A mold parameter adjustment method and system

Abstract

The present application relates to the technical field of die cutting processing, and particularly discloses a die parameter adjusting method and system, wherein the method comprises the following steps: obtaining a cutting edge sharpness feature and a material hard point feature; obtaining an edge quality score and a size deviation feature of a cut edge; obtaining a material local structure unevenness feature; estimating a die wear state feature; inputting the cutting edge sharpness feature, the material hard point feature, the edge quality score, the size deviation feature, the material local structure unevenness feature and the die wear state feature into a pre-trained adaptive parameter configuration model to output die adjusting parameters; and controlling an execution mechanism of a die cutting machine based on the die adjusting parameters to perform die cutting. The method of the present application effectively deals with quality fluctuations caused by individual differences of dies, wear and superposition of material local unevenness, improves the stability of die cutting processing and the consistency of product quality, and realizes closed-loop control and optimization of the die cutting process through sensing, analysis and intelligent decision-making.

Description

Technical Field

[0001] This application relates to the field of die-cutting technology, and more specifically, to a method and system for adjusting die parameters. Background Technology

[0002] In the field of die-cutting, traditional processes rely on empirical parameter settings, which makes it difficult to cope with the quality fluctuations caused by the superposition of micro-individual differences in molds and local non-uniformity of materials.

[0003] Specifically, even with the same model of mold, manufacturing tolerances, microscopic edge curling, and heat treatment fluctuations lead to significant individual differences in actual cutting performance. Traditional fixed-parameter strategies cannot adapt to the dynamic attenuation and regional fluctuations in blade sharpness (such as increased burr stimulation caused by localized blunting). Simultaneously, the random distribution of fiber orientation and uneven bonding in the laminated structure of rolled materials during processing cause dynamic changes in the tear resistance and deformation behavior of the material in the die-cutting area. While existing vision inspection systems can capture edge defects, they struggle to establish a real-time correlation model between the material's internal microstructure and process parameters. More critically, cumulative mold wear and localized hard spots in the material create a coupled interference. Traditional static compensation models cannot decouple the combined effects of these two factors on the force distribution, resulting in lag in parameter adjustment or even reverse compensation, ultimately hindering the mass production stability of highly consistent products.

[0004] There is currently no effective technical solution to the above problems. Summary of the Invention

[0005] The purpose of this application is to provide a method and system for adjusting mold parameters, which comprehensively senses the blade sharpness, cutting edge characteristics, local microscopic characteristics of the material, and mold wear characteristics to generate mold adjustment parameters, so as to realize the adaptive adjustment and control of the die-cutting machine.

[0006] In a first aspect, this application provides a method for adjusting mold parameters, applied in a die-cutting machine, the method comprising the following steps:

[0007] S1. Obtain transient force distribution data, material surface image data, thickness distribution data, density distribution data, and mold usage history of the die-cutting area of ​​the workpiece;

[0008] S2. Obtain the blade sharpness characteristics and material hardness characteristics based on the transient force distribution data;

[0009] S3. Obtain the edge quality score and dimensional deviation characteristics of the cut edge based on the material surface image data;

[0010] S4. Obtain the characteristics of local structural inhomogeneity of the material based on thickness distribution data and density distribution data;

[0011] S5. Estimate the wear characteristics of the mold based on the mold's usage history;

[0012] S6. Input the blade sharpness feature, material hard point feature, edge quality score, size deviation feature, material local structural non-uniformity feature, and mold wear state feature into the pre-trained adaptive parameter configuration model to output mold adjustment parameters.

[0013] S7. Based on the mold adjustment parameters, control the actuator of the die-cutting machine to perform die cutting.

[0014] The die parameter adjustment method of this application comprehensively perceives various factors affecting die-cutting quality by collecting transient stress, material images, local physical properties of materials, and historical data of the die during the die-cutting process. By integrating multi-source data and generating die adjustment parameters based on an adaptive parameter configuration model, the die-cutting machine can achieve adaptive adjustment and control. This makes the die-cutting parameters no longer fixed empirical values, but dynamically adjusted according to the current die state, material characteristics, and preliminary cutting results.

[0015] The aforementioned mold parameter adjustment method, wherein step S2 includes:

[0016] S21. Perform stress waveform analysis, peak value analysis, and distribution non-uniformity analysis on the transient force distribution data in sequence to extract the blade sharpness characteristics and material hard point characteristics.

