Method for evaluating bearing capacity in manufacturing process of safety handrail based on industrial internet of things

Abstract

The invention provides a bearing capacity assessment method for a safety handrail manufacturing process based on an industrial internet of things, and the method comprises the steps: reading a groove spacing between adjacent knurling grooves according to a texture depth value, dividing the groove spacing by the texture depth value to obtain a groove spacing depth ratio, and recognizing a knurling section with the groove spacing depth ratio lower than a safety threshold value as a stress superposition risk section; extracting an energy peak value of an acoustic emission signal in the stress superposition risk section and a maximum strain gradient in the pipe wall strain distribution data, and performing weighted fusion to obtain groove root stress superposition strength; extracting position coordinates of each measuring point entering yield for the first time from the strain distribution data, and identifying a transformation degree of a yield initial position from random distribution to linear arrangement along the knurling groove bottom by adopting a support vector machine classification algorithm to obtain a yield position concentration ratio; and according to the yield position concentration ratio and the groove root stress superposition strength, dividing a damage grade corresponding to the texture depth value, and marking the knurled section with the damage grade exceeding a preset warning grade as the initial position of the weak section.

Description

Technical Field

[0001] This invention relates to the field of information technology, and in particular to a method for assessing the load-bearing capacity of a safety handrail manufacturing process based on the Industrial Internet of Things. Background Technology

[0002] The application of Industrial Internet of Things (IIoT) technology in manufacturing is becoming increasingly widespread, especially in the production of load-bearing metal tubing such as safety handrails. Their quality directly impacts the safety of people in public places and their reliability. As everyday protective components, safety handrails must possess sufficient structural strength and surface anti-slip properties to withstand the load demands of various complex usage environments. Therefore, precise control of the manufacturing process is crucial for ensuring product safety. Currently, most manufacturers use traditional knurling processes to increase the surface friction coefficient of the tubing to prevent accidental slippage due to wet hands. However, this seemingly reasonable process choice introduces hidden structural damage. While the grooves formed during knurling increase grip, they also remove some material from the tubing wall, reducing the effective load-bearing area of ​​the cross-section. More importantly, as the knurling depth increases, the spacing and depth ratio between adjacent grooves changes. The stress concentration at the bottom of the grooves is no longer isolated but exhibits significant mutual influence and superposition effects. This superposition of stress fields causes the originally randomly distributed tiny yield points to gradually concentrate towards the bottom of the knurled grooves and connect into a line when the tubing is subjected to external forces, resulting in a significant decrease in the overall resistance of the tubing. For example, when producing a batch of safety handrails of the same specifications, if the knurling depth is increased from 0.3 mm to 0.6 mm to achieve a higher anti-slip rating, the surface friction performance is indeed improved. However, in standard load tests, some tubes show visible plastic deformation prematurely, and even local collapse occurs far below the design load capacity. The reason for this is that with the increased texture depth, the ratio of groove spacing to depth deviates from the reasonable range. This causes the stress peaks at the roots of adjacent grooves to approach and intensify each other. The damage evolution path in the weak areas of the tube wall changes from dispersed to a linear arrangement concentrated along the bottom of the groove, thus significantly weakening the load-bearing capacity. This inherent contradiction between strength and anti-slip performance caused by adjustments in process parameters makes it difficult for manufacturers to accurately predict and control the true remaining load-bearing capacity of the tube while pursuing higher anti-slip ratings. Therefore, while increasing the depth of the knurled texture to bring better grip and anti-slip effect, how to effectively identify and quantify the degree of weakening of the tube's load-bearing capacity due to the stress superposition caused by the change in the groove spacing and depth ratio, and how to capture the transformation characteristics of tube wall damage from random distribution to linear concentration at the groove bottom in real time during the production process, have become key issues to ensure reliable assessment of the load-bearing capacity during the manufacturing process of safety handrails. Summary of the Invention

[0003] This invention provides a method for load-bearing capacity assessment in the manufacturing process of safety handrails based on the Industrial Internet of Things, comprising: By collecting acoustic emission sensing units deployed on the safety handrail manufacturing production line during knurling, and simultaneously reading strain distribution data at various measuring points on the pipe wall by the strain acquisition module, texture depth values ​​are extracted from the acoustic signals, and the effective load-bearing cross-sectional area reduction rate of the pipe wall is identified from the strain distribution data. The groove spacing between adjacent knurling grooves is read based on the texture depth value. The groove spacing is divided by the texture depth value to obtain the groove spacing depth ratio. Knurling sections with a groove spacing depth ratio lower than the safety threshold are identified as stress superposition risk sections. The energy peak value of the acoustic emission signal in the stress superposition risk section and the maximum strain gradient in the pipe wall strain distribution data are extracted and weighted to obtain the stress superposition intensity of the trench root. The coordinates of the first yield point are extracted from the strain distribution data. The support vector machine classification algorithm is used to identify the degree of change of the yield initiation position from random distribution to linear arrangement along the knurling groove, and the yield position concentration is obtained. Based on the yield position concentration and the superposition strength of the groove root stress, the damage level corresponding to the texture depth value is divided, and the knurled section with the damage level exceeding the preset warning level is marked as the initial position of the weak section. Extract the reduction rate of the load-bearing cross-sectional area at the initial location of the weak section and the ratio of the slot spacing depth. Analyze the case where the reduction rate of the load-bearing cross-sectional area exceeds the critical load-bearing threshold and the ratio of the slot spacing depth is lower than the safety threshold, and confirm that the initial location of the weak section is the target weak section location. Extract the residual wall thickness of the pipe body at the target weak section location and the superimposed stress intensity of the trench root, aggregate and optimize the residual wall thickness and the superimposed stress intensity of the trench root, construct a bearing capacity assessment model, and output the remaining bearing capacity level of the pipe body.

[0004] Furthermore, the process of acquiring acoustic wave signals during knurling by an acoustic emission sensing unit deployed on the safety handrail manufacturing line, simultaneously reading strain distribution data at various measuring points on the pipe wall by a strain acquisition module, extracting texture depth values ​​from the acoustic wave signals, and identifying the reduction rate of the effective load-bearing cross-sectional area of ​​the pipe wall from the strain distribution data includes: By deploying acoustic emission sensing units on the safety handrail manufacturing production line, acoustic signals are continuously collected during the processing stage when the knurling cutter cuts into the pipe wall. The acoustic signals are divided into frequency bands according to the frequency range. The pulse vibration amplitude generated by the periodic contact between the knurling cutter teeth and the pipe wall is identified from the high frequency band. The texture depth value of the knurling groove is calculated based on the peak-valley spacing of the pulse vibration amplitude. The interval of the measurement points of the strain acquisition module is determined according to the texture depth value. Multiple strain gauges are set along the circumferential and axial directions of the pipe wall to collect the strain values ​​generated by each measurement point during the knurling process. The difference in strain values ​​between adjacent measurement points is calculated, and the area where the difference in strain values ​​exceeds the preset difference threshold is identified as the concentrated area of ​​wall thickness reduction. Read the current wall thickness measurement value corresponding to each measuring point in the wall thickness reduction concentration area, subtract the current wall thickness measurement value from the original wall thickness of the pipe to obtain the wall thickness reduction amount, divide the wall thickness reduction amount by the original wall thickness to obtain the wall thickness reduction ratio, and multiply the wall thickness reduction ratio by the original cross-sectional area of ​​the pipe to obtain the effective bearing cross-sectional area reduction rate of the pipe wall.

