A copper foil thickness analysis method and system

Abstract

The application provides a copper foil thickness analysis method and device, comprising: obtaining tension values of each layer position in the copper foil winding process and thickness measurement results of the corresponding layer position; according to the compression deformation analysis result, combining the obtained interlayer relative displacement and torsion angle to analyze the deformation mechanism, and obtaining stress distribution mode data; compensating the thickness measurement data based on the stress distribution mode data to obtain compensated thickness measurement data; setting a normal thickness range and an abnormality judgment threshold based on the compensated thickness measurement data, determining a sensitivity identification threshold parameter of thickness abnormality detection; judging the thickness abnormality state according to the sensitivity identification threshold parameter of thickness abnormality detection to obtain a thickness abnormality detection result; and determining a thickness abnormality detection combination parameter in combination with the thickness abnormality detection result and historical winding process thickness measurement data.

Description

Technical Field

[0001] This invention relates to the field of information technology, and in particular to a method and system for analyzing the thickness of copper foil. Background Technology

[0002] Copper foil, as a core material in the electronics industry, directly determines the performance of circuit boards and the quality of electronic products, holding an irreplaceable and crucial position in high-precision electronic manufacturing. With the trend towards thinner and more integrated electronic products, the requirements for copper foil thickness control accuracy are becoming increasingly stringent, making accurate thickness measurement a vital link in ensuring product quality. Traditional copper foil thickness measurement methods mainly rely on static detection or simple online measurements. These methods often ignore the influence of dynamic mechanical factors during the winding process, leading to significant deviations between the measurement results and the actual thickness. Existing measurement systems lack in-depth analysis of the multi-layer superposition compression mechanism of copper foil under winding conditions, the law of interlayer shear deformation transmission, and the coupling relationship between tension, deformation, and thickness. This makes it impossible to effectively compensate for measurement errors caused by winding tension, seriously affecting the accuracy and reliability of quality control. Specifically, during the copper foil winding process, due to factors such as changes in winding speed, response delays in the tension control system, and differences in copper foil material properties, the unevenness of tension distribution becomes prominent at different layers. The inner copper foil layers bear the cumulative pressure from the outer layers, undergoing significant compression deformation and reducing their apparent thickness. This compression intensifies with the increase in the number of winding layers. The presence of compressive deformation further induces a cumulative shear deformation effect in the intermediate copper foil. This shear deformation not only alters the geometry of the copper foil but also accumulates continuously during the winding process. Therefore, when the winding tension distribution changes dynamically, the deformation response of copper foils with different layers exhibits highly nonlinear characteristics, such as coupling between shear deformation and radial position. Traditional strategies for fixed compensation of copper foil thickness measurements with different layers cannot adapt to this dynamic change, leading to a sharp decline in measurement accuracy. How to combine the relationship between winding tension and copper foil deformation, and develop a dynamic tension compensation measurement strategy based on the deformation mechanism under different winding layers, has become a key issue in the development of precision copper foil thickness measurement technology. Summary of the Invention

[0003] This invention provides a method for analyzing the thickness of copper foil, mainly including: Obtain the tension value and thickness measurement results of each layer position during the copper foil winding process; By analyzing the correlation between the thickness variation and tension distribution of copper foil under different winding layers, the compression deformation law of copper foil under different winding layers is identified, and the compression deformation analysis results are generated based on the compression deformation law of copper foil under different winding layers. Based on the compression deformation analysis results, the deformation mechanism is analyzed in conjunction with the obtained interlayer relative displacement and torsional angle to obtain stress distribution pattern data; Based on the stress distribution pattern data, the thickness measurement data is compensated to obtain the compensated thickness measurement data. Based on the compensated thickness measurement data, the normal thickness range and anomaly judgment threshold are set to determine the sensitivity identification threshold parameters for thickness anomaly detection. The thickness anomaly state is judged based on the sensitivity identification threshold parameter of the thickness anomaly detection, and the thickness anomaly detection result is obtained. Based on the thickness anomaly detection results and historical winding process thickness measurement data, the combined parameters for thickness anomaly detection are determined.

[0004] Furthermore, the acquisition of tension values ​​and corresponding thickness measurement results at each layer position during the copper foil winding process includes: Tension sensors are placed at various layers of the copper foil winding device to collect tension signals during the winding process. The tension signals are amplified and filtered to generate a real-time tension value data sequence. The layer number identifier and acquisition time information in the real-time tension value data sequence are marked. A thickness measuring device is deployed with multiple measuring points distributed along the width direction of the copper foil to collect thickness data at each measuring point. The average thickness data of all measuring points is calculated to generate the average thickness value of the copper foil layer. The layer number identifier and measurement time corresponding to the average thickness value are marked. The real-time tension value data sequence is matched with the average thickness value. The correspondence is determined by the layer number identifier and the time difference to generate a corresponding data group containing the layer number, tension value, and thickness value. The comparison result of the tension value of the corresponding data group with the preset tension threshold is judged. The corresponding data group that exceeds the threshold and its adjacent layer number data groups are marked. All corresponding data groups are stored.

[0005] Furthermore, the compression deformation law of copper foil with different winding layers is identified by analyzing the correlation between the thickness variation and tension distribution under different winding layers. Based on the compression deformation law of copper foil with different winding layers, compression deformation analysis results are generated, including: Based on the real-time tension values ​​at different radial positions and the original thickness measurement results at the corresponding layer positions, a three-dimensional data matrix of layer number, tension, and thickness is constructed. The thickness variation value of each layer is calculated, and a quantitative correlation index between tension value and thickness variation value is generated. Based on the quantitative correlation index, combined with the winding radius and copper foil width, the radial pressure generated by each layer is calculated, and the cumulative pressure value borne by the inner copper foil is obtained by summing them. Based on the cumulative pressure value and the elastic modulus of the copper foil material obtained in advance, a compression deformation analysis result is generated.

[0006] Furthermore, based on the compression deformation analysis results, and combined with the obtained interlaminar relative displacement and torsional angle, the deformation mechanism is analyzed to obtain stress distribution pattern data, including: Extract relative displacement and torsional angle data between adjacent layers, measure the vertical distance between the surfaces of adjacent copper foil layers, calculate the tangential shear strain, and calculate the torsional shear strain by combining the torsional angle and the radius of the current layer. Add the tangential shear strain and the torsional shear strain to generate a total shear strain value. Accumulate the total shear strain value layer by layer to generate the cumulative shear deformation of each layer. Extract the compression deformation from the compression deformation analysis results, fit the relationship between the cumulative shear deformation and the compression deformation, determine the regression coefficient, substitute it into a preset coupling function form to generate a coupling relationship expression, substitute the compression deformation and the cumulative shear deformation to calculate the equivalent stress value under the combined deformation state, and generate stress distribution pattern data.