[0017] This step, through a series of processing and analysis, transforms complex transient stress data into structured, quantifiable feature data. This feature data accurately describes the sharpness of the cutting edge and the distribution of material hard spots, providing accurate input information for subsequent adaptive adjustment of mold parameters.

[0018] The aforementioned mold parameter adjustment method, wherein step S21 includes:

[0019] S211. The transient force distribution data is analyzed by fast Fourier transform to extract the main frequency component and harmonic components and construct a spectrum diagram.

[0020] S212. Calculate the spectral energy concentration based on the spectrum diagram, and convert the blade sharpness characteristics based on the spectral energy concentration.

[0021] S213. Use a peak detection algorithm to identify peak points in transient force distribution data, and record the force value and location coordinates corresponding to each peak point;

[0022] S214. Based on the force value and position coordinates of the peak point, calculate the uniformity coefficient of force distribution in the die-cut area, and determine the hard point characteristics of the material based on the uniformity coefficient of force distribution.

[0023] The above processing method decomposes the force signal into frequency components and analyzes the local peak distribution, which can distinguish and quantify the overall vibration characteristics related to the blade sharpness as well as the local force changes and distribution related to the material hard points. This solves the problem of difficulty in distinguishing and quantifying the influence of the two, and provides more targeted input for subsequent mold parameter adjustment.

[0024] The aforementioned mold parameter adjustment method, wherein step S3 includes:

[0025] S31. The material surface image data is sequentially subjected to edge detection, defect identification, and size measurement to obtain the edge quality score and size deviation characteristics of the cut edge.

[0026] The mold parameter adjustment method, wherein step S4 includes:

[0027] S41. Perform local change rate detection and abrupt change point detection on the thickness distribution data and density distribution data in sequence to obtain the local structural inhomogeneity characteristics of the material.

[0028] The aforementioned mold parameter adjustment method, wherein step S41 includes:

[0029] S411. Perform Gaussian filtering on the thickness distribution data and density distribution data respectively to obtain smoothed thickness data and density distribution data;

[0030] S412. Calculate the first difference between the smoothed thickness data and density distribution data to obtain the thickness change rate and density change rate.

[0031] S413. Count the number of local mutation points within a unit area, wherein the local mutation point is a location where the thickness change rate is greater than a preset thickness change rate threshold and / or the density change rate is greater than a preset density change rate threshold.

[0032] S414. Calculate the proportion of the local abrupt change points and convert the proportion into the characteristics of local structural inhomogeneity of the material.

[0033] The aforementioned mold parameter adjustment method includes local abrupt change points, which include thickness abrupt change points and density abrupt change points. Thickness abrupt change points are locations where the thickness change rate is greater than a preset thickness change rate threshold, and density abrupt change points are locations where the density change rate is greater than a preset density change rate threshold.

[0034] Step S414 includes:

[0035] The proportions of thickness abrupt change points and density abrupt change points are calculated separately, and the proportions of thickness abrupt change points and density abrupt change points are weighted based on preset weighting coefficients to obtain the local structural inhomogeneity characteristics of the material.

[0036] The aforementioned mold parameter adjustment method, wherein step S5 includes:

[0037] S51. Obtain the cumulative die-cutting length, average die-cutting pressure, and overload frequency of the mold based on the historical statistics of mold usage;

[0038] S52. Based on a preset wear model or wear mapping table, calculate the fatigue value of the mold according to the cumulative die-cutting length, average die-cutting pressure and overload frequency, as a characteristic of the mold wear state.

[0039] The mold parameter adjustment method further includes a step performed after step S7:

[0040] S8. Based on the visual inspection system or offline quality inspection results, obtain the actual product die-cutting quality data, and input it as a feedback signal into the adaptive parameter configuration model to iteratively optimize the adaptive parameter configuration model.

[0041] Secondly, this application also provides a mold parameter adjustment system for use in a die-cutting machine, the system comprising:

[0042] The acquisition module is used to acquire transient force distribution data, material surface image data, thickness distribution data, density distribution data, and mold usage history of the die-cutting area of ​​the workpiece.

[0043] The first analysis module is used to obtain the blade sharpness characteristics and material hard point characteristics based on the transient force distribution data.

[0044] The second analysis module is used to obtain the edge quality score and dimensional deviation characteristics of the cut edge based on the material surface image data.