[0005] Furthermore, the step of reading the groove spacing between adjacent knurling grooves based on the texture depth value, dividing the groove spacing by the texture depth value to obtain the groove spacing-depth ratio, and identifying knurling sections with a groove spacing-depth ratio lower than a safety threshold as stress superposition risk sections includes: Based on the knurling groove position corresponding to the texture depth value, the groove spacing between two adjacent knurling grooves is extracted from the pulse period of the acoustic signal. The groove spacing is divided by the texture depth value to obtain the groove spacing depth ratio. The groove spacing depth ratio value corresponding to each knurling section is recorded segment by segment along the tube axis. For each knurled section, if the groove pitch depth ratio is lower than a preset safety threshold, the knurled section is marked as a stress superposition risk section.

[0006] Furthermore, the step of extracting the energy peak value of the acoustic emission signal within the stress superposition risk zone and the maximum strain gradient in the pipe wall strain distribution data, and then weighted and fused to obtain the stress superposition intensity at the trench root, includes: For the stress superposition risk section, the energy peak value is extracted from the acoustic emission signal within the corresponding time window of the section. The energy peak value is obtained by squaring the amplitude of the acoustic signal and then summing them. The energy peak value corresponding to the maximum acoustic energy in the section is recorded. From the strain distribution data of the pipe wall in the stress superposition risk zone, the strain value difference between adjacent measuring points is read, the strain value difference is divided by the distance between measuring points to obtain the strain gradient, and the maximum value of the strain gradient in the relevant zone is identified as the maximum strain gradient. The peak energy and the maximum strain gradient are divided by their respective upper limits to obtain normalized values. The normalized values ​​are then weighted and summed according to preset energy weighting coefficients and strain weighting coefficients to obtain the stress superposition intensity at the root of the stress superposition risk zone.

[0007] Furthermore, the step of extracting the coordinates of the first yield point from the strain distribution data, and using a support vector machine classification algorithm to identify the degree of change in the yield initiation position from a random distribution to a linear arrangement along the knurling groove, yields the yield position concentration, including: The strain values ​​of each measuring point are read from the strain distribution data point by point. The strain values ​​are compared with the yield strain threshold of the material corresponding to the measuring point. If the strain value of the measuring point reaches or exceeds the yield strain threshold for the first time, the position coordinates of the measuring point on the pipe wall are recorded to obtain the yield initiation position coordinate sequence. For the yield initiation position coordinate sequence, the lateral offset distance of each coordinate point relative to the center line of the knurled groove bottom is read along the pipe body axis, and the lateral offset distance is combined to form the distribution feature vector of each yield initiation position; The spatial dispersion between each yield initiation position is calculated based on the distribution feature vector. The spatial dispersion is obtained by statistically analyzing the variance of the distance from each coordinate point to the center line of the knurling groove bottom. The spatial dispersion is used as a quantitative indicator to characterize the distribution pattern of the yield initiation position. A support vector machine classification algorithm is adopted, using the distribution feature vector and the spatial dispersion as input features, and pre-labeled random distribution samples and linearly arranged samples as training data to obtain the classification boundary that distinguishes between random distribution and linear arrangement. Based on the classification boundary, the confidence score of the current yield initiation position sequence belonging to the linear arrangement is output, and the confidence score is used as the yield position concentration.

[0008] Furthermore, the step of classifying the damage level corresponding to the texture depth value based on the yield position concentration and the superimposed stress intensity at the groove root, and marking the knurled section with the damage level exceeding the preset warning level as the initial location of the weak section, includes: Based on the yield position concentration and the stress superposition intensity of the groove root, a two-dimensional damage assessment coordinate system is established, with the yield position concentration as the horizontal axis value and the stress superposition intensity of the groove root as the vertical axis value. The location points of each knurled section are marked in the two-dimensional damage assessment coordinate system. Multiple damage level regions are pre-defined in the two-dimensional damage assessment coordinates. The damage level of each knurled section is determined based on the damage level region into which its position point falls. For each knurled section, if the damage level exceeds the preset warning level, the knurled section is marked as a preliminary weak section location.

[0009] Furthermore, the step of extracting the ratio of the reduction rate of the load-bearing cross-sectional area at the initial location of the weak section to the ratio of the slot spacing depth, analyzing the case where the reduction rate of the load-bearing cross-sectional area exceeds the critical load-bearing threshold and the ratio of the slot spacing depth is lower than the safety threshold, and confirming the initial location of the weak section as the target weak section location, includes: For the initial location of the weak section, the reduction rate of the load-bearing cross-sectional area and the ratio of the slot spacing depth corresponding to the section are extracted. It is determined that the reduction rate of the load-bearing cross-sectional area exceeds the preset critical load threshold, and at the same time, it is determined that the ratio of the slot spacing depth is lower than the preset safety threshold. If the reduction rate of the load-bearing cross-sectional area exceeds the critical load-bearing threshold and the groove spacing depth ratio is lower than the safety threshold, then the preliminary location of the weak section is confirmed as the target weak section location.

[0010] Furthermore, the extraction of the residual wall thickness of the pipe at the target weak section location and the superimposed stress intensity of the trench root, the aggregation and optimization of the residual wall thickness and the superimposed stress intensity of the trench root, the construction of a bearing capacity assessment model, and the output of the remaining bearing capacity level of the pipe include: For the target weak section, the residual wall thickness of the section is extracted from the pipe wall strain distribution data. The residual wall thickness is the remaining thickness value after subtracting the wall thickness reduction from the original wall thickness of the pipe body. At the same time, the stress superposition intensity of the groove root corresponding to the section is read. The residual wall thickness and the stress superposition strength of the trench root are subjected to aggregate optimization processing. The residual wall thickness is divided by the original wall thickness of the pipe to obtain the wall thickness retention ratio. The wall thickness retention ratio is multiplied by the normalized compensation value of the stress superposition strength of the trench root to obtain the comprehensive bearing index. The normalized compensation value is a normalized value minus the stress superposition strength of the trench root. Read the texture depth value, slot spacing depth ratio, and yield position concentration corresponding to the location of the target weak section, and combine the texture depth value, slot spacing depth ratio, yield position concentration, and comprehensive bearing capacity index to form a bearing capacity assessment input vector; The random forest regression algorithm is used, with the bearing capacity assessment input vector as the input feature and the pre-labeled pipe bearing capacity test data as the training label for training. It outputs continuous bearing capacity prediction values ​​and maps the bearing capacity prediction values ​​to the remaining bearing capacity level of the pipe according to the preset bearing capacity level division interval.

