Method for monitoring operation state of waste regeneration equipment based on Internet of Things

Abstract

The invention provides a waste regeneration equipment operation state monitoring method based on the Internet of Things, which comprises the following steps: acquiring image information of a wear area on the surface of a crushing cavity, performing image processing on the image information of the wear area, and extracting wear key features; analyzing the distribution proportion of local sharp wear and uniform wear forms according to the wear key features, and determining wear form classification; identifying stress distribution of different wear forms, performing finite element analysis to obtain a stress distribution cloud picture and stress gradient distribution, and determining a stress abnormal region; carrying out correlation analysis on the stress abnormal region and the wear form classification, and generating a potential distribution interval of the equipment fault risk; and aiming at key influence factors of quality stability control, the sampling frequency and the monitoring point position of the Internet of Things sensor are adjusted, and the adjusted wear degree real-time data and particle size change information are obtained.

Description

Technical Field

[0001] This invention relates to the field of information technology, and in particular to a method for monitoring the operational status of waste recycling equipment based on the Internet of Things. Background Technology

[0002] With the acceleration of urbanization, the demand for the treatment of large quantities of waste concrete is increasing. Concrete crushing equipment, as a core tool, has a significant impact on the entire treatment process due to the performance and stability of its crushing chamber. Research in this technological field is of great practical significance and strategic value for improving the resource utilization rate of construction waste, reducing environmental pollution load, and promoting sustainable development. However, current methods in waste concrete treatment often neglect the dynamic interaction between material characteristics and equipment operating status, especially when facing concrete of different strength grades and complex compositional variations. They lack in-depth analysis of equipment wear mechanisms and real-time response mechanisms. When the crushing teeth experience localized sharp wear, their crushing capacity changes, and the crushing process, which originally produced uniform particle sizes, may become uneven. Severely worn areas of the crushing teeth may be unable to effectively crush larger concrete blocks, resulting in oversized particles in the product, while less worn areas may produce excessive fine powder. This uneven particle size distribution not only affects the engineering performance of recycled aggregates but may also lead to a decrease in the efficiency of subsequent grading and screening processes. This non-uniform particle size distribution not only affects the engineering properties of recycled aggregates but may also lead to a decrease in the efficiency of subsequent grading and screening processes. Therefore, how to dynamically adjust the acquisition strategy of IoT sensors to monitor the wear degree of the crushing chamber and the changes in product particle size distribution in real time, and optimize preventive maintenance cycles and product quality control based on this, has become a key problem that urgently needs to be solved in this study. Summary of the Invention

[0003] This invention provides a method for monitoring the operational status of waste recycling equipment based on the Internet of Things, mainly including:

[0004] Image information of the wear area on the surface of the crushing chamber is collected, and the image information of the wear area is processed to extract key wear features;

[0005] Based on the analysis of key wear characteristics, the distribution ratio of localized sharp wear and uniform wear patterns is determined to classify wear patterns.

[0006] By identifying the stress distribution of different wear patterns, finite element analysis is performed to obtain stress distribution cloud maps and stress gradient distribution, and stress anomaly regions are determined.

[0007] By performing correlation analysis between stress anomaly regions and wear morphology classifications, potential distribution ranges of equipment failure risks can be generated.

[0008] Real-time acquisition of product particle size distribution data within the potential distribution range of equipment failure risks; analysis of particle size distribution fluctuation characteristics; and determination of key influencing factors for quality stability control based on product quality standard requirements and customer particle size specification needs.

[0009] To address the key factors affecting quality stability control, the sampling frequency and monitoring points of the IoT sensors were adjusted to obtain real-time data on wear level and particle size change information after the adjustment.

[0010] By analyzing real-time data on wear degree and particle size change information, and verifying and correcting the initial wear morphology change trend, the optimal maintenance time window under the current operating conditions of the equipment is predicted, and the quality stability control parameter settings are updated to obtain an optimized operating status monitoring scheme.

[0011] Furthermore, the acquisition of image information of the wear area on the surface of the crushing chamber, and the image processing of the wear area to extract key wear features, including:

[0012] Multi-angle images of the wear area on the surface of the crushing chamber are acquired. An edge detection algorithm is used to identify the contour of the wear area, and the mean and variance of the grayscale values ​​of pixels within the contour are calculated to divide the area into heavily worn and lightly worn sub-regions. For each of the heavily worn and lightly worn sub-regions, pixel calibration is performed to establish a mapping between image coordinates and physical dimensions. The actual wear area of ​​each sub-region is calculated and accumulated to obtain the total wear area. Based on the multi-angle images, the positional differences of feature points are calculated to generate disparity values. Three-dimensional point cloud data is reconstructed using the disparity values ​​and camera calibration parameters. Reference plane parameters are extracted, and the vertical distance from each point in the point cloud to the reference plane is calculated. The maximum vertical distance is determined as the wear depth. Local plane fitting is performed on the three-dimensional point cloud data, and the angle distribution between the normal vector of the local plane and the normal vector of the reference plane is calculated to determine the surface roughness value.

[0013] Furthermore, the step of analyzing the distribution ratio of localized sharp wear and uniform wear patterns based on key wear characteristics to determine wear pattern classification includes:

[0014] A depth distribution map is constructed based on wear depth data. The depth gradient value is calculated, and the proportion of sharp wear points with depth gradient values ​​exceeding a threshold is statistically analyzed. Connected regions are marked on the depth distribution map, the average depth of each connected region is calculated, and the number of points with depths exceeding twice the average depth is statistically analyzed to determine the peak density value. Based on the peak density value and the coordinates of the region center, the average distance and aggregation index of the center points of high peak density regions are calculated. The wear region boundary point sequence is extracted, and the rate of change of the direction angle of the boundary point connection is calculated to determine the distribution uniformity. Based on the aggregation index and the distribution uniformity, the wear morphology classification is determined.

[0015] Furthermore, the process of identifying stress distributions of different wear patterns, performing finite element analysis to obtain stress distribution cloud maps and stress gradient distributions, and determining stress anomaly regions includes:

[0016] The geometric features of the wear morphology classification regions are identified, and adaptive meshing is performed on these regions. Mesh sizes are set for sharp wear regions and uniform wear regions, and material property parameters are assigned to the mesh units. Boundary conditions are applied to construct stress-strain relationships, and stress tensor data for each unit are calculated to generate the stress distribution cloud map. The ratio of the principal stress difference to the distance between adjacent units is calculated, and stress concentration locations where the stress gradient exceeds a threshold are marked. The principal stress peaks are extracted. The principal stress peaks are statistically analyzed, and regions exceeding the material yield strength threshold or stress gradient exceeding a critical value are identified as stress anomaly regions.