[0007] Furthermore, the compensation of the thickness measurement data based on the stress distribution pattern data to obtain the compensated thickness measurement data includes: Extract stress values ​​and real-time tension values, combine them with deformation data from the compression deformation analysis results, calculate thickness deviation, and generate a dynamic compensation parameter library; based on real-time tension values ​​and layer positions, search for matching categories in the dynamic compensation parameter library, extract the average tension, average deformation, and average thickness deviation of that category, and calculate the dynamic compensation coefficient; use the dynamic compensation coefficient to correct the original thickness measurement value, and obtain the compensated thickness measurement data.

[0008] Furthermore, the calculation of the dynamic compensation coefficient includes: finding a category that matches the real-time tension value, extracting the mean value of the parameters of the category, calculating the distance ratio between the real-time tension value and the mean value of the category, and generating the dynamic compensation coefficient.

[0009] Furthermore, the determination of thickness anomaly detection combination parameters by combining the thickness anomaly detection results and historical winding process thickness measurement data includes: Combining the anomaly detection results and historical winding process thickness measurement data, the frequency of anomalies at each layer position is calculated by dividing the number of anomalies at each position by the total number of measurements for that layer. Layers with anomaly frequencies greater than twice the historical average anomaly frequency are identified as key adjustment layers, and deviation characteristic values ​​are extracted from these key adjustment layers. The current compensation coefficient for the key adjustment layer is then searched in the dynamic compensation parameter library. The current compensation coefficient is multiplied by the deviation characteristic value to obtain a new compensation coefficient, and the parameters for the corresponding layer in the dynamic compensation parameter library are updated. The updated dynamic compensation parameter library is used to perform compensation calculations on the original thickness data of the most recent production batch. The number of anomalies detected in the compensated thickness data is counted, and the thickness anomaly detection combination parameters are determined.

[0010] This invention provides a copper foil thickness analysis system, which employs the method described above and mainly includes: The data acquisition module is used to acquire the tension value and thickness measurement results of each layer position during the copper foil winding process. The correlation analysis module is used to identify the compression deformation law of copper foil with different winding layers by analyzing the correlation between the thickness change and tension distribution of copper foil with different winding layers, and to generate compression deformation analysis results based on the compression deformation law of copper foil with different winding layers. The compression deformation analysis module is used to analyze the deformation mechanism based on the compression deformation analysis results, combined with the obtained interlayer relative displacement and torsional angle, and obtain stress distribution pattern data. The thickness compensation module is used to compensate the thickness measurement data based on the stress distribution pattern data to obtain the compensated thickness measurement data. The data compensation module is used to set the normal thickness range and anomaly judgment threshold based on the compensated thickness measurement data, and to determine the sensitivity identification threshold parameter for thickness anomaly detection. The threshold setting module is used to determine the thickness anomaly state based on the sensitivity identification threshold parameter of the thickness anomaly detection, and obtain the thickness anomaly detection result. An anomaly detection module is used to determine the thickness anomaly detection combination parameters by combining the thickness anomaly detection results with the thickness measurement data of the historical winding process.

[0011] A computer-readable storage medium storing a computer program that, when executed by a processor, implements the method described above.

[0012] A computer program product includes a computer program that, when executed by a processor, implements the steps of the above-described method.

[0013] The technical solutions provided by the embodiments of the present invention may include the following beneficial effects: This invention discloses a method and system for copper foil thickness analysis. Addressing the complex correlation between thickness variation and tension distribution during copper foil winding, and the impact of interlayer compression and shear deformation coupling on measurement accuracy, the method acquires tension sensor and thickness measurement data in real time. It analyzes the correlation between thickness variation and tension distribution under different layer numbers, constructs a nonlinear relationship model between compression deformation and layer position, calculates the cumulative pressure of the inner copper foil, and then establishes a coupled model of compression and shear deformation using interlayer relative displacement and torsional angle to predict stress distribution under combined deformation. When the cumulative shear deformation exceeds a threshold, finite element analysis is used to obtain the stress distribution pattern, and a dynamic compensation parameter library based on tension sensing is constructed to dynamically adjust the thickness measurement compensation coefficient. Using the compensated thickness data as input, a benchmark thickness range model is established, and deviations are monitored and model parameters are updated in real time. The compensation parameter library and threshold are optimized through abnormal frequency and positional deviation distribution, achieving high-precision dynamic measurement and anomaly detection. This invention significantly improves the accuracy of copper foil thickness measurement and the robustness of quality monitoring, providing reliable technical support for high-performance copper foil production. Attached Figure Description

[0014] Figure 1 This is a flowchart of a copper foil thickness analysis method according to the present invention.

[0015] Figure 2 This is a schematic diagram of the structure of a copper foil thickness analysis system according to the present invention. Detailed Implementation

[0016] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be described in detail below with reference to the accompanying drawings and specific embodiments.

[0017] like Figure 1-2 This embodiment of a copper foil thickness analysis method and apparatus may specifically include: S101. Obtain the tension value and thickness measurement results of each layer position during the copper foil winding process.

[0018] Tension sensors are placed at each layer position in the copper foil winding device. The sensors collect tension signals during the winding process at a preset sampling frequency. The collected tension signals are amplified and filtered by a signal conditioning circuit to obtain a real-time tension value data sequence for each layer position. Each tension value data includes a layer number identifier and the acquisition time information. A laser thickness gauge or ultrasonic thickness gauge is used to measure the thickness of the wound copper foil. The thickness gauge has multiple measurement points evenly distributed along the width direction of the copper foil, acquiring the thickness data at each measurement point. The sum of the thickness data at all measurement points is calculated using an arithmetic mean method, divided by the number of measurement points, to obtain the average thickness value of that layer of copper foil. The layer number identifier and measurement time corresponding to this thickness value are recorded. Based on the layer number identifier and acquisition time information in the real-time tension value data sequence, it is matched with the layer number identifier and measurement time of the average thickness value. When the layer number identifiers are the same and the time difference is within a preset range, the tension value and thickness value are determined to correspond, forming a corresponding data set containing the layer number, tension value, and thickness value. For each corresponding data group, it is determined whether its tension value exceeds the preset tension threshold. If it exceeds the threshold, the corresponding data group and the corresponding data groups of the adjacent layers are marked as key data and stored. If it does not exceed the threshold, the corresponding data group is stored in the standard format to obtain a complete record of the tension sensor data and thickness measurement data of each layer position during the copper foil winding process.