[0011] The technical solutions provided by the embodiments of the present invention may include the following beneficial effects: This invention discloses a load-bearing capacity assessment method for the manufacturing process of safety handrails based on the Industrial Internet of Things (IIoT). Addressing the core business problem of uneven texture depth, excessively small groove spacing leading to severe stress superposition at the groove root, and premature yielding concentration during knurling, which significantly reduces the load-bearing capacity of the pipe body, this invention uses an acoustic emission sensing unit to collect knurling acoustic signals in real time and extract texture depth values. Simultaneously, it collects pipe wall strain distribution data, calculates the groove spacing-to-depth ratio to identify stress superposition risk sections, and then integrates the acoustic emission energy peak value and the maximum strain gradient to quantify the stress superposition intensity at the groove root. Furthermore, a support vector machine algorithm is used to analyze the degree of change from random yield initiation position to linear concentration at the groove root, obtaining the yield position concentration degree. Based on this, and together with the stress superposition intensity at the groove root, damage levels are classified, and target weak sections with a critical reduction rate in load-bearing cross-sectional area and a groove spacing-to-depth ratio below the safety threshold are identified. Finally, a load-bearing capacity assessment model is constructed using texture depth value, groove spacing-to-depth ratio, and yield position concentration degree as inputs to accurately output the remaining load-bearing capacity level of the pipe body. This invention enables online damage detection, precise location of weak sections, and dynamic assessment of remaining load-bearing capacity throughout the knurling process, effectively avoiding the risk of early failure of safety handrails caused by knurling defects, and significantly improving the product's structural reliability and the quality control level of the manufacturing process. Attached Figure Description

[0012] Figure 1 This is a flowchart of a load-bearing capacity assessment method for the manufacturing process of a safety handrail based on the Industrial Internet of Things, according to the present invention.

[0013] Figure 2 This is a schematic diagram of a load-bearing capacity assessment method for the manufacturing process of a safety handrail based on the Industrial Internet of Things, according to the present invention.

[0014] Figure 3 This is another schematic diagram of a load-bearing capacity assessment method for the manufacturing process of a safety handrail based on the Industrial Internet of Things according to the present invention. Detailed Implementation

[0015] To further understand the content of this invention, a detailed description of the invention is provided in conjunction with the accompanying drawings and embodiments. The specific embodiments described herein are for illustrative purposes only and are not intended to limit the invention. It should also be noted that, for ease of description, only the parts relevant to the invention are shown in the accompanying drawings.

[0016] like Figures 1-3 This embodiment of a load-bearing capacity assessment method for the manufacturing process of safety handrails based on the Industrial Internet of Things may specifically include: S101. Acoustic emission sensing units deployed on the safety handrail manufacturing line collect acoustic signals during knurling, and strain acquisition modules simultaneously read strain distribution data at various measuring points on the pipe wall. Texture depth values ​​are extracted from the acoustic signals, and the effective load-bearing cross-sectional area reduction rate of the pipe wall is identified from the strain distribution data.

[0017] Acoustic emission sensing units deployed on the safety handrail manufacturing line continuously collect acoustic signals during the knurling process where the knurling cutter cuts into the pipe wall. These signals are divided into frequency bands, and the pulse vibration amplitude generated by the periodic contact between the knurling cutter teeth and the pipe wall is identified from the high-frequency band. The texture depth of the knurling groove is calculated based on the peak-to-valley spacing of the pulse vibration amplitude. The spacing of the strain acquisition module's measuring points is determined based on the texture depth value. Multiple strain gauges are set along the circumferential and axial directions of the pipe wall to collect strain values ​​generated at each measuring point during the knurling process. The difference in strain values ​​between adjacent measuring points is calculated, and areas where the strain value difference exceeds a preset difference threshold are identified as concentrated wall thickness reduction areas. For these concentrated wall thickness reduction areas, the current wall thickness measurement value corresponding to each measuring point within the area is read. The original wall thickness is subtracted from the current wall thickness measurement value to obtain the wall thickness reduction amount. The wall thickness reduction amount is divided by the original wall thickness to obtain the wall thickness reduction ratio. The wall thickness reduction ratio is multiplied by the original cross-sectional area of ​​the pipe to obtain the effective load-bearing cross-sectional area reduction rate of the pipe wall.

[0018] At the knurling station of the safety handrail manufacturing line, an acoustic emission sensing unit is fixedly installed on the tube surface at a preset distance from the knurling cutter. When the cutting teeth of the knurling cutter cut into the tube wall to form a groove, the periodic contact between the cutting teeth and the metal material will generate high-frequency pulse vibration. This pulse vibration propagates along the tube wall in the form of sound waves and is captured by the sensing unit.

[0019] Specifically, the raw acoustic wave signal output by the acoustic emission sensing unit contains multiple frequency components. The high-frequency band corresponds to the impact response generated when the blade cuts into the pipe wall, while the low-frequency band corresponds to the bending vibration generated by the overall force on the pipe. By dividing the acoustic wave signal into frequency bands and separating the high-frequency band signal, the pulse vibration amplitude sequence is extracted.

[0020] In one embodiment, each time the knurling cutter completes a cutting motion, a peak amplitude appears in the high-frequency signal, and a trough amplitude is formed between two adjacent cutting motions. The peak amplitude of the pulse vibration is positively correlated with the depth of the cutter cutting into the tube wall, and the peak-to-trough spacing reflects the time interval between the cutter cutting. Based on the rotational speed parameters of the knurling cutter and the time length of the peak-to-trough spacing, combined with the calibration curve of peak amplitude versus material cutting depth, the texture depth value of the current knurling groove is calculated. The formula for calculating the texture depth value is: D texture =f(A peak )=k×A peak +b, where Dtexture A represents the texture depth value (unit: mm). peak denoted as peak amplitude of the pulse vibration (unit: mV), k is the slope of the calibration curve, and b is the intercept. This calibration curve is obtained by pre-calibrating using a standard sample of known depth.

[0021] For example, when the peak amplitude of the pulse vibration increases, it indicates that the cutting depth of the blade increases, and the corresponding texture depth value increases accordingly.

[0022] In one possible implementation, the strain acquisition module determines the spacing of the measuring points based on the texture depth value. The larger the texture depth value, the wider the stress influence range between adjacent knurled grooves, and the spacing of the measuring points is correspondingly reduced to increase the monitoring density. Several sets of strain gauges are uniformly arranged along the circumference of the pipe wall, and multiple measurement sections are set along the axial direction of the pipe body within the knurled section to form a strain measuring point array covering the entire knurled processing area.