[0017] Furthermore, the correlation analysis between stress anomaly regions and wear morphology classifications to generate potential distribution ranges for equipment failure risks includes:

[0018] Calculate the spatial overlap area between the stress anomaly region and the wear morphology classification region, determine the correlation strength coefficient, mark the overlap between severe concentrated wear and high stress anomaly region, and generate the wear and stress coupling influence area distribution map; calculate the product of wear depth and stress peak value, determine the influence weight, and mark high-risk coupling areas exceeding the threshold; combine historical fault data and operating condition parameters to calculate the actual fault probability; based on the actual fault probability and the spatial distribution of high-risk coupling areas, calculate the cumulative fault risk value per unit area, and classify it into four levels from low to high: low risk, medium risk, high risk, and extremely high risk, with the corresponding risk value range being the potential distribution range of equipment fault risk.

[0019] Furthermore, the product particle size distribution data within the potential distribution range of the real-time acquisition equipment failure risk is analyzed to assess the fluctuation characteristics of the particle size distribution. Based on product quality standards and customer particle size specification requirements, key influencing factors for quality stability control are determined, including:

[0020] During the high-risk and extremely high-risk periods of the potential distribution range, particle size distribution data of the crushed products are collected in real time, and the mass percentage and risk level of each particle size segment are recorded to construct a particle size distribution time series. The mass percentage difference between adjacent moments in the particle size distribution time series is calculated to determine the fluctuation amplitude sequence of each particle size segment, and the standard deviation of the mass proportion of each screening level is statistically analyzed as the fluctuation characteristic value. The particle size distribution is compared with the product quality standard to calculate the comprehensive quality deviation value. The fault risk level, wear mode category and stress anomaly degree when the comprehensive quality deviation value exceeds the threshold are extracted to determine the key influencing factors of quality stability control.

[0021] Furthermore, regarding the key influencing factors for quality stability control, adjusting the sampling frequency and monitoring points of the IoT sensor to obtain real-time data on wear degree and particle size change information after adjustment includes:

[0022] Based on the wear pattern type and stress anomaly degree, identify the crushing cavity area where the value exceeds the threshold, add IoT sensors, and divide the monitoring points into high, medium and low priority points; set differentiated sampling frequencies for the high priority points, medium priority points and low priority points, and obtain real-time data on wear degree and particle size change information of the adjusted sensor network.

[0023] Furthermore, the analysis of real-time wear data and particle size change information, combined with preliminary wear morphology change trend verification and correction, predicts the optimal maintenance time window under the current operating conditions of the equipment, updates the quality stability control parameter settings, and obtains an optimized operating status monitoring scheme, including:

[0024] Time-series analysis is performed on the real-time wear level data and the particle size change information to calculate the cross-correlation coefficient sequence, determine the time delay, and calculate the correlation value according to the wear morphology. Equipment load data is acquired, the load comprehensive index is calculated, and the linear relationship between the load comprehensive index and the wear rate increment is fitted to determine the correlation coefficient. Based on the wear rate and the load comprehensive index, the remaining time before the wear depth reaches the threshold is calculated, the most recent downtime period is selected as the maintenance time window, and a scheduling instruction containing maintenance type and spare parts list is generated. The quality control sampling frequency, alarm limit, and data upload priority are adjusted to formulate an optimized operation status monitoring scheme.

[0025] The technical solutions provided by the embodiments of the present invention may include the following beneficial effects:

[0026] This invention discloses an intelligent monitoring and maintenance optimization method for the coupled influence of crushing chamber wear and equipment operating status, solving the complex correlation between crushing chamber wear morphology, stress distribution, and product particle size fluctuation. By acquiring wear images of the chamber surface, key features such as wear depth, area, and roughness are extracted. Combined with finite element analysis, a stress cloud map is generated to identify stress concentration areas and correlate them with wear morphology, constructing a wear-stress coupling distribution map. Based on a risk assessment matrix and operating conditions, the method predicts equipment failure risk ranges, monitors particle size distribution fluctuations in real time, and determines key factors for quality stability. By adjusting the sampling frequency and monitoring points of IoT sensors, the correlation between wear rate and particle size fluctuation and the impact of equipment load are analyzed to predict the optimal maintenance time window, generate targeted maintenance scheduling instructions, and optimize the operating status monitoring scheme. This invention achieves accurate equipment failure prediction and quality stability control through multi-dimensional coupled analysis and dynamic adjustment of wear-stress-particle size, significantly improving equipment operating efficiency and product consistency. Attached Figure Description

[0027] Figure 1 This is a flowchart of a method for monitoring the operational status of waste recycling equipment based on the Internet of Things, according to the present invention.

[0028] Figure 2 This is a schematic diagram of a waste recycling equipment operation status monitoring method based on the Internet of Things according to the present invention. Detailed Implementation

[0029] The technical solution of the present invention will be clearly and completely described below with reference to the embodiments. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0030] like Figure 1-2 This embodiment of a method for monitoring the operational status of waste recycling equipment based on the Internet of Things may specifically include:

[0031] Step S101: Collect image information of the wear area on the surface of the crushing chamber, perform image processing on the image information of the wear area, and extract key wear features.

[0032] Multi-angle images of the wear area on the surface of the crushing chamber were acquired. The Canny edge detection algorithm was used to identify the contour of the wear area. Gray-level values ​​of pixels within the contour were statistically analyzed, and the mean and variance of gray-level values ​​were calculated. Areas with a gray-level mean below a preset threshold were classified as heavily worn sub-regions, and areas with a gray-level mean above the preset threshold were classified as lightly worn sub-regions. For each sub-region, pixel calibration was performed to establish a mapping relationship between image coordinates and actual physical dimensions. The actual wear area of ​​each sub-region was calculated using a contour extraction algorithm. The areas of the heavily worn and lightly worn sub-regions were summed to obtain the total wear area data. Based on the acquired multi-angle images, disparity values ​​were calculated according to the positional differences of the same feature point in different angle images. The disparity values ​​and camera calibration parameters were used to perform 3D coordinate transformation to reconstruct 3D point cloud data of the wear surface. Reference plane parameters were extracted from the unworn edge region of the chamber, and the vertical distance from each point in the point cloud to the reference plane was calculated. The maximum vertical distance value was determined as the wear depth. Local plane fitting is performed on the 3D point cloud data, the normal vector of each local plane is calculated, the distribution of the angle between all normal vectors and the normal vector of the reference plane is statistically analyzed, the standard deviation of the angle distribution is calculated, and the surface roughness value is determined according to the mapping relationship between the standard deviation and the preset roughness reference table, thus obtaining key wear features including wear depth, total wear area, and surface roughness.