[0019] Specifically, tension control and thickness monitoring are key factors in ensuring product quality during copper foil winding. The placement of tension sensors needs to take into account the structural characteristics of the winding device; typically, multiple strain gauge or piezoelectric sensors are installed at different radii of the winding shaft. These sensors can detect changes in tensile stress of the copper foil in real time during the winding process.

[0020] Specifically, the signal conditioning circuit plays a crucial role in tension signal processing. The raw tension signal often contains high-frequency noise and electromagnetic interference. Operational amplifiers amplify the signal, bringing the weak sensor output signal to the voltage level required for subsequent processing. Filtering then uses a low-pass filter to remove high-frequency interference, retaining the low-frequency signal components that reflect actual tension changes. The conditioned signal forms continuous time-series data, with each data point containing a precise time stamp and corresponding layer number information.

[0021] In one possible implementation, a laser thickness gauge emits a laser beam onto the surface of a copper foil and calculates the thickness based on the time difference of the reflected light. Multiple measurement points arranged along the width of the copper foil can capture variations in its lateral thickness. The arithmetic averaging method is used to reduce random errors from single-point measurements; by summing the thickness values ​​of all measurement points and dividing by the total number of points, a more stable and reliable average thickness value is obtained. This multi-point measurement and averaging approach effectively reduces measurement deviations caused by localized unevenness on the copper foil surface.

[0022] It is important to note that time matching between tension and thickness data is crucial for achieving accurate correspondence. Due to potential physical differences in the location of tension sensors and thickness measuring devices, the time taken for the same layer of copper foil to pass through the two sensors may differ. By setting a reasonable time window, when the layer number identifiers of the two data points are the same and the time difference is within an acceptable range, they can be considered measurement results for the same layer of copper foil. This matching mechanism ensures accurate data correspondence, providing a reliable foundation for subsequent quality analysis.

[0023] Preferably, the tension threshold is set based on the mechanical properties of the copper foil material and the requirements of the manufacturing process. When the tension value of a certain layer exceeds the threshold, it indicates that there may be an abnormality at that location, such as material defects or equipment failure. In this case, the data of that layer and adjacent layers are marked and stored in a key manner, which helps with subsequent quality traceability and problem analysis. Data within the normal range is stored in a standard format, which ensures data integrity and avoids wasting storage resources.

[0024] S102. By analyzing the correlation between the thickness variation and tension distribution of copper foil under different winding layers, the compression deformation law of copper foil under different layers is identified, and the compression deformation analysis results are generated based on the compression deformation law of copper foil under different layers. Based on the real-time tension values ​​at different radial positions and the original thickness measurements at corresponding layer positions, a three-dimensional data matrix of layer number, tension, and thickness is constructed. The difference between the original thickness and the standard thickness of each layer is calculated to obtain the thickness variation value. The correlation coefficient between the tension value and the thickness variation value of each layer is calculated using the Pearson correlation coefficient. When the absolute value of the correlation coefficient is greater than a preset threshold, it is determined that the layer has a significant tension-thickness correlation characteristic, and the correlation quantification index of each layer is obtained. Based on the correlation quantification index, for layers with significant correlation, the winding radius and copper foil width parameters of the layer are obtained. A radial pressure calculation method is used, where the radial pressure is equal to the tension value of the layer divided by the winding radius and then multiplied by the copper foil width. The radial pressure generated by each layer is accumulated from the outermost layer inward to obtain the cumulative pressure value borne by each inner layer position. The theoretical compression of each layer is calculated using the cumulative pressure value and the elastic modulus of the copper foil material. The compressed thickness is obtained by subtracting the theoretical compression from the original thickness measurement. The difference between the original thickness and the compressed thickness is the compression deformation. The least squares method is used to fit a power function relationship between the compression deformation and the layer position, determining the power exponent and proportionality coefficient to obtain a nonlinear expression describing the compression deformation law of the copper foil. The power function expression for the compression deformation and the layer position is as follows: The formula denoted by y represents the compression deformation, x represents the layer position, a represents the proportionality coefficient, and b represents the power exponent. This formula describes the nonlinear power function law governing the compression deformation of copper foil as a function of the layer position. Based on this nonlinear relationship expression and the value of any layer position, the theoretical compression deformation of that layer is calculated by substituting it into the expression. The original thickness measurement of the layer is then subtracted from the theoretical compression deformation to obtain the predicted actual thickness considering the compression effect. Finally, the compression deformation of each layer, the cumulative pressure value, and the predicted actual thickness are summarized to form the compression deformation analysis results.

[0025] Specifically, the analysis of compression deformation during copper foil winding involves complex relationships between multiple physical quantities. The Pearson correlation coefficient, as a statistical indicator measuring the degree of linear correlation between two variables, ranges from -1 to 1. When the calculated correlation coefficient is close to 1 or -1, it indicates a strong correlation between tension and thickness change; this strong correlation means that changes in tension directly lead to changes in copper foil thickness.

[0026] Specifically, the construction of a three-dimensional data matrix is ​​fundamental to achieving accurate analysis. Each element of this matrix contains tension and thickness measurements at specific layer locations, allowing for batch processing of large amounts of data through matrix operations. Calculating thickness variation requires a pre-defined standard thickness as a reference, typically derived from copper foil production specifications or quality standards. When the measured thickness is less than the standard thickness, the thickness variation is negative, indicating that the copper foil is compressed; conversely, it indicates potential tension or other anomalies.

[0027] In one possible implementation, the radial pressure is calculated based on the circumferential stress theory in materials mechanics. During the winding process, the tension of the outer copper foil is converted into radial pressure on the inner layers. This pressure transmission mechanism is similar to the stress distribution in a wound pressure vessel, where each layer of copper foil bears pressure from the outer layers and exerts pressure on the inner layers. The winding radius gradually increases with the number of layers, and this change in radius directly affects the magnitude and distribution of the pressure.

[0028] It should be noted that the cumulative pressure value is calculated using a layer-by-layer superposition method. Starting from the outermost layer, the radial pressure generated by each layer is transmitted to the inner layers, and the total pressure borne by the inner layers is the sum of the pressures from all the outer layers. This cumulative effect results in the innermost layer experiencing the greatest pressure and being most prone to significant compressive deformation. The elastic modulus of the copper foil material is a key parameter connecting pressure and deformation, reflecting the material's ability to resist elastic deformation.