[0023] It should be noted that each measuring point outputs strain values ​​in real time during the knurling process. The difference in strain values ​​between adjacent measuring points reflects the degree of change in the pipe wall thickness in that area. When the knurling groove cuts deeper, the local thickness of the pipe wall decreases, and the strain value at this measuring point is significantly higher than that at surrounding measuring points, thus increasing the difference in strain values ​​between adjacent measuring points. The difference in strain values ​​is compared with a preset difference threshold. Areas exceeding this threshold are identified as concentrated areas of wall thickness reduction, indicating that the thickness loss of the pipe wall due to knurling is significant in these areas. For these concentrated areas of wall thickness reduction, the current wall thickness measurement value corresponding to each measuring point within the area is read. The original wall thickness before processing is subtracted from the current wall thickness measurement value to obtain the wall thickness reduction amount at each measuring point. The wall thickness reduction amount is divided by the original wall thickness to obtain the wall thickness reduction ratio. Multiplying the wall thickness reduction ratio by the original cross-sectional area of ​​the pipe body yields the reduction rate of the effective load-bearing cross-sectional area of ​​the pipe wall in this area, which is used to characterize the degree of weakening of the pipe's load-bearing capacity by the knurling process.

[0024] S102. Read the groove spacing between adjacent knurling grooves based on the texture depth value, divide the groove spacing by the texture depth value to obtain the groove spacing depth ratio, and identify knurling sections with a groove spacing depth ratio lower than the safety threshold as stress superposition risk sections.

[0025] Based on the knurling groove positions corresponding to the texture depth value, the groove spacing between two adjacent knurling grooves is extracted from the pulse period of the acoustic signal. The groove spacing is divided by the texture depth value to obtain the groove spacing depth ratio. The groove spacing depth ratio values ​​corresponding to each knurling segment are recorded segment by segment along the pipe axis. For the groove spacing depth ratio value of each knurling segment, it is determined whether the groove spacing depth ratio is lower than a preset safety threshold. If the groove spacing depth ratio is lower than the safety threshold, the knurling segment is marked as a stress superposition risk segment.

[0026] During the knurling process, the knurling tool's teeth cut into the pipe wall at fixed intervals to form continuous grooves. The distance between two adjacent knurling grooves is the groove spacing. In the acoustic emission sensor unit, each time the teeth cut into the pipe wall, a pulse peak is generated. Based on the time interval between adjacent pulse peaks and the feed rate of the knurling tool, the corresponding groove spacing value is calculated.

[0027] It should be noted that the groove spacing-to-depth ratio reflects the relative relationship between the material thickness and groove depth between adjacent knurled grooves. When this ratio is large, the wall material retained between adjacent grooves is thicker, and the stress on each groove bottom area is relatively independent; when this ratio decreases to a certain extent, the stress-affected areas at the bottom of adjacent grooves begin to overlap, resulting in stress superposition.

[0028] In one embodiment, the knurling area is divided into several sections along the tube axis, and the groove spacing depth ratio is calculated for each section and compared with a preset safety threshold. Sections below the safety threshold are marked as stress superposition risk sections, indicating that there may be stress interaction between adjacent knurling grooves in the section.

[0029] S103. Extract the peak energy of the acoustic emission signal in the stress superposition risk section and the maximum strain gradient in the pipe wall strain distribution data, and weighted fuse them to obtain the stress superposition intensity of the trench root.

[0030] For the stress superposition risk section, the energy peak value is extracted from the acoustic emission signal within the corresponding time window of the section. The energy peak value is obtained by squaring the amplitude of the acoustic wave signal and summing them. The energy peak value corresponding to the maximum acoustic wave energy in the section is recorded. From the strain distribution data of the pipe wall in the stress superposition risk section, the strain value difference between adjacent measuring points is read. The strain value difference is divided by the distance between measuring points to obtain the strain gradient. The maximum value of the strain gradient in the section is identified as the maximum strain gradient. The energy peak value and the maximum strain gradient are divided by their respective upper limits to obtain normalized values. The normalized values ​​are weighted and summed according to preset energy weighting coefficients and strain weighting coefficients. The sum of the energy weighting coefficient and the strain weighting coefficient is one to obtain the stress superposition intensity of the trench root in the stress superposition risk section.

[0031] Within the stress superposition risk zone, the acoustic emission signal reflects the microscopic deformation and internal stress release state of the tube wall material during the knurling process. When the knurling tool cuts into the tube wall, the lattice slip and dislocation movement inside the material generate elastic waves, which are captured by the acoustic emission sensing unit to form an acoustic signal.

[0032] Specifically, the process of extracting the energy peak involves squaring the amplitude of the acoustic signal point by point within the time window and then summing the results along the time axis. The summation result reflects the total energy carried by the acoustic signal during that period. The moment when the acoustic energy reaches its maximum value during the entire processing corresponds to the stage where the stress release of the pipe wall material is most concentrated, and the value recorded at this time is the energy peak.

[0033] In one embodiment, the strain gradient characterizes the degree of drastic change in strain state between adjacent locations on the pipe wall. After reading the strain values ​​of each measuring point within the stress superposition risk zone, the difference between the strain values ​​of two adjacent measuring points is calculated and then divided by the physical distance between the two measuring points to obtain the strain gradient at that location. This process is repeated for all adjacent measuring point pairs within the zone, and the maximum strain gradient among them is taken as the maximum strain gradient.

[0034] It should be noted that the peak energy and the maximum strain gradient have different dimensions and numerical ranges, and direct addition cannot reflect their relative contributions. Normalization involves dividing the peak energy by the upper limit of the energy range of the acoustic emission sensing unit and dividing the maximum strain gradient by the upper limit of the gradient range of the strain acquisition module, so that both values ​​fall within the range of zero to one.

[0035] In one possible implementation, the energy weighting coefficient and the strain weighting coefficient are determined based on the sensitivity of the knurling process to the acoustic response and strain response. The sum of the two is one. The normalized energy peak value and the maximum strain gradient are multiplied by the corresponding weighting coefficients and then added together to obtain the stress superposition intensity at the groove root. This value comprehensively reflects the degree of mutual influence of stress in the bottom areas of adjacent knurled grooves within the stress superposition risk zone.

[0036] S104. Extract the coordinates of the first yield point from the strain distribution data, and use the support vector machine classification algorithm to identify the degree of change of the yield initiation position from random distribution to linear arrangement along the knurling groove bottom line, so as to obtain the yield position concentration.

[0037] The strain values ​​at each measuring point are read from the strain distribution data. These strain values ​​are compared with the yield strain threshold of the material corresponding to that measuring point. If the strain value at a measuring point reaches or exceeds the yield strain threshold for the first time, the coordinates of that measuring point on the pipe wall are recorded, resulting in a yield initiation position coordinate sequence. For this yield initiation position coordinate sequence, the lateral offset distance of each coordinate point relative to the center line of the knurled groove bottom is read along the pipe axis. These lateral offset distances are combined to form a distribution feature vector for each yield initiation position. The spatial dispersion between each yield initiation position is calculated based on the distribution feature vector. This spatial dispersion is obtained by statistically analyzing the variance of the distance from each coordinate point to the center line of the knurled groove bottom, and is used as a quantitative indicator characterizing the distribution pattern of the yield initiation positions. A support vector machine (SVM) classification algorithm is used, with the distribution feature vector and spatial dispersion as input features, and pre-labeled randomly distributed samples and linearly arranged samples as training data. A classification boundary distinguishing between random distribution and linear arrangement is obtained. Based on the classification boundary, the confidence score that the current yield initiation position sequence belongs to a linear arrangement is output, and this confidence score is used as the yield position concentration.