[0033] For example, during long-term operation, the inner wall of the crushing chamber will wear down due to repeated impacts and friction from materials. Accurately acquiring wear characteristics is crucial for assessing equipment condition and developing maintenance strategies. Multi-angle image acquisition is fundamental to obtaining complete wear information. By arranging multiple cameras around the chamber and photographing the same wear area from different perspectives, three-dimensional information of the wear surface can be captured.

[0034] In one possible implementation, the Canny edge detection algorithm accurately extracts the boundary contours of worn areas by calculating the magnitude and direction of the image gradient and utilizing non-maximum suppression and double threshold detection. This algorithm exhibits good robustness to illumination changes, making it particularly suitable for handling uneven illumination within cavities. During grayscale value statistics, the algorithm iterates through each pixel within the contour, accumulating its grayscale value and calculating the mean and variance. When the grayscale mean is below a set threshold, it indicates less reflected light in the area, typically corresponding to deeper wear depressions; conversely, it indicates less wear. This grayscale feature-based region segmentation provides a basis for subsequent differential processing.

[0035] Specifically, pixel calibration establishes a mapping between the image plane and the physical world. By placing a calibration plate of known size inside the cavity, the algorithm can calculate the actual physical size corresponding to each pixel. After contour extraction, the actual wear area of ​​each sub-region can be obtained by counting the number of pixels within the contour and multiplying it by the physical area of ​​a single pixel. By summing the areas of heavily worn and lightly worn regions, not only can the total wear area be obtained, but also the distribution characteristics of wear can be understood. This is of guiding significance for analyzing the causes of wear and optimizing cavity design.

[0036] It should be noted that disparity calculation utilizes the principles of stereo vision. When the same feature point is identified in images from different angles, the difference in its position on the image plane reflects the depth information of that point. Using intrinsic and extrinsic parameters obtained through camera calibration, including focal length, principal point coordinates, and the relative positional relationship between cameras, the algorithm can convert disparity values ​​into three-dimensional coordinates. After reconstructing the 3D point cloud, multiple points are extracted from the unworn area of ​​the cavity edge, and the reference plane parameters are obtained through least-squares fitting. The vertical distance from each point in the point cloud to this reference plane is calculated, and the maximum value represents the wear depth, which directly reflects the severity of wear. Surface roughness is evaluated through local plane fitting. The algorithm selects a local neighborhood in the point cloud data and performs plane fitting on the points within that neighborhood to obtain the normal vector of that local region. The angle between all local normal vectors and the reference plane normal vector is calculated; a larger angle indicates a more inclined surface. The standard deviation of the angle distribution reflects the uniformity of surface undulation; a larger standard deviation indicates a rougher surface. By using a pre-established table comparing standard deviation with roughness values, statistical results can be converted into roughness indices that conform to industrial standards, thus providing a complete system of characteristic parameters for the quantitative assessment of wear.

[0037] Step S102: Analyze the distribution ratio of local sharp wear and uniform wear patterns based on key wear characteristics to determine the wear pattern classification.

[0038] A depth distribution map is constructed based on wear depth data. The ratio of the depth difference to the distance between adjacent wear points is calculated as the depth gradient value. When the depth gradient value exceeds a preset gradient threshold, it is identified as a sharp wear point. The proportion of sharp wear points to the total number of wear points is counted to obtain the sharp wear percentage data. Connected regions are marked based on the depth distribution map. Adjacent points with a depth difference less than a preset difference threshold are grouped into the same connected region. The average depth of all points in each connected region is calculated as the region's average depth. The number of points in each connected region whose depth exceeds twice the region's average depth is counted, and divided by the region's area to obtain the peak density value. Based on the peak density value and the region's center coordinates, the average distance between the center points of all high peak density regions is calculated. The clustering index is obtained by dividing the total wear area by the number of high peak density regions and then by the average distance. Simultaneously, the wear region boundary point sequence is extracted, and the rate of change of the direction angle of the line connecting adjacent boundary points is calculated as the curvature value. The standard deviation of all curvature values ​​is counted as the distribution uniformity. Wear patterns are determined using aggregation index and distribution uniformity. If the aggregation index is less than 0.3 and the distribution uniformity is less than 0.2, it is judged as mild uniform wear. If the aggregation index is between 0.3 and 0.7, it is judged as moderate local wear. If the aggregation index is greater than 0.7 and the peak density exceeds the preset density threshold, it is judged as severe concentrated wear. If the proportion of sharp wear exceeds 0.6, it is judged as abnormally sharp wear.

[0039] For example, the calculation of depth gradient values ​​is a key indicator for identifying wear patterns.

[0040] In one possible implementation, the gradient value is obtained by calculating the ratio of the depth difference between two adjacent sampling points to the horizontal distance. When deep pits or grooves appear on the surface of the crushing cavity, the depth changes drastically over a short distance, and the gradient value increases significantly. A gradient threshold of 0.5 is set, meaning that when the depth change reaches half the horizontal distance, the location is identified as a sharp wear point. This method effectively distinguishes between slowly transitioning uniform wear and abruptly sharp wear. Connected region labeling uses the eight-neighborhood connectivity criterion, grouping adjacent points with similar depths into the same region.

[0041] Specifically, when the depth difference between two adjacent points is less than 0.2 mm, they are considered to belong to the same wear region. This grouping method can divide the wear surface into multiple independent wear units, each with similar wear characteristics. By statistically analyzing the area and depth distribution of each connected region, the spatial distribution characteristics of wear can be understood.