[0029] Preferably, a power function is chosen to describe the relationship between compression deformation and the number of layers during the least squares fitting process. The power function effectively reflects nonlinear characteristics, particularly the pattern of rapid deformation growth in the early stages of winding followed by stabilization. The power exponent parameter determines the curvature of the curve, while the proportionality coefficient controls the overall deformation. By fitting measured data points, the optimal parameter combination can be obtained, minimizing the deviation between the fitted curve and the actual data. The advantage of this analytical method is its ability to predict the compression deformation at any layer position, even if direct measurement data is unavailable at that location. By substituting the layer values ​​into the fitted expression, the theoretical deformation can be calculated, and thus the actual thickness after compression can be estimated. This prediction helps identify risk areas where over-compression may occur.

[0030] S103. Based on the compression deformation analysis results, and combined with the obtained interlayer relative displacement and torsional angle, the deformation mechanism is analyzed to obtain stress distribution pattern data.

[0031] Based on the compression deformation analysis results, relative displacement and torsional angle data between adjacent layers are obtained. The vertical distance between the surfaces of adjacent copper foil layers is measured as the interlayer distance. Tangential shear strain is calculated based on the relationship that shear strain equals the relative displacement between layers divided by the interlayer distance. Torsional shear strain is calculated by multiplying the torsional angle by the radius of the current layer and then dividing by the interlayer distance. The tangential and torsional shear strains are added together to obtain the total shear strain value. The total shear strain value of each layer is accumulated from the inner layer to the outer layer to obtain the cumulative shear deformation of each layer. Based on the cumulative shear deformation and the compression deformation in the compression deformation analysis results, a multiple regression method is used to fit the relationship between shear deformation and compression deformation. The regression coefficient is determined as the influence coefficient of compression deformation on shear deformation. The compression deformation, shear deformation, and influence coefficient are substituted into a preset coupling function form to obtain the coupling relationship expression describing the interaction between the two deformations and its parameter values. Based on the coupling relationship expression and its parameter values, the compression deformation and shear deformation of each layer are substituted into the expression to calculate the equivalent stress value under the composite deformation state. If the cumulative shear deformation of a certain layer exceeds the preset shear deformation threshold, a geometric model for finite element analysis is established for that layer, triangular or quadrilateral elements are divided, displacement constraints and stress loads are applied to the model boundary, and the stress distribution pattern data of each element in the region is obtained by solving the nodal displacement and element stress.

[0032] Specifically, shear deformation is a complex mechanical phenomenon during copper foil winding, primarily stemming from the relative motion between adjacent layers. When the outer copper foil undergoes circumferential strain under tension, it induces tangential displacement in the inner layers; this displacement difference creates the shear effect. The relative displacement between layers can be measured by setting markers on the copper foil surface and using image recognition technology to track the positional changes of these markers during the winding process.

[0033] Specifically, the calculation of tangential shear strain is based on the shear deformation theory in mechanics of materials. When two adjacent layers slip relative to each other, the shear strain is equal to the ratio of the relative displacement to the interlayer distance. This ratio reflects the degree of shear deformation of the material. Torsional shear strain takes into account the torsional effect during the winding process. When the copper foil is wound at a certain helical angle, it generates additional shear force. The product of the torsional angle and the radius represents the circumferential displacement caused by torsion, and dividing by the interlayer distance yields the shear strain component caused by torsion. In one possible implementation, the application of multiple regression requires the collection of a large amount of experimental data. By changing process parameters such as winding tension and speed, the compression and shear deformation under different conditions are measured to establish a data sample set. In the regression analysis, compression deformation is used as the independent variable and shear deformation as the dependent variable. The optimal regression coefficient is determined by minimizing the sum of squared errors between the predicted and measured values.

[0034] S represents the sum of squared errors, n represents the sample size, and Y represents the sum of squared errors. i Y represents the measured value of the i-th shear deformation. i The cap represents the predicted value of the i-th shear deformation, X. i Let βi represent the i-th compression deformation, β0 represent the regression intercept coefficient, and β1 represent the regression slope coefficient. This formula determines the optimal regression coefficients by minimizing the sum of squared errors between the predicted and measured values. These coefficients quantify the influence of compression deformation on shear deformation.

[0035] It should be noted that the construction of the coupling relationship expression considers the interaction mechanism between the two deformations. Compressive deformation alters the interlayer contact state and friction conditions, thus affecting the development of shear deformation. The coupling function is typically in polynomial form, containing linear and cross terms, and can describe the nonlinear relationship between deformations. By substituting actual deformation data into the expression, the shear deformation response under a specific compressive state can be predicted.

[0036] Preferably, mesh generation in finite element analysis is the process of discretizing a continuous copper foil region. Triangular elements are suitable for irregular boundary regions, while quadrilateral elements offer higher computational accuracy in regular regions. The choice of element size requires a balance between computational accuracy and efficiency; denser meshes are used in stress concentration regions, while sparser meshes are used in regions with gradual stress changes. Boundary conditions simulate actual constraint states, such as fixed-end constraints limiting node displacement, while stress loads simulate external forces. The calculation of equivalent stress comprehensively considers the combined effects of normal stress and shear stress, transforming multi-directional stress states into single values ​​to facilitate the assessment of the material's stress state. When the cumulative shear deformation exceeds a threshold, it indicates potential material degradation or damage in that region, requiring detailed stress analysis. The finite element solution process establishes the global stiffness matrix and load vector, solves linear equations to obtain nodal displacements, and then calculates element stresses, forming a stress distribution map of the entire analysis region.

[0037] S104. Based on the stress distribution pattern data, the thickness measurement data is compensated to obtain the compensated thickness measurement data.