[0038] During the manufacturing process of safety handrails, the tube wall material gradually accumulates strain under the continuous action of knurling. When the strain value at a certain measuring point reaches the yield limit of the material, that location begins to enter the plastic deformation stage. The yield strain threshold is a predetermined value based on the characteristics of the metal material used in the tube; different tube materials correspond to different yield strain thresholds.

[0039] Specifically, the strain acquisition module monitors the strain value changes at each measuring point in real time, comparing the current strain value with the preset yield strain threshold point by point. When the strain value at a measuring point first reaches or exceeds the yield strain threshold, the position information of that measuring point in the pipe wall coordinate system is recorded, including its position along the pipe axis and its position along the circumference. After traversing all measuring points, a yield initiation position coordinate sequence is formed, which contains all the position points that first enter the yield state during the knurling process.

[0040] In one embodiment, the distribution feature vector is constructed based on the spatial relationship between the yield initiation position and the centerline of the knurled groove bottom. The centerline of the knurled groove bottom is a virtual reference line extending along the pipe body axis, located on the line connecting the deepest recesses of each knurled groove. For each coordinate point in the yield initiation position coordinate sequence, its lateral offset distance from the centerline of the knurled groove bottom is calculated. This lateral offset distance reflects the degree to which the yield position deviates from the knurled groove bottom. The lateral offset distances of all yield initiation positions are arranged and combined in positional order to form the distribution feature vector.

[0041] It should be noted that spatial dispersion is a statistical indicator used to quantify the distribution pattern of yield initiation positions. When the yield initiation positions are randomly distributed, the distances from each coordinate point to the center line of the knurling groove bottom vary greatly, resulting in a high variance. When the yield initiation positions are linearly arranged along the knurling groove bottom, the coordinate points are concentrated near the center line, resulting in a low variance. The calculation process for spatial dispersion involves taking the arithmetic mean of all lateral offset distances in the distribution feature vector, then calculating the sum of the squares of the differences between each lateral offset distance and this mean, and finally dividing by the total number of coordinate points to obtain the variance value.

[0042] In one possible implementation, the training process of the Support Vector Machine (SVM) classification algorithm relies on pre-labeled sample data. Randomly distributed samples are coordinate sequences of yield positions selected from historical processing records, characterized by high spatial dispersion and a large range of lateral offset distance fluctuations. Linearly arranged samples are coordinate sequences of yield positions concentrated along the bottom of the knurling groove, selected from historical processing records, characterized by low spatial dispersion and a relatively small range of lateral offset distances. The SVM classification algorithm uses both the distribution feature vector and spatial dispersion as input features, and the sample class label as a supervision signal. During training, the SVM finds a hyperplane to separate the randomly distributed samples from the linearly arranged samples; this hyperplane is the classification boundary. The location of the classification boundary is determined by the support vectors, which are the sample points closest to the classification boundary. After training, for a new yield initiation position coordinate sequence, its distribution feature vector and spatial dispersion are input into the trained SVM. The confidence score of whether the input point belongs to the linearly arranged category is calculated based on the distance from the input point to the classification boundary. A higher confidence score indicates that the yield positions are more concentrated near the bottom of the knurling groove.

[0043] For example, after the knurling of a batch of safety handrails is completed, the yield initiation position coordinate sequence of the tube is extracted and a distribution feature vector is constructed. After calculating the spatial dispersion, the vector is input into a support vector machine classification algorithm. If the output confidence value is high, it indicates that the yield initiation position of the tube has changed from a random distribution to a linear arrangement along the bottom of the knurling groove, and the yield position concentration is high.

[0044] Understandably, the concentration of yield points reflects the degree to which damage to the pipe wall material evolves from a dispersed state to a concentrated state. In the early stages of knurling, strain accumulation is relatively uniform throughout the pipe wall, and the yield initiation points are randomly distributed throughout the pipe wall. As the knurling depth increases, the stress concentration effect in the groove bottom region gradually becomes apparent, and the yield initiation points begin to converge towards the bottom of the knurling groove, exhibiting a linear arrangement.

[0045] S105. Based on the concentration of yield position and the superposition strength of groove root stress, classify the damage level corresponding to the texture depth value, and mark the knurled section with the damage level exceeding the preset warning level as the initial position of the weak section.

[0046] Based on the yield point concentration and the superimposed stress intensity at the groove root, a two-dimensional damage assessment coordinate system is established, with the yield point concentration as the horizontal axis and the superimposed stress intensity at the groove root as the vertical axis. The location points of each knurled section are marked on this coordinate system. Multiple damage level regions are pre-defined on the coordinate system, including a minor damage region near the origin, a moderate damage region in the middle of the coordinate system, and a severe damage region away from the origin. The damage level of each knurled section is determined based on the damage level region its location point falls into at the current texture depth. For each knurled section, it is determined whether the damage level exceeds a preset warning level. If the damage level exceeds the preset warning level, the knurled section is marked as a preliminary weak point.

[0047] Yield location concentration and groove root stress superposition intensity are two indicators reflecting the damage state of the pipe wall from different dimensions. Yield location concentration characterizes the spatial distribution of the yield initiation position of the pipe wall; a higher value indicates that the yield position is more concentrated at the bottom of the knurled groove. Groove root stress superposition intensity characterizes the degree of mutual influence of stress in the bottom areas of adjacent knurled grooves; a higher value indicates a more significant stress superposition effect.

[0048] In one embodiment, when establishing a two-dimensional damage assessment coordinate system, a Cartesian coordinate system is formed with the yield position concentration as the horizontal axis and the groove root stress superposition intensity as the vertical axis. For each knurled section, its yield position concentration value and groove root stress superposition intensity value are read, and the corresponding position points are marked in this coordinate system, thereby visually presenting the damage state of each knurled section in a two-dimensional plane.

[0049] It should be noted that the division of damage level zones is pre-defined in a two-dimensional damage assessment coordinate system. The minor damage zone is located near the origin, corresponding to a low level of yield strength concentration and trench root stress superposition intensity, indicating relatively mild pipe wall damage. The moderate damage zone is located in the middle of the coordinate system, corresponding to a moderate level of both indicators. The severe damage zone is located away from the origin, corresponding to a high level of yield strength concentration and trench root stress superposition intensity, indicating relatively severe pipe wall damage. The boundaries of the three zones are pre-determined based on material properties and process requirements.