[0042] It's important to note that peak density calculation reflects the severity of localized wear. Within each connected region, the average depth of all points is first calculated, then the number of points with depths exceeding twice the average is counted. These depth anomalies typically correspond to the most severely worn locations, such as the center of a pit formed by repeated material impacts. Dividing the number of anomalies by the region area yields the peak density value, which quantitatively describes the degree of wear concentration in that region. Higher peak density indicates more depth anomalies and more uneven wear within the region. The calculation of the clustering index involves several spatial statistics. First, all high-wear regions with peak densities exceeding a threshold are identified, and the average distance between the center points of these regions is calculated. This distance reflects the dispersion of severely worn areas. Then, the total wear area is divided by the number of high-wear regions to obtain the average influence range of a single high-wear region, which is then divided by the average distance to obtain the clustering index. A larger clustering index indicates a more concentrated distribution of high-wear regions and a more localized wear pattern. Boundary curvature calculation is used to assess the regularity of the wear region's contour. A point sequence is extracted along the boundary of the wear region, and the rate of change of the angle formed by three adjacent points is calculated as the local curvature value. The standard deviation of all curvature values ​​is calculated to obtain the distribution uniformity index. Uniform wear typically has smooth boundaries with small curvature variations and low standard deviations; while localized wear has irregular boundaries with large curvature variations and high standard deviations. By comprehensively using four indicators—aggregation index, distribution uniformity, peak density, and the proportion of sharp wear—the wear morphology can be accurately determined. Mild uniform wear is characterized by low aggregation index and low distribution uniformity; moderate localized wear has an intermediate aggregation index; severe concentrated wear not only has a high aggregation index but also a peak density exceeding the threshold; and abnormally sharp wear is characterized by a high proportion of sharp wear. This multi-indicator comprehensive determination method improves the accuracy and reliability of wear morphology classification.

[0043] Step S103: Identify the stress distribution of different wear patterns, perform finite element analysis to obtain stress distribution cloud map and stress gradient distribution, and determine the stress anomaly area.

[0044] The geometric characteristics of different wear morphologies are identified, including four types of regions: mild uniform wear, moderate localized wear, severe concentrated wear, and abnormally sharp wear. Adaptive meshing is performed on each type of wear region. The mesh size for sharp wear regions is set to one-tenth of the maximum depth of that region, while the mesh size for uniform wear regions is set to a preset standard value. Elastic modulus and Poisson's ratio parameters are assigned to the mesh element materials. Boundary conditions are applied to the mesh model based on the actual working conditions of the crushing chamber, including displacement constraints in non-wear regions and normal pressure and tangential friction force determined according to material properties on the wear surface. Stress-strain relationships are constructed using elastic modulus and Poisson's ratio, and the nodal displacement field is solved. The stress tensor data of each element is calculated through the strain-displacement relationship and the stress-strain relationship. The principal stress values ​​of each element's stress tensor data are used as the stress values ​​for that element, generating a stress distribution cloud map. The ratio of the difference in principal stress between adjacent elements to the distance between the element centers is calculated as the local stress gradient. When the stress gradient exceeds a preset gradient threshold, it is marked as a stress concentration location, and the peak principal stress at that location is extracted. The peak values ​​of principal stresses at each stress concentration location are statistically analyzed. When the peak value exceeds 0.8 times the yield strength of the material, or when the stress gradient exceeds a preset critical value, the continuous area where the gradient at that location and its surrounding area exceeds the threshold is identified as a stress anomaly area. The spatial coordinates and corresponding peak stress intensities of all stress anomaly areas are recorded.

[0045] For example, adaptive mesh generation is a fundamental task in finite element analysis, and its core lies in the rational allocation of computational resources according to the characteristics of different wear patterns.

[0046] In one possible implementation, for abnormally sharp wear regions, due to their drastic depth variations and large stress gradients, a fine mesh is required to accurately capture stress changes. The mesh size is set to one-tenth of the maximum depth of the region; that is, if the maximum depth of a sharp wear pit is 5 mm, the mesh size for that region should be set to 0.5 mm. This refinement can achieve sufficient computational accuracy in stress concentration areas. In contrast, the stress distribution in uniform wear regions is relatively gentle, and a larger standard mesh can meet the accuracy requirements, saving computational resources while ensuring analysis efficiency. The setting of material property parameters directly affects the accuracy of stress calculation results. The elastic modulus reflects the material's ability to resist elastic deformation, while Poisson's ratio describes the ratio of transverse strain to longitudinal strain under uniaxial stress. The crushing chamber is usually made of high-manganese steel, with an elastic modulus of approximately 200 GPa and a Poisson's ratio of approximately 0.3. These two parameters together constitute the constitutive relation of the material, i.e., the mathematical relationship between stress and strain.

[0047] It should be noted that the boundary conditions applied must conform to the actual working conditions. The non-wear region is typically a fixed part of the cavity; applying displacement constraints means that this region remains stationary during the analysis. The loads on the wear surface include normal pressure and tangential friction. The normal pressure comes from the impact of the material, and its magnitude depends on the mass, velocity, and impact angle of the material. The tangential friction is generated by the sliding of the material on the cavity surface and is proportional to the friction coefficient and the normal pressure. The stress solution process follows the basic principles of the finite element method. The displacement values ​​of each node are obtained by solving the global stiffness equation, then the element strain is calculated using the strain-displacement relationship, and finally the stress tensor is obtained through the stress-strain relationship. The stress tensor is a symmetric matrix containing six independent components, describing the complete stress state within the element. The principal stresses are the eigenvalues ​​of the stress tensor, representing the maximum, intermediate, and minimum normal stress values ​​at that point. The maximum principal stress is often a key indicator for determining whether the material has failed.

[0048] Specifically, the calculation of stress gradient reveals the spatial rate of stress variation. The difference in principal stress between adjacent elements reflects the amount of stress change, which, when divided by the distance between the element centers, yields the rate of stress change per unit length. When this rate of change exceeds a preset threshold, it indicates a sharp change in stress at that location, typically corresponding to abrupt geometric changes or areas of load concentration. These stress concentration locations are weak points in the structure, prone to cracking and ultimately failure. The determination of stress anomaly regions employs a dual standard. On one hand, when the peak principal stress reaches 0.8 times the material's yield strength, the material is close to the critical state of plastic deformation; on the other hand, even if the absolute stress value is not high, if the stress gradient exceeds the critical value, fatigue failure may still occur due to stress concentration effects.

[0049] Step S104 involves performing correlation analysis between the stress anomaly region and wear pattern classification to generate the potential distribution range of equipment failure risk.