[0038] Using the stress distribution pattern data, stress values ​​and corresponding real-time tension values ​​at each layer are extracted. Deformation data for each layer is obtained from the compression deformation analysis results. The difference between the actual measured thickness and the design thickness is calculated as the thickness deviation. Data points with similar stress, tension, and thickness deviation characteristics are grouped into the same category using the K-means clustering method. The mean tension, mean deformation, and mean thickness deviation of all data points within each category are calculated. A correspondence table between category identifiers and the three mean parameters is established, forming a dynamic compensation parameter library. Based on the real-time tension value and layer position at the current measurement location, the category closest to the current tension value is searched in the dynamic compensation parameter library. The mean tension, mean deformation, and mean thickness deviation of this category are extracted. When the real-time tension value is between the mean tension values ​​of two adjacent categories, the mean thickness deviation of the two categories is weighted and averaged according to the distance ratio between the real-time tension value and the mean tension values ​​of the two categories to obtain the dynamic compensation coefficient for the current location. The original thickness measurement value is corrected using the dynamic compensation coefficient. The original measurement value is subtracted from the dynamic compensation coefficient to obtain the compensated thickness value. The difference between the tension value at the current time and the previous sampling time is calculated. If the absolute value of the difference exceeds the preset tension change threshold, the parameter library is returned to search for the corresponding category again. The updated dynamic compensation coefficient is calculated based on the new tension value. The thickness measurement value is corrected again using the updated compensation coefficient to obtain the dynamically compensated thickness measurement data.

[0039] Specifically, the construction of the dynamic compensation parameter library is based on statistical analysis of a large amount of historical measurement data. During the copper foil winding process, different tension conditions will cause the material to undergo different degrees of elastic deformation, which directly affects the accuracy of thickness measurement. The design thickness refers to the standard specification thickness during copper foil manufacturing, while the actual measured thickness will deviate due to stress.

[0040] Specifically, the K-means clustering method groups data points through iterative optimization. Initially, K centroids are randomly selected. Then, the Euclidean distance from each data point to each centroid is calculated, assigning the data point to the category represented by the nearest centroid. Subsequently, the centroid position of each category is recalculated, which is the average of all dimensions of all data points within that category. This process is repeated until the centroid positions no longer change significantly. In copper foil thickness compensation applications, each data point contains three dimensions: stress value, tension value, and thickness deviation value.

[0041] In one possible implementation, the choice of the number of categories K needs to balance accuracy and efficiency. Too few categories will lead to excessive differences in data within the same category, reducing compensation accuracy; too many categories will increase the complexity of storage and retrieval. The optimal value of K can usually be determined using the elbow rule, which involves plotting the sum of squares within groups for different K values ​​and selecting the location where the curve shows a clear inflection point.

[0042] It should be noted that the weighted average calculation method embodies the idea of ​​linear interpolation. When the real-time tension value lies between the tension averages of two adjacent categories, the compensation coefficient should not abruptly change, but rather transition smoothly. Assuming the tension averages of two adjacent categories are T1 and T2, and the corresponding thickness deviation averages are D1 and D2, with a real-time tension value of T, then the interpolation weights can be determined based on the relative distances of T to T1 and T2. The closer T is to T1, the greater the weight of D1; conversely, the farther away T is from T1, the greater the weight of D2.

[0043] Preferably, the setting of the tension change threshold needs to consider the response characteristics of the measurement system and the mechanical properties of the material. If the threshold is set too small, it will lead to frequent parameter updates, increasing the computational burden and potentially introducing noise; if the threshold is too large, it will not be able to respond promptly to significant changes in tension, affecting the compensation effect. The threshold is usually set to 2 to 3 times the normal tension fluctuation range. The compensation coefficient is applied by subtraction rather than addition because copper foil is usually compressed and thinned under tension, and the actual measured value is less than the true thickness. By subtracting a positive compensation coefficient, the true thickness when unaffected by stress can be restored. This compensation mechanism can eliminate the interference of tension changes on thickness measurement and improve the reliability of measurement data. The dynamic update mechanism ensures that the compensation parameters can adapt to changes in working conditions during the winding process. When a significant change in tension is detected, the parameter is repositioned in the parameter library to obtain a new compensation coefficient. This real-time adjustment capability enables the thickness measurement to maintain high accuracy throughout the winding process, and stable and reliable measurement results can be obtained even under large tension fluctuations.

[0044] S105. Based on the compensated thickness measurement data, set the normal thickness range and the anomaly judgment threshold, and determine the sensitivity identification threshold parameter for thickness anomaly detection.

[0045] Using compensated thickness measurement data as input, a thickness data sequence is formed by arranging the data in chronological order. The arithmetic mean of all data points in the sequence is calculated as the central reference value. The variance is obtained by calculating the sum of squares of the deviations of each data point from the central reference value and dividing it by the total number of data points. The standard deviation is obtained by taking the square root of the variance. The standard deviation, calculated by adding or subtracting a preset multiple from the central reference value, is determined as the upper and lower limits of the normal thickness range. A reference thickness range parameter set is constructed, including the central reference value, standard deviation, and upper and lower limits. For the thickness data sequence and the reference thickness range parameter set, a data window of fixed length and a window movement step size are set. The average value of the data points within each window is calculated as the local mean at each window position. The difference between the local means of adjacent windows is calculated and divided by the window movement step size to obtain the local change gradient. The local means and local change gradients at all window positions are statistically analyzed to form a statistical feature sequence describing the temporal variation characteristics of the thickness data. Based on the local change gradient data in the statistical feature sequence, all gradient absolute values ​​are sorted from smallest to largest, and the gradient value corresponding to the preset percentile is selected as the gradient anomaly detection threshold. When the local change gradient exceeds the detection threshold for more than a preset number of consecutive windows, it is determined that the thickness change trend is abnormal. Combining the upper and lower limits of the range in the reference thickness range parameter group and the gradient anomaly detection threshold, the sensitivity identification threshold parameter for thickness anomaly detection is determined.

[0046] Specifically, the establishment of the baseline thickness range model relies on the normal distribution theory in statistics. During copper foil production, thickness measurements typically fluctuate around a central value, and this fluctuation exhibits a normal distribution under normal production conditions. The central baseline value represents the expected thickness, while the standard deviation reflects the dispersion of the thickness fluctuation.

[0047] Specifically, the calculation of variance involves the deviation of each measurement point from the mean. Assuming there are 100 thickness measurements, first calculate the arithmetic mean of these 100 values, then calculate the difference between each value and the mean, square these differences, sum them, and divide by 100 to obtain the variance. The standard deviation is the square root of the variance; it has the same dimensions as the original data and more intuitively reflects the dispersion of the data. The preset multiplier is usually chosen to be 2.5 or 3, based on the confidence interval theory in statistics, which can cover most normal data points.

[0048] In one possible implementation, the sliding window method requires appropriately setting window parameters. The window length determines the amount of data for local statistics; too short a window can lead to unstable statistical results, while too long a window may mask local variation characteristics. The window movement step size determines the level of analysis granularity; a step size of 1 indicates point-by-point movement, capable of capturing subtle changes; a step size greater than 1 improves computational efficiency but may miss some points of change. The calculation of the local change gradient reflects the rate of change of thickness over time; rapid gradient changes often indicate anomalies in the production process.