[0050] Specifically, when a knurled section falls within a moderately or severely damaged area, a judgment is made based on a preset warning level. If the damage level of the knurled section exceeds the preset warning level, it is marked as a preliminary weak section location, indicating that the section is at risk of reduced load-bearing capacity at the current texture depth value.

[0051] S106. Extract the ratio of the reduction rate of the load-bearing cross-sectional area to the depth ratio of the slot spacing at the initial location of the weak section, and analyze whether the reduction rate of the load-bearing cross-sectional area exceeds the critical load-bearing threshold and whether the depth ratio of the slot spacing is lower than the safety threshold. If so, the initial location of the weak section is confirmed as the target weak section location.

[0052] For the initial location of the weak section, the reduction rate of the load-bearing cross-sectional area and the ratio of the slot spacing depth corresponding to that section are extracted. It is determined whether the reduction rate of the load-bearing cross-sectional area exceeds a preset critical load-bearing threshold, and simultaneously, whether the ratio of the slot spacing depth is lower than a preset safety threshold. If the reduction rate of the load-bearing cross-sectional area exceeds the critical load-bearing threshold and the ratio of the slot spacing depth is lower than the safety threshold, then the initial location of the weak section is confirmed as the target weak section location.

[0053] The initial location of weak sections is the knurled section with potential load-bearing risk, screened according to the damage level. For each initial location of a weak section, the reduction rate of the load-bearing cross-sectional area and the groove spacing depth ratio of that section are read from the previous processing results. These two parameters reflect the degree of material loss of the pipe wall and the degree of stress influence between adjacent knurled grooves, respectively.

[0054] In one embodiment, the critical load-bearing threshold is a pre-set value based on the design load-bearing requirements of the safety handrail. When the reduction rate of the load-bearing cross-sectional area exceeds this threshold, it indicates that the effective load-bearing cross-section of the pipe wall has been reduced to a degree that may affect the overall load-bearing capacity. The safety threshold is a pre-set value based on the condition that the stresses between the knurled grooves are independent. When the groove spacing depth ratio is lower than this threshold, it indicates that the stress-affected areas at the bottom of adjacent knurled grooves have overlapped.

[0055] It should be noted that the initial location of the weak section is only confirmed as the target weak section location when both conditions are met simultaneously: the reduction rate of the bearing cross-sectional area exceeds the critical bearing threshold and the ratio of the slot spacing to the depth is lower than the safety threshold. This indicates that the section simultaneously presents two risk factors: excessive cross-sectional loss and significant stress superposition effect.

[0056] S107. Extract the residual wall thickness and groove root stress superposition strength of the pipe body at the target weak section location, aggregate and optimize the residual wall thickness and groove root stress superposition strength, construct a bearing capacity assessment model with texture depth value, groove spacing depth ratio and yield position concentration as input, and output the residual bearing capacity level of the pipe body.

[0057] For the target weak section, the residual wall thickness of that section is extracted from the pipe wall strain distribution data. This residual wall thickness is the remaining thickness after subtracting the wall thickness reduction from the original pipe wall thickness. Simultaneously, the stress superposition intensity at the groove root corresponding to that section is read. The residual wall thickness and the stress superposition intensity at the groove root are aggregated and optimized. The residual wall thickness is divided by the original pipe wall thickness to obtain the wall thickness retention ratio. This retention ratio is multiplied by the normalized complement of the stress superposition intensity at the groove root to obtain the comprehensive bearing capacity index. The normalized complement is a normalized value minus the stress superposition intensity at the groove root. The texture depth value, groove spacing depth ratio, and yield position concentration corresponding to the target weak section are read. These values ​​are combined with the comprehensive bearing capacity index to form a bearing capacity assessment input vector. The random forest regression algorithm is used, with the bearing capacity assessment input vector as the input feature and the pre-labeled pipe bearing capacity test data as the training label for training. It outputs continuous bearing capacity prediction values ​​and maps the bearing capacity prediction values ​​to the remaining bearing capacity level of the pipe according to the preset bearing capacity level division interval.

[0058] The target weak section is a knurled section identified as having a load-bearing risk after screening under dual threshold conditions. The dual threshold conditions refer to a screening logic where the reduction rate of the load-bearing cross-sectional area exceeds the critical load-bearing threshold and the groove spacing-to-depth ratio is lower than the safety threshold. For each target weak section, the residual wall thickness after processing is read from the pipe wall strain distribution data. The residual wall thickness is the actual wall thickness retained after the thickness reduction caused by material cutting during the knurling process.

[0059] In one embodiment, the aggregation optimization process merges two indicators with different physical meanings—residual wall thickness and trench root stress superposition intensity—into a single comprehensive load-bearing index. The wall thickness retention ratio is obtained by dividing the residual wall thickness by the original pipe wall thickness; this ratio reflects the degree of material retention, with a higher value indicating less material loss. The normalized value of the trench root stress superposition intensity has already been obtained in the previous processing. The normalization complement is obtained by subtracting this normalized value from 1; a higher normalization complement indicates a weaker stress superposition effect. Multiplying the wall thickness retention ratio by the normalization complement yields a dimensionless comprehensive load-bearing index, which is directly used in subsequent optimization models. This index simultaneously considers the combined impact of material retention and stress concentration on the pipe's load-bearing capacity, improving economic practicality.

[0060] It should be noted that the construction of the load-bearing capacity assessment input vector integrates multiple key factors affecting the pipe's load-bearing capacity into a unified feature representation. The texture depth value reflects the cutting depth of the knurling grooves, the groove spacing-to-depth ratio reflects the geometric relationship between adjacent knurling grooves, the yield position concentration reflects the spatial distribution of pipe wall damage, and the comprehensive load-bearing index reflects the combined state of material and stress. These four feature values ​​are arranged and combined in a fixed order to form a four-dimensional load-bearing capacity assessment input vector.

[0061] Specifically, the Random Forest Regression algorithm is a machine learning method based on decision tree ensembles. It outputs the final predicted value by constructing multiple decision trees and averaging their predictions. During the training phase, the load-bearing capacity assessment input vectors of each pipe in historical production records are used as input features, and the actual load-bearing capacity values ​​measured under standard load tests are used as training labels. The Random Forest Regression algorithm randomly selects multiple sample subsets from the training data, constructs a decision tree for each subset, and each decision tree randomly selects some features from four input features to calculate the splitting conditions when splitting at a node. After training, for a new target weak section, its load-bearing capacity assessment input vector is input into the trained random forest. Each decision tree outputs a load-bearing capacity prediction value, and the arithmetic mean of the prediction values ​​from all decision trees is taken to obtain the predicted load-bearing capacity value for that section.