[0050] The spatial coordinates of the stress anomaly region and the boundary data of the wear morphology classification region are obtained. The spatial overlap area of ​​the two is calculated, and the overlap area is divided by the total area of ​​the stress anomaly region and the total area of ​​the wear region, respectively. The average of the two ratios is taken as the correlation strength coefficient. When a heavily concentrated wear region overlaps with a high stress anomaly region, the correlation strength coefficient is multiplied by a preset amplification factor to obtain the coupling coefficient. Based on the coupling coefficient and regional spatial coordinate data, the coupling coefficient values ​​of each region are marked in the coordinate system, generating a distribution map of the wear and stress coupling influence area. The product of the wear depth and the stress peak value of each region is calculated, and the product is divided by the average of the product of all regions to obtain the influence weight. Regions with influence weights exceeding a preset weight threshold are marked as high-risk coupling zones. A risk assessment matrix is ​​constructed for high-risk coupling zones. The number of rows in the matrix corresponds to the four wear morphology classifications, and the number of columns corresponds to the stress level classification. The failure probability benchmark value of each cell is determined by historical failure statistics. The current operating condition parameters, including material hardness coefficient, feed rate, and cumulative running time, are obtained. The correction coefficient is calculated based on the deviation of the operating condition parameters from the standard operating condition. The actual failure probability is obtained by multiplying the correction coefficient by the benchmark value. Based on the actual failure probability and the spatial distribution of high-risk coupling zones, the cumulative failure risk value per unit area is calculated. The cumulative risk value is divided into four levels from low to high: low risk, medium risk, high risk, and extremely high risk. The corresponding risk value range is the potential distribution range of equipment failure risk.

[0051] For example, the calculation of the correlation strength coefficient reflects the spatial correlation between the two failure mechanisms of wear and stress.

[0052] In one possible implementation, when stress anomaly regions overlap with wear regions, it indicates that the location is simultaneously affected by both geometric defects and mechanical loads. The degree of spatial correlation can be quantified by calculating the ratio of the overlapping area to the total area of ​​each region. Taking the average of the two ratios as the correlation strength coefficient avoids the bias that a single indicator might introduce. When severe concentrated wear overlaps with high stress anomaly regions, the failure risk increases exponentially; therefore, a magnification factor is needed to reflect this nonlinear relationship. The process of generating the coupled influence region distribution map transforms abstract numerical values ​​into an intuitive spatial distribution.

[0053] Specifically, in the unfolded coordinate system of the crushing chamber, each region is assigned a different color depth or contour density based on its coupling coefficient value. The product of wear depth and peak stress reflects the overall damage level of that region. This product relationship conforms to the physical mechanism of material failure, namely, the greater the stress and the deeper the defect, the higher the probability of failure. Normalization is performed by dividing by the average of the products of all regions, making the influence weights of different regions comparable.

[0054] It should be noted that the risk assessment matrix is ​​constructed based on statistical analysis of a large amount of historical failure data. The rows of the matrix correspond to four wear patterns: mild uniform wear, moderate localized wear, severe concentrated wear, and abnormally sharp wear; the columns are divided into low, medium, and high levels according to stress levels. The baseline failure probability value for each cell is derived from historical statistics under the same operating conditions, reflecting the likelihood of failure caused by a specific wear-stress combination. This two-dimensional matrix structure can comprehensively cover various possible failure mode combinations. The correction of failure probability by operating condition parameters reflects the impact of actual operating conditions. The material hardness coefficient reflects the abrasive ability of the crushed material on the cavity; the higher the hardness, the faster the wear rate. The feed rate determines the impact frequency and energy per unit time; the cumulative running time reflects the cumulative effect of fatigue damage. The correction coefficient is calculated by comparing the difference between the current operating condition and the standard operating condition; for example, if the material hardness is 20% higher than the standard value, the correction coefficient increases accordingly. The calculation of the cumulative failure risk value considers the spatial distribution characteristics of the risk. The more high-risk coupling zones per unit area, the denser the failure points, and the greater the possibility of overall equipment failure. By statistically analyzing the failure probabilities of different areas and performing spatial integration, a cumulative value reflecting the overall risk level is obtained. The risk level classification adopts a four-level system, with each level corresponding to a specific risk value range. For example, a cumulative risk value of 0-0.25 is low risk, 0.25-0.5 is medium risk, 0.5-0.75 is high risk, and 0.75-1.0 is extremely high risk.

[0055] Step S105: Collect product particle size distribution data in real time within the potential distribution range of equipment failure risk, analyze the fluctuation characteristics of particle size distribution, and determine the key influencing factors of quality stability control based on product quality standard requirements and customer particle size specification requirements.

[0056] Based on the potential distribution range of equipment failure risks, real-time data acquisition is initiated during high-risk and extremely high-risk periods. Particle size distribution data of the crushed products is acquired at preset time intervals using a particle size analyzer. The mass percentage of each particle size segment and its corresponding risk level are recorded, constructing a particle size distribution time series with risk labels. The particle size distribution time series is analyzed, calculating the difference in mass percentage of the same particle size segment at adjacent times to obtain the fluctuation amplitude sequence for each particle size segment. The analyzer data is verified using a screening device. Based on the screening levels set according to customer particle size specifications, the standard deviation of the mass percentage of each level is calculated as the fluctuation characteristic value for that level. The actual particle size distribution is compared with the percentages of each level specified in the product quality standard. The absolute value of the difference between the actual percentage and the standard percentage for each level is calculated, and a weighted sum is obtained to obtain the comprehensive quality deviation value. When the deviation value exceeds the allowable threshold, the failure risk level, wear pattern type, and stress anomaly degree data for the corresponding time are extracted. Establish a correlation matrix between quality deviation value and failure risk level, wear pattern type, and stress anomaly degree. Calculate the Pearson correlation coefficient between each factor and quality deviation. Sort the factors according to the absolute value of the correlation coefficient from largest to smallest, and identify the factors with correlation coefficients exceeding the preset significance threshold as key influencing factors for quality stability control.

[0057] For example, the correlation mechanism between the fault risk distribution range and particle size acquisition is the core of quality monitoring.

[0058] In one possible implementation, when the equipment is in a high-risk or extremely high-risk range, wear and stress concentration in the crushing chamber directly affect the crushing effect, leading to abnormal product particle size distribution. By initiating intensive sampling during these critical periods, the immediate impact of equipment status changes on product quality can be captured. The particle size analyzer uses the principle of laser diffraction; when the crushed product passes through the detection area, particles of different sizes produce different light scattering patterns. By analyzing the intensity distribution of the scattered light, particle size distribution data can be obtained. The introduction of risk labels gives the particle size data a state attribute.

[0059] Specifically, each set of particle size distribution data is labeled with the risk level corresponding to the collection time, such as low risk corresponding to value 1, medium risk to value 2, high risk to value 3, and extremely high risk to value 4. This labeling method establishes a data bridge between equipment status and product quality, laying the foundation for subsequent correlation analysis. The construction of the time series not only records the dynamic changes in particle size distribution but also retains the temporal correspondence with the equipment risk status.