[0045] The third analysis module is used to obtain the local structural inhomogeneity characteristics of the material based on the thickness distribution data and density distribution data;

[0046] The fourth analysis module is used to estimate the wear characteristics of the mold based on the mold's usage history;

[0047] The parameter generation module is used to input the blade sharpness feature, material hard point feature, edge quality score, size deviation feature, material local structural non-uniformity feature, and mold wear state feature into a pre-trained adaptive parameter configuration model to output mold adjustment parameters.

[0048] The actuator control module is used to control the actuator of the die-cutting machine to perform die-cutting based on the mold adjustment parameters.

[0049] The die parameter adjustment system of this application comprehensively perceives various factors affecting die-cutting quality by collecting transient stress, material images, local physical properties of materials, and historical data of the die during the die-cutting process. By integrating multi-source data and generating die adjustment parameters based on an adaptive parameter configuration model, the die-cutting machine achieves adaptive adjustment and control. This makes the die-cutting parameters no longer fixed empirical values, but dynamically adjusted according to the current die state, material characteristics, and preliminary cutting results.

[0050] As described above, this application provides a method and system for adjusting die parameters. The method comprehensively perceives various factors affecting die-cutting quality by collecting transient stress, material images, local material physical properties, and historical die data during the die-cutting process. It then integrates multi-source data to generate die adjustment parameters based on an adaptive parameter configuration model, achieving adaptive adjustment control of the die-cutting machine. This ensures that die-cutting parameters are no longer fixed empirical values ​​but are dynamically adjusted based on the current die state, material characteristics, and preliminary cutting results. This adaptive adjustment mechanism effectively addresses quality fluctuations caused by individual die differences, wear, and the superposition of local material inhomogeneities, improving the stability of die-cutting processing and the consistency of product quality. Through perception, analysis, and intelligent decision-making, this application achieves closed-loop control and optimization of the die-cutting process. Attached Figure Description

[0051] Figure 1 A flowchart illustrating a method for adjusting mold parameters provided in some embodiments of this application.

[0052] Figure 2 A flowchart of a method for adjusting mold parameters provided in some embodiments of this application.

[0053] Figure 3 This is a schematic diagram of the structure of a mold parameter adjustment system provided in some embodiments of this application.

[0054] Figure 4 A schematic diagram of the structure of a mold parameter adjustment system provided in some embodiments of this application.

[0055] Reference numerals: 201, Acquisition module; 202, First analysis module; 203, Second analysis module; 204, Third analysis module; 205, Fourth analysis module; 206, Parameter generation module; 207, Actuator control module; 208, Model optimization module; 209, Normalization processing module. Detailed Implementation

[0056] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of this application, and not all of the embodiments. The components of the embodiments of this application described and shown in the accompanying drawings can generally be arranged and designed in various different configurations. Therefore, the following detailed description of the embodiments of this application provided in the accompanying drawings is not intended to limit the scope of the claimed application, but merely represents selected embodiments of this application. All other embodiments obtained by those skilled in the art based on the embodiments of this application without inventive effort are within the scope of protection of this application.

[0057] It should be noted that similar reference numerals and letters in the following figures indicate similar items; therefore, once an item is defined in one figure, it does not need to be further defined and explained in subsequent figures. Furthermore, in the description of this application, terms such as "first," "second," etc., are used only to distinguish descriptions and should not be construed as indicating or implying relative importance.

[0058] Firstly, please refer to Figure 1 and Figure 2 This application provides a method for adjusting mold parameters in a die-cutting machine, comprising the following steps:

[0059] S1. Obtain transient force distribution data, material surface image data, thickness distribution data, density distribution data, and mold usage history of the die-cutting area of ​​the workpiece;

[0060] S2. Obtain the blade sharpness characteristics and material hardness characteristics based on transient force distribution data;

[0061] S3. Obtain the edge quality score and dimensional deviation characteristics of the cut edge based on the material surface image data;

[0062] S4. Obtain the characteristics of local structural inhomogeneity of the material based on thickness distribution data and density distribution data;

[0063] S5. Estimate the wear characteristics of the mold based on its usage history;

[0064] S6. Input the blade sharpness features, material hard point features, edge quality score, dimensional deviation features, material local structural non-uniformity features, and mold wear state features into the pre-trained adaptive parameter configuration model to output mold adjustment parameters.

[0065] S7. The die-cutting machine's actuator is controlled based on the mold adjustment parameters to perform die cutting.