[0062] In one possible implementation, the load-bearing capacity level division intervals are pre-defined based on the design load-bearing requirements of the safety handrail. The range of predicted load-bearing capacity values ​​is divided into several continuous intervals, each interval corresponding to a load-bearing capacity level. For example, from high to low, these intervals are divided into sufficient load-bearing capacity level, normal load-bearing capacity level, marginal load-bearing capacity level, and insufficient load-bearing capacity level. Based on the interval into which the predicted load-bearing capacity values ​​output by the random forest regression algorithm fall, the remaining load-bearing capacity level of the pipe body corresponding to the target weak section is determined.

[0063] For example, after a batch of safety handrails has completed the knurling process, the residual wall thickness and the stress superposition strength of the groove root are extracted for the identified target weak section. After calculating the comprehensive bearing index, it is combined with the texture depth value, groove spacing depth ratio, and yield position concentration to form a bearing capacity assessment input vector. The vector is then input into the random forest regression algorithm to obtain the bearing capacity prediction value. The remaining bearing capacity level of the pipe body is output according to the level division interval.

[0064] Understandably, the output of the remaining load-bearing capacity rating of the tube can be used as a basis for judging the quality of the safety handrail. Tubes with lower load-bearing capacity ratings should be further processed or marked before being put into use, so as to achieve a quantitative assessment of the load-bearing capacity changes caused by knurling during the manufacturing process.

[0065] If the technical solution of this application involves the collection, storage, use, processing, transmission, provision, disclosure, or deletion of personal information, the products using this technical solution have clearly and understandably informed the users of the personal information processing rules before processing personal information, and have obtained the individuals' voluntary consent in accordance with the law. If the technical solution of this application involves sensitive personal information (such as biometrics, religious beliefs, specific identities, medical and health information, financial accounts, and location tracking), the products using this solution have obtained the individuals' separate consent before processing sensitive personal information, and have also met the requirement of "express consent," ensuring that individuals make authorization decisions voluntarily based on full knowledge.

[0066] Specific implementation methods include, but are not limited to, the following: setting up clear and prominent signs at personal information collection devices such as cameras and sensors to inform relevant personnel that they have entered the scope of personal information collection and that their personal information will be collected and processed. If an individual voluntarily enters the collection scope after being informed, it is deemed that they have agreed to the collection of their personal information; or using obvious icons, text descriptions, or other means on the terminal device or system interface for personal information processing to inform them of the rules for personal information processing, and obtaining the individual's explicit authorization through interactive methods such as pop-up prompts, check confirmation boxes, or asking the individual to upload their personal information themselves.

[0067] The aforementioned personal information processing rules should include, but are not limited to, the name and contact information of the personal information processor, the specific purpose of personal information processing, the processing method, the types of personal information processed, the retention period, and the methods and procedures for individuals to exercise their relevant rights.

[0068] Although the present invention has been described in detail above with general descriptions and specific embodiments, modifications or improvements can be made to it, which will be obvious to those skilled in the art. Therefore, all such modifications or improvements made without departing from the spirit of the present invention fall within the scope of protection claimed by the present invention.

Claims

1. A method for assessing the load-bearing capacity of a safety handrail manufacturing process based on the Industrial Internet of Things, characterized in that, include: By collecting acoustic emission sensing units deployed on the safety handrail manufacturing production line during knurling, and simultaneously reading strain distribution data at various measuring points on the pipe wall by the strain acquisition module, texture depth values ​​are extracted from the acoustic signals, and the effective load-bearing cross-sectional area reduction rate of the pipe wall is identified from the strain distribution data. The groove spacing between adjacent knurling grooves is read based on the texture depth value. The groove spacing is divided by the texture depth value to obtain the groove spacing depth ratio. Knurling sections with a groove spacing depth ratio lower than the safety threshold are identified as stress superposition risk sections. The energy peak value of the acoustic emission signal in the stress superposition risk section and the maximum strain gradient in the pipe wall strain distribution data are extracted and weighted to obtain the stress superposition intensity of the trench root. The coordinates of the first yield point are extracted from the strain distribution data. The support vector machine classification algorithm is used to identify the degree of change of the yield initiation position from random distribution to linear arrangement along the knurling groove, and the yield position concentration is obtained. Based on the yield position concentration and the superposition strength of the groove root stress, the damage level corresponding to the texture depth value is divided, and the knurled section with the damage level exceeding the preset warning level is marked as the initial position of the weak section. Extract the reduction rate of the load-bearing cross-sectional area at the initial location of the weak section and the ratio of the slot spacing depth. Analyze the case where the reduction rate of the load-bearing cross-sectional area exceeds the critical load-bearing threshold and the ratio of the slot spacing depth is lower than the safety threshold, and confirm that the initial location of the weak section is the target weak section location. Extract the residual wall thickness of the pipe body at the target weak section location and the superimposed stress intensity of the trench root, aggregate and optimize the residual wall thickness and the superimposed stress intensity of the trench root, construct a bearing capacity assessment model, and output the remaining bearing capacity level of the pipe body.

2. The load-bearing capacity assessment method for the manufacturing process of safety handrails based on the Industrial Internet of Things, as described in claim 1, is characterized in that... The process involves collecting acoustic emission signals during knurling using an acoustic emission sensing unit deployed on the safety handrail manufacturing line, simultaneously reading strain distribution data at various measuring points on the pipe wall using a strain acquisition module, extracting texture depth values ​​from the acoustic signals, and identifying the reduction rate of the effective load-bearing cross-sectional area of ​​the pipe wall from the strain distribution data. This includes: By deploying acoustic emission sensing units on the safety handrail manufacturing production line, acoustic signals are continuously collected during the processing stage when the knurling cutter cuts into the pipe wall. The acoustic signals are divided into frequency bands according to the frequency range. The pulse vibration amplitude generated by the periodic contact between the knurling cutter teeth and the pipe wall is identified from the high frequency band. The texture depth value of the knurling groove is calculated based on the peak-valley spacing of the pulse vibration amplitude. The interval of the measurement points of the strain acquisition module is determined according to the texture depth value. Multiple strain gauges are set along the circumferential and axial directions of the pipe wall to collect the strain values ​​generated by each measurement point during the knurling process. The difference in strain values ​​between adjacent measurement points is calculated, and the area where the difference in strain values ​​exceeds the preset difference threshold is identified as the concentrated area of ​​wall thickness reduction. Read the current wall thickness measurement value corresponding to each measuring point in the wall thickness reduction concentration area, subtract the current wall thickness measurement value from the original wall thickness of the pipe to obtain the wall thickness reduction amount, divide the wall thickness reduction amount by the original wall thickness to obtain the wall thickness reduction ratio, and multiply the wall thickness reduction ratio by the original cross-sectional area of ​​the pipe to obtain the effective bearing cross-sectional area reduction rate of the pipe wall.