[0060] It should be noted that the calculation of fluctuation amplitude reveals the stability characteristics of particle size distribution. The difference in the mass percentage of the same particle size range at adjacent moments reflects the degree of instantaneous change in that particle size range. When the wear of the crushing chamber intensifies, the crushing gap increases, leading to an increase in large-size products; while stress concentration may cause local over-crushing, producing excessive fine powder. The verification role of the screening device is to provide accurate data on physical separation, complementing the optical measurements of the detector and improving data reliability. The calculation of the overall mass deviation value uses a weighted summation method, with the weights set according to the customer's sensitivity to different particle size ranges.

[0061] For example, a customer requires 60% of particles to be 5-10mm in size, but the actual measured particle size is 55%, a deviation of 5%; the required particle size is 30% for 10-20mm, but the actual measured particle size is 32%, a deviation of 2%. If the customer is more sensitive to the 5-10mm particle size and assigns it a weight of 0.7, and assigns a weight of 0.3 to the 10-20mm particle size, the overall deviation value is 5 × 0.7 + 2 × 0.3 = 4.1. This weighting method ensures consistency between quality evaluation and customer needs. The Pearson correlation coefficient quantifies the degree of influence of each factor on quality stability. The correlation coefficient is calculated based on the covariance and standard deviation between the quality deviation value series and the series of influencing factors. A correlation coefficient close to 1 indicates a positive correlation, close to -1 indicates a negative correlation, and close to 0 indicates no correlation. By setting a significance threshold such as 0.6, strongly correlated factors are screened out.

[0062] For example, if the correlation coefficient between wear pattern type and quality deviation is 0.85, the correlation coefficient between stress anomaly degree is 0.72, and other factors are below 0.6, then wear pattern and stress anomaly are determined to be the key influencing factors.

[0063] Step S106: For key influencing factors of quality stability control, adjust the sampling frequency and monitoring points of the IoT sensor to obtain real-time data on wear degree and particle size change information after adjustment.

[0064] Based on the identified wear morphology and stress anomaly levels—two key influencing factors—areas within the crushing chamber where these values ​​exceed preset thresholds are designated as key monitoring areas. IoT sensors are added to these key monitoring areas, and monitoring points are categorized into high, medium, and low priorities based on the correlation coefficients of each factor. Different sampling frequencies are set for monitoring points of different priorities: high-priority points with a correlation coefficient greater than 0.8 are sampled once per minute; medium-priority points with a correlation coefficient between 0.6 and 0.8 are sampled once every three minutes; and low-priority points with a correlation coefficient less than 0.6 are sampled once every five minutes. Wear depth data aligned to time at each point is obtained through a timestamp synchronization mechanism. Using this time-aligned wear depth data, the wear rate is calculated by subtracting the previous moment's depth value from the current moment's depth value and then dividing by the time interval. Simultaneously, the particle size distribution of the product is collected, and the root mean square error between the actual distribution and the standard distribution is calculated as the particle size deviation value. A real-time mapping table is created by mapping the wear rate and particle size deviation values ​​at the same moment. Based on the real-time mapping table, when the wear rate at a certain point exceeds the preset rate threshold or the corresponding particle size deviation exceeds the quality control limit, the sensor sampling frequency, monitoring point coordinates, wear depth value, wear rate value and particle size distribution data at that moment are extracted to obtain the real-time data of wear degree and particle size change information of the adjusted sensor network.

[0065] For example, the identification of key monitoring areas is based on data on key influencing factors derived from previous analysis.

[0066] In one possible implementation, an area is designated as a key monitoring area when its wear pattern is determined to be either severe concentrated wear or abnormally sharp wear, and the degree of stress anomaly exceeds 0.8 times the material's yield strength. This dual criterion ensures that monitoring resources are concentrated at the locations most likely to experience failure. The addition of IoT sensors follows the principles of coverage and economy, with one sensor deployed at the center and edge of each key area to form a regional monitoring network. The correlation coefficient serves as a quantitative indicator in prioritization.

[0067] Specifically, the correlation coefficient reflects the strength of the linear relationship between each influencing factor and product quality deviation. When the correlation coefficient of wear morphology reaches 0.85, it indicates that the change in wear state is highly correlated with the particle size distribution deviation, and should be given the highest monitoring priority. This data-driven prioritization method avoids the subjectivity of experience-based judgment, making the allocation of monitoring resources more scientific and reasonable.

[0068] It should be noted that the differentiated sampling frequency settings fully consider the balance between data value and transmission cost. High-priority points are sampled once per minute to capture rapidly changing wear processes; medium-priority points are sampled every three minutes to meet routine monitoring needs; and low-priority points are sampled every five minutes, primarily serving as auxiliary verification. The timestamp synchronization mechanism is implemented through a unified clock server, with all sensors recording a time stamp accurate to the second when collecting data, ensuring that data from different locations correspond precisely in the time dimension. The wear rate is calculated using a differential method, reflecting the dynamic characteristics of the wear process.

[0069] For example, if the wear depth at a monitoring point is 2.5 mm at 10:00 and 2.52 mm at 10:01, the wear rate is 0.02 mm / min. This real-time rate calculation can promptly detect accelerated wear. The root mean square error (RMSE), as a measure of particle size deviation, comprehensively considers the deviation across different particle size segments. The calculation involves summing the squared differences between the actual and standard distributions in each particle size segment, taking the square root, and dividing by the number of particle size segments to obtain a comprehensive deviation value. The establishment of a real-time mapping table enables a dynamic correlation between equipment status and product quality. Each record in the table contains four fields: timestamp, monitoring point number, wear rate, and particle size deviation value. By continuously updating this mapping table, the changing patterns between wear rate and particle size deviation can be observed. A sudden increase in the wear rate at a certain point often indicates that the crushing effect in that area is about to deteriorate, and the particle size deviation will also increase accordingly. The real-time extraction and recording of monitoring data forms a complete status tracking chain. When an early warning condition is triggered, not only is the current abnormal data recorded, but also the sensor's configuration parameters, including sampling frequency and location coordinates, are saved.

[0070] Step S107: Analyze the real-time data on wear degree and particle size change information, verify and correct the initial wear morphology change trend, predict the optimal maintenance time window under the current operating conditions of the equipment, update the quality stability control parameter settings, and obtain an optimized operating status monitoring scheme.