[0066] Specifically, step S1 is the data acquisition stage. Acquiring transient force distribution data reflects the real-time mechanical state during the die-cutting process and is directly related to blade sharpness and material hardness. Acquiring material surface image data provides intuitive information about the die-cutting results, used to assess edge quality and dimensions. Acquiring thickness and density distribution data reflects the physical inhomogeneity of the material itself. Die usage history provides clues about the overall state of the die changing over time. This data forms the basis for subsequent analysis and parameter adjustment. Thus, it is possible to comprehensively understand the key influencing factors in the die-cutting process.

[0067] More specifically, step S2 characterizes the cutting performance of the blade and the properties of high-resistance regions in the material by processing the transient force distribution data to obtain blade sharpness and material hardness characteristics. This step transforms the raw force data into physically meaningful features, which helps to quantify the microscopic state of the mold and the material.

[0068] More specifically, step S3 characterizes the state and dimensional accuracy of the cut edge by processing the edge quality score and dimensional deviation features of the cut edge obtained from the material surface image data. This step quantifies the smoothness, burrs, and other quality indicators of the cut edge, as well as the deviation between the actual and target dimensions, through image analysis. These features directly measure the product quality under the current die-cutting parameters, providing direct quality feedback.

[0069] More specifically, step S4 uses the local structural inhomogeneity characteristics of the material obtained by processing thickness and density distribution data to quantify changes in the material's internal structure. This step quantifies the material's inherent variability, which helps in understanding the material's impact on the die-cutting process.

[0070] More specifically, step S5 uses the mold wear condition characteristics obtained through historical mold usage estimates to quantify the degree of mold aging. This step utilizes historical data to assess the macroscopic condition of the mold, providing information on its long-term condition.

[0071] More specifically, step S6 takes the acquired blade sharpness features, material hard point features, edge quality score, dimensional deviation features, material local structural inhomogeneity features, and die wear state features as inputs and feeds them into a pre-trained adaptive parameter configuration model. This allows the adaptive parameter configuration model to output specific die adjustment parameters based on the learned features and the required die adjustment parameters. Specifically, the adaptive parameter configuration model learns the relationship between different feature combinations and optimal die-cutting parameters, and can comprehensively consider the die state, material properties, and current cutting results to output die adjustment parameters tailored to the current working condition.

[0072] More specifically, step S7 controls the actuator of the die-cutting machine to perform die-cutting based on these mold adjustment parameters. The process may involve driving the servo motor and hydraulic cylinder of the die-cutting machine, adjusting the die-cutting pressure, speed, or mold gap to optimize the die-cutting process.

[0073] More specifically, the mold parameter adjustment method of this application aims to solve the problem of unstable die-cutting quality caused by changes in mold and material by collecting multi-source data and using a model to adjust parameters.

[0074] The die parameter adjustment method of this application comprehensively perceives various factors affecting die-cutting quality by collecting transient stress, material images, local physical properties of the material, and historical die data during the die-cutting process. It then integrates multi-source data to generate die adjustment parameters based on an adaptive parameter configuration model, achieving adaptive adjustment control of the die-cutting machine. This ensures that die-cutting parameters are no longer fixed empirical values ​​but are dynamically adjusted based on the current die state, material characteristics, and preliminary cutting results. This adaptive adjustment mechanism effectively addresses quality fluctuations caused by individual die differences, wear, and the superposition of local material inhomogeneities, improving the stability of die-cutting processing and the consistency of product quality. Through perception, analysis, and intelligent decision-making, the method of this application achieves closed-loop control and optimization of the die-cutting process.

[0075] In some preferred embodiments, step S2 includes:

[0076] S21. Perform stress waveform analysis, peak value analysis, and distribution non-uniformity analysis on the transient force distribution data in sequence to extract the blade sharpness characteristics and material hard point characteristics.

[0077] Specifically, in some implementations, transient force distribution data can be filtered and denoised before subsequent force waveform analysis, peak analysis, and distribution non-uniformity analysis are performed, thereby removing interference components from the data and improving data quality.

[0078] More specifically, this step performs force waveform analysis on transient force distribution data, which can extract patterns that reflect changes in force over time or location, and these patterns are associated with the blade sharpness.

[0079] More specifically, this step performs a distribution non-uniformity analysis, quantifying the distribution of force within the die-cutting area. This distribution is correlated with material hard spots or localized blunting of the cutting edge. Through these analytical steps, characteristic data reflecting the sharpness of the cutting edge and the state of material hard spots are extracted from the transient force data.