3. The load-bearing capacity assessment method for the manufacturing process of safety handrails based on the Industrial Internet of Things, as described in claim 1, is characterized in that... The step of reading the groove spacing between adjacent knurling grooves based on the texture depth value, dividing the groove spacing by the texture depth value to obtain the groove spacing-depth ratio, and identifying knurling sections with a groove spacing-depth ratio lower than a safety threshold as stress superposition risk sections includes: Based on the knurling groove position corresponding to the texture depth value, the groove spacing between two adjacent knurling grooves is extracted from the pulse period of the acoustic signal. The groove spacing is divided by the texture depth value to obtain the groove spacing depth ratio. The groove spacing depth ratio value corresponding to each knurling section is recorded segment by segment along the tube axis. For each knurled section, if the groove pitch depth ratio is lower than a preset safety threshold, the knurled section is marked as a stress superposition risk section.

4. The load-bearing capacity assessment method for the manufacturing process of safety handrails based on the Industrial Internet of Things as described in claim 1, characterized in that, The step of extracting the peak energy of the acoustic emission signal within the stress superposition risk zone and the maximum strain gradient in the pipe wall strain distribution data, and then weighting and fusing them to obtain the stress superposition intensity at the trench root, includes: For the stress superposition risk section, the energy peak value is extracted from the acoustic emission signal within the corresponding time window of the section. The energy peak value is obtained by squaring the amplitude of the acoustic signal and then summing them. The energy peak value corresponding to the maximum acoustic energy in the section is recorded. From the strain distribution data of the pipe wall in the stress superposition risk zone, the strain value difference between adjacent measuring points is read, the strain value difference is divided by the distance between measuring points to obtain the strain gradient, and the maximum value of the strain gradient in the relevant zone is identified as the maximum strain gradient. The peak energy and the maximum strain gradient are divided by their respective upper limits to obtain normalized values. The normalized values ​​are then weighted and summed according to preset energy weighting coefficients and strain weighting coefficients to obtain the stress superposition intensity at the root of the stress superposition risk zone.

5. The load-bearing capacity assessment method for the manufacturing process of safety handrails based on the Industrial Internet of Things as described in claim 1, characterized in that, The method involves extracting the coordinates of each measuring point's initial yield point from the strain distribution data, and using a support vector machine classification algorithm to identify the degree of change in the yield initiation position from a random distribution to a linear arrangement along the knurling groove, thereby obtaining the yield position concentration, including: The strain values ​​of each measuring point are read from the strain distribution data point by point. The strain values ​​are compared with the yield strain threshold of the material corresponding to the measuring point. If the strain value of the measuring point reaches or exceeds the yield strain threshold for the first time, the position coordinates of the measuring point on the pipe wall are recorded to obtain the yield initiation position coordinate sequence. For the yield initiation position coordinate sequence, the lateral offset distance of each coordinate point relative to the center line of the knurled groove bottom is read along the pipe body axis, and the lateral offset distance is combined to form the distribution feature vector of each yield initiation position; The spatial dispersion between each yield initiation position is calculated based on the distribution feature vector. The spatial dispersion is obtained by statistically analyzing the variance of the distance from each coordinate point to the center line of the knurling groove bottom. The spatial dispersion is used as a quantitative indicator to characterize the distribution pattern of the yield initiation position. A support vector machine classification algorithm is adopted, using the distribution feature vector and the spatial dispersion as input features, and pre-labeled random distribution samples and linearly arranged samples as training data to obtain the classification boundary that distinguishes between random distribution and linear arrangement. Based on the classification boundary, the confidence score of the current yield initiation position sequence belonging to the linear arrangement is output, and the confidence score is used as the yield position concentration.

6. The load-bearing capacity assessment method for the manufacturing process of safety handrails based on the Industrial Internet of Things as described in claim 1, characterized in that, The step of classifying the damage level corresponding to the texture depth value based on the yield position concentration and the superimposed stress intensity of the groove root, and marking the knurled section with the damage level exceeding the preset warning level as the initial location of the weak section, includes: Based on the yield position concentration and the stress superposition intensity of the groove root, a two-dimensional damage assessment coordinate system is established, with the yield position concentration as the horizontal axis value and the stress superposition intensity of the groove root as the vertical axis value. The location points of each knurled section are marked in the two-dimensional damage assessment coordinate system. Multiple damage level regions are pre-defined in the two-dimensional damage assessment coordinates. The damage level of each knurled section is determined based on the damage level region into which its position point falls. For each knurled section, if the damage level exceeds the preset warning level, the knurled section is marked as a preliminary weak section location.

7. The load-bearing capacity assessment method for the manufacturing process of safety handrails based on the Industrial Internet of Things as described in claim 1, characterized in that, The step of extracting the ratio of the reduction rate of the load-bearing cross-sectional area at the initial location of the weak section to the ratio of the slot spacing depth, analyzing the case where the reduction rate of the load-bearing cross-sectional area exceeds the critical load-bearing threshold and the ratio of the slot spacing depth is lower than the safety threshold, and confirming the initial location of the weak section as the target weak section location includes: For the initial location of the weak section, the reduction rate of the load-bearing cross-sectional area and the ratio of the slot spacing depth corresponding to the section are extracted. It is determined that the reduction rate of the load-bearing cross-sectional area exceeds the preset critical load threshold, and at the same time, it is determined that the ratio of the slot spacing depth is lower than the preset safety threshold. If the reduction rate of the load-bearing cross-sectional area exceeds the critical load-bearing threshold and the groove spacing depth ratio is lower than the safety threshold, then the preliminary location of the weak section is confirmed as the target weak section location.

8. The load-bearing capacity assessment method for the manufacturing process of safety handrails based on the Industrial Internet of Things, as described in claim 1, is characterized in that... The process involves extracting the residual wall thickness of the pipe at the target weak section location and the superimposed stress intensity at the trench root, then performing aggregate optimization on the residual wall thickness and the superimposed stress intensity at the trench root to construct a bearing capacity assessment model and output the remaining bearing capacity level of the pipe, including: For the target weak section, the residual wall thickness of the section is extracted from the pipe wall strain distribution data. The residual wall thickness is the remaining thickness value after subtracting the wall thickness reduction from the original wall thickness of the pipe body. At the same time, the stress superposition intensity of the groove root corresponding to the section is read. The residual wall thickness and the stress superposition strength of the trench root are subjected to aggregate optimization processing. The residual wall thickness is divided by the original wall thickness of the pipe to obtain the wall thickness retention ratio. The wall thickness retention ratio is multiplied by the normalized compensation value of the stress superposition strength of the trench root to obtain the comprehensive bearing index. The normalized compensation value is a normalized value minus the stress superposition strength of the trench root. Read the texture depth value, slot spacing depth ratio, and yield position concentration corresponding to the location of the target weak section, and combine the texture depth value, slot spacing depth ratio, yield position concentration, and comprehensive bearing capacity index to form a bearing capacity assessment input vector; The random forest regression algorithm is used, with the bearing capacity assessment input vector as the input feature and the pre-labeled pipe bearing capacity test data as the training label for training. It outputs continuous bearing capacity prediction values ​​and maps the bearing capacity prediction values ​​to the remaining bearing capacity level of the pipe according to the preset bearing capacity level division interval.

Images