[0071] Time-series analysis was performed on the real-time data sequence of wear degree and the information sequence of particle size change. The cross-correlation coefficient sequence was obtained by calculating the product of the covariance of the two sequences at different time offsets and their respective standard deviations. The time delay corresponding to the maximum correlation coefficient was determined. Combined with the obtained wear morphology classification results, the correlation was calculated according to different wear morphologies, obtaining the correlation values ​​between wear rate and particle size fluctuation under each morphology. Based on the correlation values, real-time load data of the equipment was obtained. The load composite index was obtained by multiplying the feed rate by the rated feed rate, the operating speed by the rated speed, and the working pressure by the rated pressure. The least squares method was used to fit the linear relationship between the load composite index and the wear rate increment, and the slope was the correlation coefficient. When the coefficient exceeded a preset threshold, a predictive relationship between load and wear rate was constructed. Based on the predictive relationship, the current wear rate, and the load status, the remaining time before the wear depth reaches the maximum allowable value was calculated. The production plan was queried to obtain the future scheduled downtime. The nearest downtime before the remaining time was selected as the maintenance time window, generating scheduling instructions that included maintenance types such as liner replacement, clearance adjustment, or full overhaul, as well as the required man-hours and spare parts list. Based on the maintenance cycle in the scheduling instructions, the quality control sampling frequency is adjusted to one-tenth of the maintenance cycle, the alarm limit is set to 80% of the normal range, the data upload priority of each monitoring point is reallocated, and an optimized operation status monitoring scheme is formulated, which includes a new sampling frequency, alarm threshold, data compression and transmission rules, and abnormal response procedures.

[0072] For example, cross-correlation analysis has important applications in time series data processing.

[0073] In one possible implementation, cross-correlation calculations of the wear rate sequence and the particle size fluctuation sequence reveal the time dependency between them. When the wear of the crushing chamber intensifies, the crushing effect is reflected in the product particle size after a certain time delay. By calculating the covariance of the two sequences at different time offsets, the delay time corresponding to the strongest correlation can be found.

[0074] For example, after the wear rate at a certain monitoring point increases, the particle size distribution typically begins to change significantly after 15-20 minutes. This time delay reflects the residence time of the material from crushing to discharge. The calculation of the load composite index embodies the idea of ​​multi-parameter normalization.

[0075] Specifically, a dimensionless comprehensive index is obtained by multiplying the ratio of actual operating parameters to rated parameters. When the feed rate is 90% of the rated value, the rotational speed is 95% of the rated value, and the pressure is 85% of the rated value, the load comprehensive index is 0.9 × 0.95 × 0.85 = 0.73. This normalization process allows parameters with different dimensions to be compared and calculated on the same scale. Least squares fitting finds the optimal linear relationship parameters by minimizing the squared error between the predicted and actual values.

[0076] It is important to note that the accuracy of wear prediction directly impacts the rationality of maintenance decisions. By establishing the relationship between load and wear rate, wear trends can be predicted based on future production plans. Assuming a current wear depth of 3 mm, a maximum allowable wear depth of 8 mm, and a current wear rate of 0.1 mm / day, under a load index of 0.8, the predicted remaining service life is approximately 50 days. This quantitative prediction provides a scientific basis for maintenance planning. The choice of maintenance type depends on the degree and distribution characteristics of wear. Replacing liners is suitable for situations where wear is nearing its limit; adjusting clearances is suitable for situations where wear distribution is uneven but the overall system is still usable; and comprehensive overhauls are implemented when multiple parts experience severe wear simultaneously. The maintenance scheduling instructions detail the required man-hours, such as 8 hours for liner replacement and 4 hours for clearance adjustment, along with a corresponding spare parts list to ensure smooth maintenance. The optimized monitoring scheme fully considers the impact of the maintenance cycle on monitoring needs.

[0077] For example, if the established maintenance cycle is 30 days, the quality control sampling frequency is adjusted to once every 3 days, which can promptly detect anomalies while avoiding oversampling. Alarm limits are set at 80% of the normal range, meaning that an alert is triggered when a certain indicator reaches 80% of the normal upper limit, allowing ample time for corrective action. Data compression and transmission rules are set with different compression ratios based on the importance and rate of change of the data; critical data is transmitted in real time, while routine data is uploaded in batches at scheduled times. The established anomaly response process ensures a rapid response from problem discovery to resolution. When monitoring data triggers an alarm, the system automatically increases the monitoring frequency in the relevant area and simultaneously notifies maintenance personnel to prepare the necessary tools and spare parts. This proactive monitoring and maintenance model transforms traditional reactive maintenance into preventative maintenance, significantly improving the reliability of equipment operation and the stability of product quality.

[0078] Based on the embodiments of the present invention described above, and through the above description, those skilled in the art can make various changes and modifications without departing from the technical concept of the present invention. The technical scope of the present invention is not limited to the contents of the specification, but must be determined according to the scope of the claims.