[0049] It's important to note that the choice of percentile directly affects the sensitivity of anomaly detection. After sorting all the absolute values ​​of gradients, the 90th percentile means that 90% of the gradient values ​​are less than this threshold, and only 10% of the gradients are considered anomalies. Choosing a higher percentile reduces the false alarm rate but may miss real anomalies; choosing a lower percentile has the opposite effect. This trade-off needs to be determined based on the actual quality requirements of production.

[0050] Preferably, the anomaly detection mechanism using continuous windows can distinguish between occasional fluctuations and trend anomalies. A gradient exceeding the limit in a single window may be caused by measurement noise or transient disturbances, while anomalies in multiple consecutive windows indicate a systemic problem. The preset number is typically set to 3 to 5 consecutive windows; this design filters out occasional interference while promptly identifying genuine quality issues. Determining the sensitivity identification threshold parameters is a comprehensive process. It includes not only the upper and lower limits based on statistical distribution but also a gradient threshold based on time-series changes. This dual detection mechanism can identify anomalies from different dimensions: range detection focuses on whether the absolute value deviates from the normal range, while gradient detection focuses on whether the rate of change is abnormal. The two detection methods complement each other, improving the accuracy and reliability of anomaly identification. By adjusting the various threshold parameters, it is possible to adapt to the requirements of different product specifications and quality standards, achieving flexible quality control.

[0051] S106. Determine the thickness anomaly state based on the sensitivity identification threshold parameter for thickness anomaly detection, and obtain the thickness anomaly detection result.

[0052] Based on the sensitivity identification threshold parameter and the reference thickness range parameter set for anomaly detection, the compensated thickness data at the current moment is obtained. The difference between this data and the center reference value in the reference thickness range parameter set is calculated as the current deviation value. The sensitivity identification threshold is adjusted according to the current layer position to obtain a dynamic threshold. If the absolute value of the current deviation value exceeds the dynamic threshold, the current measurement point is determined to be in an abnormal state, and the layer position, timestamp, and deviation value of the measurement point are recorded to form an anomaly record. Based on the anomaly record, the deviation values ​​are stored in the anomaly data cache in chronological order of occurrence. At the same time, the frequency of anomalies at the same layer position is statistically analyzed. When the number of anomaly data points in the cache reaches the preset update trigger number, the arithmetic mean of all deviation values ​​in the cache is calculated as the deviation mean, and the standard deviation of the deviation values ​​is calculated as the deviation dispersion to obtain the anomaly statistical parameter set. Using the aforementioned anomaly statistical parameter set and a preset exponential smoothing coefficient, the center reference value in the reference thickness range parameter set is updated. The updated center reference value is equal to the product of the original center reference value and the exponential smoothing coefficient, plus the product of the mean deviation and the complement of the exponential smoothing coefficient. The standard deviation parameter is updated using the same method. The updated reference thickness range parameter set is stored and used for subsequent detection. All anomaly records, anomaly statistical parameter sets, and updated reference thickness range parameter sets are summarized to obtain the thickness anomaly detection result.

[0053] Specifically, the dynamic threshold adjustment mechanism during real-time monitoring is key to achieving accurate anomaly detection. Copper foil layers at different locations experience varying cumulative pressure, resulting in different normal thickness fluctuation ranges. Inner copper foil experiences greater pressure, allowing for a smaller allowable thickness deviation; outer copper foil experiences less pressure, allowing for a more relaxed tolerance range. This differentiated threshold setting better aligns with actual production practices.

[0054] Specifically, the dynamic threshold is calculated based on the mapping relationship between layer position and sensitivity identification threshold. Assuming the basic sensitivity identification threshold is T0, the dynamic threshold for the nth layer of copper foil can be adjusted by an adjustment coefficient. The adjustment coefficient is determined based on the historical anomaly rate and physical characteristics of that layer. The adjustment coefficient for inner layers is usually less than 1, making the threshold more stringent; the adjustment coefficient for outer layers may be greater than 1, appropriately relaxing the detection standard. This hierarchical threshold management improves the targeting of detection.

[0055] In one possible implementation, the exception data cache is designed using a circular queue structure, which ensures data sequence while preventing unlimited growth of storage space. When new exception data arrives, if the cache is full, the oldest data is removed. The selection of the preset update trigger number needs to balance update frequency and statistical reliability. Too small a number will lead to frequent updates and unstable statistical results, while too large a number will reduce the system's response speed. It is usually set to 20 to 50 exception points, a range that ensures statistical significance while reflecting changes in production status in a timely manner.

[0056] It's important to note that the application of exponential smoothing in parameter updates reflects a trade-off between old and new information. The exponential smoothing coefficient determines the relative importance of historical and new data. When the coefficient is close to 1, the updated parameters primarily retain the original values, resulting in slow changes; when the coefficient is close to 0, the impact of new data is greater, leading to more aggressive parameter updates. The complement of the exponential smoothing coefficient is 1 minus the smoothing coefficient, representing the weight of the new data. This update mechanism allows the system to maintain stability while adapting to gradual changes in production conditions.

[0057] Preferably, the utilization of layer location information is not only reflected in dynamic threshold adjustment but also in anomaly pattern recognition. By statistically analyzing the frequency of anomalies at different layers, it is possible to identify systemic problems at certain specific layers. If the anomaly frequency of a certain layer is significantly higher than that of other layers, it may indicate that the equipment at that location needs maintenance or that process parameters need adjustment. Timestamp information is used to analyze the temporal distribution characteristics of anomalies and identify periodic or trend-based quality fluctuations. The comprehensive output of anomaly detection results provides a comprehensive basis for subsequent quality control decisions. Anomaly records provide specific problem location information, facilitating traceability and analysis; the anomaly statistical parameter set reflects the current quality fluctuation status; and the updated baseline thickness range parameter set ensures that the detection standard can adapt to changes in production status. This multi-level information output mechanism not only meets the needs of real-time monitoring but also provides data support for long-term quality improvement. Through continuous learning and parameter updates, detection performance is continuously optimized, reducing false alarms and missed alarms.

[0058] S107. Based on the thickness anomaly detection results and historical winding process thickness measurement data, determine the thickness anomaly detection combination parameters.

