriminative ability score of each weighted term is calculated based on the incremental statistical model, and the discriminative ability score is normalized to adjust the weight coefficient of the dynamic incremental factor of the interference.

10. A fault diagnosis system for an ultra-high molecular weight polyethylene (UHMWPE) pulverizing device, applied in the fault diagnosis method for the UHMWPE pulverizing device according to any one of claims 1-9, characterized in that, The system includes: The image recognition module is configured to acquire the impact contact image of the tool, obtain the indentation area map of the impact contact image based on the image recognition, and extract the residual value of the indentation depth and the springback recovery height. The springback factor module is configured to calculate the indentation springback factor based on the residual value of the indentation depth and the springback recovery height. The particle size feature module is configured to acquire the particle size distribution map of the material, extract the main peak drift value, the proportion of perturbed particles, and the full width at half maximum (FWHM) value, and construct the particle size perturbation factor based on the main peak drift value, the proportion of perturbed particles, and the FWHM value. The difference analysis module is configured to obtain the indentation rebound factor and particle size perturbation factor of the current batch and the adjacent batches before and after it. Based on the difference between the indentation rebound factor and the difference between the particle size perturbation factor of the current batch and the adjacent batches before and after it, the interference dynamic increment factor is constructed by weighted summation. The status judgment module is configured to calculate the impact response deviation value based on the changing trend of the interference dynamic increment factor, and compare it with the interval threshold to determine whether the current tool has a damage trend or a critical fault state.

Images