Claims

1. A method for monitoring the operating state of waste recycling equipment based on the Internet of Things, characterized in that, The method comprises: Collecting image information of the worn area of the surface of the crushing cavity, image processing the image information of the worn area, and extracting key wear features, including wear depth, wear area, and surface roughness; Analyzing the distribution proportion of local sharp wear and uniform wear morphology according to the key wear features to determine the wear morphology classification; Identifying the stress distribution of different wear morphologies, performing finite element analysis to obtain a stress distribution cloud map and a stress gradient distribution, and determining a stress abnormal area; Correlating the stress abnormal area with the wear morphology classification to analyze and generate a potential distribution interval of equipment failure risk; Real-time collecting product particle size distribution data of the potential distribution interval of equipment failure risk, analyzing the fluctuation characteristics of the particle size distribution, and determining key influencing factors of quality stability control according to product quality standards and customer particle size specifications; Adjusting the sampling frequency and monitoring points of the Internet of Things sensor according to the key influencing factors of quality stability control, and obtaining adjusted real-time data of wear degree and particle size change information; Analyzing the real-time data of wear degree and particle size change information, verifying and correcting the preliminary wear morphology change trend, predicting the optimal maintenance time window under the current working condition of the equipment, updating the quality stability control parameter settings, and obtaining an optimized operation state monitoring scheme. 2.The method of claim 1, wherein, The collecting image information of the worn area of the surface of the crushing cavity, image processing the image information of the worn area, and extracting key wear features comprise: Collecting multi-angle images of the worn area of the surface of the crushing cavity, identifying the wear area contour through an edge detection algorithm, calculating the gray mean and variance of the pixel points inside the contour, dividing the severe wear sub-area and the slight wear sub-area, establishing an image coordinate and physical size mapping for the severe wear sub-area and the slight wear sub-area through pixel calibration, calculating the actual wear area of each sub-area, and accumulating to obtain the total wear area; based on the multi-angle images, calculating the position difference of feature points to generate a parallax value, reconstructing three-dimensional point cloud data using the parallax value and camera calibration parameters, extracting reference plane parameters, calculating the vertical distance of each point in the point cloud to the reference plane, and determining the maximum vertical distance as the wear depth; performing local plane fitting on the three-dimensional point cloud data, calculating the included angle distribution of the normal vector of the local plane and the normal vector of the reference plane, and determining the surface roughness value. 3.The method of claim 1, wherein, The analyzing the distribution proportion of local sharp wear and uniform wear morphology according to the key wear features to determine the wear morphology classification comprises: Based on the wear depth data, constructing a depth distribution graph, calculating a depth gradient value, and counting the proportion of sharp wear points whose depth gradient value exceeds a threshold value; performing connected region labeling on the depth distribution graph, calculating the average depth of each connected region, counting the number of points whose depth exceeds twice the average depth, and determining a peak density value; according to the peak density value and the region center coordinates, calculating the average distance and aggregation index of the high peak density region center points, extracting a boundary point sequence of the wear area, calculating the direction angle change rate of the boundary point connecting line, and determining the distribution uniformity; according to the aggregation index and the distribution uniformity, determining the wear morphology classification. 4.The method of claim 1, wherein, The stress distribution of different wear morphologies is identified, finite element analysis is performed to obtain a stress distribution cloud map and a stress gradient distribution, and a stress anomaly area is determined, including: The geometric features of the wear morphology classification area are identified, adaptive mesh division is performed on the wear morphology classification area, the mesh size of the sharp wear area and the uniform wear area is set, and the grid element material attribute parameters are assigned; the boundary conditions are applied, the stress-strain relationship is constructed, the stress tensor data of each element is calculated, and the stress distribution cloud map is generated; the ratio of the principal stress difference value to the distance of adjacent elements is calculated, the stress concentration position where the stress gradient exceeds the threshold value is marked, and the principal stress peak value is extracted; the principal stress peak value is counted, and the area where the material yield strength threshold value or the stress gradient exceeds the critical value is determined as the stress anomaly area. 5.The method of claim 1, wherein, The stress anomaly area and the wear morphology classification are associated and analyzed to generate a potential distribution interval of equipment failure risk, including: The spatial overlap area of the stress anomaly area and the wear morphology classification area is calculated, the correlation strength coefficient is determined, the overlap of severe concentrated wear and high stress anomaly area is marked, and the wear and stress coupling influence area distribution map is generated; the product of the wear depth and the stress peak value is calculated, the influence weight is determined, and the high-risk coupling area exceeding the threshold value is marked; the actual failure probability is calculated by combining historical failure data and operating condition parameters; according to the actual failure probability and the distribution number of the high-risk coupling area in space, the cumulative failure risk value per unit area is calculated, and the cumulative risk value is divided into low risk, medium risk, high risk and extremely high risk from low to high, corresponding to the potential distribution interval of equipment failure risk. 6.The method of claim 1, wherein, The product particle size distribution data of the potential distribution interval of equipment failure risk is collected in real time, the fluctuation characteristics of the particle size distribution are analyzed, and the key influencing factors of quality stability control are determined according to the product quality standard requirements and customer particle size specification requirements, including: In the high-risk and extremely high-risk period of the potential distribution interval, the particle size distribution data of the broken product is collected in real time, the mass percentage and risk level of each particle size section are recorded, and the particle size distribution time series is constructed; the mass percentage difference value of adjacent time points in the particle size distribution time series is calculated, the fluctuation amplitude sequence of each particle size section is determined, and the standard deviation of the screening level mass percentage is taken as the fluctuation characteristic value; the comprehensive quality deviation value is calculated by comparing the particle size distribution with the product quality standard, and the failure risk level, wear morphology category and stress anomaly degree when the comprehensive quality deviation value exceeds the threshold value are extracted to determine the key influencing factors of quality stability control. 7.The method of claim 1, wherein, The key influencing factors of quality stability control are adjusted, the sampling frequency and monitoring points of the Internet of Things sensors are adjusted, and the adjusted wear degree real-time data and particle size change information are obtained, including: According to the wear morphology category and the stress anomaly degree, the crushing cavity area with numerical value exceeding the threshold value is identified, the Internet of Things sensors are added, and high, medium and low priority monitoring points are divided; the differential sampling frequency of the high, medium and low priority points is set, and the adjusted sensor network wear degree real-time data and particle size change information are obtained. 8.The method of claim 1, wherein, The wear degree real-time data and the particle size change information are analyzed, combined with preliminary wear morphology change trend verification correction, the best maintenance time window of the equipment under the current working condition is predicted, the quality stability control parameter setting is updated, and an optimized operation state monitoring scheme is obtained, including: The wear degree real-time data and the particle size change information are analyzed, combined with preliminary wear morphology change trend verification correction, the best maintenance time window of the equipment under the current working condition is predicted, the quality stability control parameter setting is updated, and an optimized operation state monitoring scheme is obtained, including: The wear degree real-time data and the particle size change information are analyzed, combined with preliminary wear morphology change trend verification correction, the best maintenance time window of the equipment under the current working condition is predicted, the quality stability control parameter setting is updated, and an optimized operation state monitoring scheme is obtained, including: The wear degree real-time data and the particle size change information are analyzed, combined with preliminary wear morphology change trend verification correction, the best maintenance time window of the equipment under the current working condition is predicted, the quality stability control parameter setting is updated, and an optimized operation state monitoring scheme is obtained, including: The wear degree real-time data and the particle size change information are analyzed, combined with preliminary wear morphology change trend verification correction, the best maintenance time window of the equipment under the current working condition is predicted, the quality stability control parameter setting is updated, and an optimized operation state monitoring scheme is obtained, including: The wear degree real-time data and the particle size change information are analyzed, combined with preliminary wear morphology change trend verification correction, the best maintenance time window of the equipment under the current working condition is predicted, the quality stability control parameter setting is updated, and an optimized operation state monitoring scheme is obtained, including: The wear degree real-time data and the particle size change information are analyzed, combined with preliminary wear morphology change trend verification correction, the best maintenance time window of the equipment under the current working condition is predicted, the quality stability control parameter setting is updated, and an optimized operation state monitoring scheme is obtained, including: The wear degree real-time data and the particle size change information are analyzed, combined with preliminary wear morphology change trend verification correction, the best maintenance time window of the equipment under the current working condition is predicted, the quality stability control parameter setting is updated,

Images