[0059] Combining the anomaly detection results and historical winding thickness measurement data, the frequency of anomalies at each layer is calculated by dividing the number of anomalies at each position by the total number of measurements for that layer. The absolute value of the deviation at each anomaly point is calculated, and the average of all absolute deviations for the same layer is taken as the average deviation for that layer. The absolute value of the deviation at each anomaly point is divided by the average deviation for that layer to obtain a deviation ratio sequence. A quality feature dataset containing the layer number, anomaly frequency, and deviation ratio sequence is constructed. Based on the layer anomaly frequency in the quality feature dataset, layers with anomaly frequencies greater than twice the historical average anomaly frequency are identified as key adjustment layers. The 95th percentile is extracted from the deviation ratio sequence as the deviation feature value for that layer. The current compensation coefficient for the key adjustment layer is found in the dynamic compensation parameter library. The current compensation coefficient is multiplied by the deviation feature value to obtain a new compensation coefficient, and the parameters for the corresponding layer in the dynamic compensation parameter library are updated. The updated dynamic compensation parameter library is used to calculate the compensation for the original thickness data of the most recent production batch. The number of anomalies detected after compensation is counted. If the number of anomalies detected is reduced by more than a preset ratio compared to before the update, the updated parameters are retained. If the preset ratio is not reached, the anomaly judgment threshold is increased or decreased by a preset step size. The compensation and statistical process is repeated until the number of anomalies detected meets the requirements, and the thickness anomaly detection combination parameters are determined.

[0060] Specifically, the feedback and adjustment mechanism for quality monitoring is based on in-depth analysis of historical data. The calculation of layer anomaly frequencies reveals differences in quality stability at different locations. Certain layers may exhibit higher anomaly rates due to equipment wear, process parameter deviations, or changes in material properties. By quantifying these differences, the system can identify weak points that require focused attention.

[0061] Specifically, the construction of the deviation ratio sequence provides a standardized method for assessing the degree of anomalies. Assuming a measurement point in layer 10 has a deviation of 0.5 micrometers, while the average deviation for that layer is 0.2 micrometers, the deviation ratio is 2.5, indicating that the deviation at that point is 2.5 times the average level. This relativistic approach eliminates the differences in absolute deviations between different layers, making the comparison of anomaly degrees more fair and reasonable.

[0062] In one possible implementation, using twice the historical average frequency of anomalies as an identification criterion is statistically significant. Normal production fluctuations typically do not cause the frequency of anomalies to double; when the frequency of anomalies at a certain level reaches this level, it often indicates a systemic problem. The choice of the 95th percentile ensures the robustness of the adjustment; it represents that 95% of deviations are less than this value, excluding only 5% of extreme cases, thus considering most anomalies while avoiding the excessive influence of a few extreme values.

[0063] It should be noted that the compensation coefficients are updated using a multiplicative adjustment rather than an additive adjustment. This is because the compensation requirement is usually proportional to the original deviation. If the deviation characteristic value of a certain layer is 1.8, the original compensation coefficient is 0.5, and the updated compensation coefficient is 0.9, it means that this layer needs a stronger compensation force to correct the systematic deviation. This adjustment method maintains the relative relationship of compensation and avoids overcompensation or undercompensation.

[0064] Preferably, the validation process for feedback adjustment uses data from the most recent production batch, rather than all historical data. This has several advantages. Recent data better reflects the current production status, avoiding interference from factors such as equipment aging or process improvements that may exist in earlier data. Simultaneously, using a smaller dataset for validation allows for rapid feedback results, improving the timeliness of adjustments. The preset ratio needs to balance improvement effectiveness and system stability. Requiring an anomaly detection count to decrease by more than 50% before accepting an update may lead to the rejection of many beneficial small improvements; conversely, a threshold that is too low, such as requiring only a 10% reduction, may accept false improvements caused by random fluctuations. Typically, an improvement of 20% to 30% is set as the acceptance standard. The incremental adjustment of the threshold forms an iterative optimization process. If there are still many anomalies after the compensation parameter update, the judgment threshold may be too strict and needs to be appropriately relaxed; conversely, if there are too few anomalies, there may be a risk of missed detections, requiring the threshold to be tightened. This two-way adjustment mechanism ensures that the optimal balance between detection sensitivity and stability can be found, achieving synergistic optimization of measurement accuracy and detection performance.

[0065] This invention provides a copper foil thickness analysis system, which employs the above-mentioned method and mainly includes: The data acquisition module is used to acquire the tension value and thickness measurement results of each layer position during the copper foil winding process. The correlation analysis module is used to identify the compression deformation law of copper foil with different winding layers by analyzing the correlation between the thickness change and tension distribution of copper foil with different winding layers, and to generate compression deformation analysis results based on the compression deformation law of copper foil with different winding layers. The compression deformation analysis module is used to analyze the deformation mechanism based on the compression deformation analysis results, combined with the obtained interlayer relative displacement and torsional angle, and obtain stress distribution pattern data. The thickness compensation module is used to compensate the thickness measurement data based on the stress distribution pattern data to obtain the compensated thickness measurement data. The data compensation module is used to set the normal thickness range and anomaly judgment threshold based on the compensated thickness measurement data, and to determine the sensitivity identification threshold parameter for thickness anomaly detection. The threshold setting module is used to determine the thickness anomaly state based on the sensitivity identification threshold parameter of the thickness anomaly detection, and obtain the thickness anomaly detection result. An anomaly detection module is used to determine the thickness anomaly detection combination parameters by combining the thickness anomaly detection results with the thickness measurement data of the historical winding process.

[0066] A computer-readable storage medium storing a computer program that, when executed by a processor, implements the method described above.

[0067] A computer program product includes a computer program that, when executed by a processor, implements the steps of the above-described method.

[0068] The above description is only a preferred embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any equivalent substitutions or modifications made by those skilled in the art within the scope of the technology disclosed in the present invention, based on the technical solution and inventive concept of the present invention, should be covered within the scope of protection of the present invention.

Claims

1. A method for analyzing the thickness of copper foil, characterized in that, The method includes: Obtain the tension value and thickness measurement results of each layer position during the copper foil winding process; By analyzing the correlation between the thickness variation and tension distribution of copper foil under different winding layers, the compression deformation law of copper foil under different winding layers is identified, and the compression deformation analysis results are generated based on the compression deformation law of copper foil under different winding layers. Based on the compression deformation analysis results, the deformation mechanism is analyzed in conjunction with the obtained interlayer relative displacement and torsional angle to obtain stress distribution pattern data. Based on the stress distribution pattern data, the thickness measurement data is compensated to obtain the compensated thickness measurement data. Based on the compensated thickness measurement data, the normal thickness range and anomaly judgment threshold are set to determine the sensitivity identification threshold parameters for thickness anomaly detection. The thickness anomaly state is judged based on the sensitivity identification threshold parameter of the thickness anomaly detection, and the thickness anomaly detection result is obtained. Based on the thickness anomaly detection results and historical winding process thickness measurement data, the combined parameters for thickness anomaly detection are determined; The method identifies the compression deformation pattern of copper foil with different winding layers by analyzing the correlation between the thickness variation and tension distribution under different winding layers. Based on this pattern, compression deformation analysis results are generated, including: constructing a three-dimensional data matrix of layers, tension, and thickness based on real-time tension values ​​at different radial positions and the original thickness measurements at corresponding layer positions; calculating the thickness variation of each layer; generating a quantitative correlation index between tension and thickness variation; calculating the radial pressure generated by each layer based on the quantitative correlation index, combined with the winding radius and copper foil width; and summing these values ​​to obtain the cumulative pressure value borne by the inner copper foil layer; and generating compression deformation analysis results based on the cumulative pressure value and the pre-acquired elastic modulus of the copper foil material. The step of analyzing the deformation mechanism based on the compression deformation analysis results, combined with the obtained interlayer relative displacement and torsional angle, and obtaining stress distribution pattern data includes: extracting relative displacement data and torsional angle data of adjacent layers, measuring the vertical distance between the surfaces of adjacent copper foil layers, calculating the tangential shear strain, calculating the torsional shear strain by combining the torsional angle and the radius of the current layer, adding the tangential shear strain and the torsional shear strain to generate a total shear strain value, accumulating the total shear strain value layer by layer to generate the cumulative shear deformation of each layer; extracting the compression deformation amount from the compression deformation analysis results, fitting the relationship between the cumulative shear deformation amount and the compression deformation amount, determining the regression coefficient, substituting it into a preset coupling function form to generate a coupling relationship expression, substituting the compression deformation amount and the cumulative shear deformation amount to calculate the equivalent stress value under the composite deformation state, and generating stress distribution pattern data. The compensation of thickness measurement data based on the stress distribution pattern data to obtain compensated thickness measurement data includes: extracting stress values ​​and real-time tension values, combining the deformation data in the compression deformation analysis results to calculate the thickness deviation, and generating a dynamic compensation parameter library; based on the real-time tension value and layer position, searching for a matching category in the dynamic compensation parameter library, extracting the average tension, average deformation, and average thickness deviation of that category, and calculating the dynamic compensation coefficient; and using the dynamic compensation coefficient to correct the original thickness measurement value to obtain compensated thickness measurement data.

2. The method for analyzing copper foil thickness according to claim 1, characterized in that, The acquisition of tension values ​​and corresponding thickness measurements at various layers during the copper foil winding process includes: Tension sensors are placed at various layers of the copper foil winding device to collect tension signals during the winding process. The tension signals are amplified and filtered to generate a real-time tension value data sequence. The layer number identifier and acquisition time information in the real-time tension value data sequence are marked. A thickness measuring device is deployed with multiple measuring points distributed along the width direction of the copper foil to collect thickness data at each measuring point. The average thickness data of all measuring points is calculated to generate the average thickness value of the copper foil layer. The layer number identifier and measurement time corresponding to the average thickness value are marked. The real-time tension value data sequence is matched with the average thickness value. The correspondence is determined by the layer number identifier and the time difference to generate a corresponding data group containing the layer number, tension value, and thickness value. The comparison result of the tension value of the corresponding data group with the preset tension threshold is judged. The corresponding data group that exceeds the threshold and its adjacent layer number data groups are marked. All corresponding data groups are stored.

3. The method for analyzing copper foil thickness according to claim 1, characterized in that, The calculation of the dynamic compensation coefficient includes: finding a category that matches the real-time tension value, extracting the mean value of the parameters of the category, calculating the distance ratio between the real-time tension value and the mean value of the category, and generating the dynamic compensation coefficient.

4. The method for analyzing copper foil thickness according to claim 1, characterized in that, The thickness anomaly detection combination parameters are determined by combining the thickness anomaly detection results with historical winding process thickness measurement data, including: Combining the anomaly detection results and historical winding process thickness measurement data, the frequency of anomalies at each layer position is calculated by dividing the number of anomalies at each position by the total number of measurements for that layer. Layers with anomaly frequencies greater than twice the historical average anomaly frequency are identified as key adjustment layers, and deviation characteristic values ​​are extracted from these key adjustment layers. The current compensation coefficient for the key adjustment layer is then searched in the dynamic compensation parameter library. The current compensation coefficient is multiplied by the deviation characteristic value to obtain a new compensation coefficient, and the parameters for the corresponding layer in the dynamic compensation parameter library are updated. The updated dynamic compensation parameter library is used to perform compensation calculations on the original thickness data of the most recent production batch. The number of anomalies detected in the compensated thickness data is counted, and the thickness anomaly detection combination parameters are determined.

5. A copper foil thickness analysis system, which employs the method as described in claim 1, characterized in that, The system includes: The data acquisition module is used to acquire the tension value and thickness measurement results of each layer position during the copper foil winding process. The correlation analysis module is used to identify the compression deformation law of copper foil with different winding layers by analyzing the correlation between the thickness change and tension distribution of copper foil with different winding layers, and to generate compression deformation analysis results based on the compression deformation law of copper foil with different winding layers. The compression deformation analysis module is used to analyze the deformation mechanism based on the compression deformation analysis results, combined with the obtained interlayer relative displacement and torsional angle, and obtain stress distribution pattern data. The thickness compensation module is used to compensate the thickness measurement data based on the stress distribution pattern data to obtain the compensated thickness measurement data. The data compensation module is used to set the normal thickness range and anomaly judgment threshold based on the compensated thickness measurement data, and to determine the sensitivity identification threshold parameter for thickness anomaly detection. The threshold setting module is used to determine the thickness anomaly state based on the sensitivity identification threshold parameter of the thickness anomaly detection, and obtain the thickness anomaly detection result. An anomaly detection module is used to determine the thickness anomaly detection combination parameters by combining the thickness anomaly detection results with the thickness measurement data of the historical winding process.

6. A computer-readable storage medium storing a computer program that, when executed by a processor, implements the method as described in any one of claims 1-4.

7. A computer program product, comprising a computer program, characterized in that, When the computer program is executed by a processor, it implements the steps of the method of claim 1.

Images