[0080] More specifically, step S21 addresses the problem of accurately extracting information reflecting blade sharpness and material hardness by directly analyzing raw data through multi-stage processing and analysis of transient force distribution data collected during the die-cutting process. First, force waveform analysis is performed on the force data, such as analyzing the rising edge, falling edge, plateau characteristics, or frequency components of the force curve. These waveform characteristics are related to the resistance characteristics of the blade when cutting the material, thus reflecting the blade sharpness. Then, a peak detection algorithm is used to identify local maximum points in the force data, recording the force values ​​and spatial locations corresponding to these peak points. These peak points indicate high-resistance areas encountered during die-cutting, potentially corresponding to material hardness. Finally, based on the identified peak points and their force values, the non-uniformity of the force distribution is calculated to reflect the concentration of material hardness or the influence of local blade dulling through the degree of non-uniformity. This step, through a series of processing and analysis, transforms complex transient stress data into structured, quantifiable feature data. This feature data accurately describes the sharpness of the cutting edge and the distribution of material hard spots, providing accurate input information for subsequent adaptive adjustment of mold parameters.

[0081] In some preferred embodiments, step S21 includes:

[0082] S211. Use Fast Fourier Transform to perform force waveform analysis on transient force distribution data, extract the main frequency component and harmonic components, and construct a spectrum diagram;

[0083] S212. Calculate the spectral energy concentration based on the spectrum diagram, and convert the blade sharpness characteristics based on the spectral energy concentration.

[0084] S213. Use a peak detection algorithm to identify peak points in transient force distribution data, and record the force value and location coordinates corresponding to each peak point;

[0085] S214. Based on the stress value and position coordinates of the peak point, calculate the uniformity coefficient of stress distribution in the die-cut area, and determine the hard point characteristics of the material based on the uniformity coefficient of stress distribution.

[0086] Specifically, step S211 uses Fast Fourier Transform to perform force waveform analysis on transient force distribution data, extracts the main frequency component and harmonic components, and constructs a spectrum diagram, thereby converting the time domain information of transient force distribution data into frequency domain information to reveal the distribution characteristics of force with frequency.

[0087] More specifically, step S212 calculates the concentration of spectral energy by analyzing the energy distribution of the spectrum. The concentration of spectral energy in a specific frequency range is related to the sharpness of the blade. Calculating the concentration of spectral energy in a specific frequency range can be used to convert the blade sharpness characteristics.

[0088] More specifically, step S213 uses a peak detection algorithm to identify peak points in the transient force distribution data and records the force values ​​and locations of these peak points. Step S214 calculates the force distribution uniformity coefficient of the die-cutting area based on the force value and location coordinates of the peak points, and determines the material hard point characteristics based on the force distribution uniformity coefficient. By analyzing the distribution of peak forces and calculating the force distribution uniformity coefficient, the influence and distribution of material hard points can be quantified.

[0089] More specifically, the above processing method decomposes the force signal into frequency components and analyzes the local peak distribution, which can distinguish and quantify the overall vibration characteristics related to the blade sharpness as well as the local force changes and distribution related to the material hard points. This solves the problem of difficulty in distinguishing and quantifying the influence of the two, and provides more targeted input for subsequent mold parameter adjustment.

[0090] In some preferred embodiments, step S214 includes:

[0091] S2141. Calculate the deviation between the force value at each peak point and the average force value in the die-cut area;

[0092] S2142. Calculate the sum of squares of the deviation values ​​based on each deviation value, and divide the sum of squares of the deviation values ​​by the number of peak points to obtain the force variance value;

[0093] S2143. Divide the average force by the force variance to obtain the force distribution uniformity coefficient. Determine the material hard point distribution uniformity based on the force distribution uniformity coefficient and the preset mapping table.

[0094] Specifically, step S2141 quantifies the degree to which each local stress point deviates from the overall average level by calculating the deviation between the stress value at each peak point and the average stress value in the die-cutting area.

[0095] More specifically, the variance value calculated in step S2141 provides a statistical measure of the magnitude of force fluctuations, reflecting the magnitude of force fluctuations around the average value. Squaring the deviation can amplify the impact of larger deviations.

[0096] More specifically, the force distribution uniformity coefficient calculated in step S2143 combines the average force level with the degree of force fluctuation, providing a quantitative uniformity index. Based on this quantitative uniformity coefficient, combined with a preset mapping table, the distribution uniformity of material hard points can be objectively determined, thereby providing a basis for subsequent mold parameter adjustment and improving the accuracy of parameter adjustment.

[0097] In some preferred embodiments, step S3 includes:

[0098] S31. Perform edge detection, defect identification, and size measurement on the material surface image data in sequence to obtain the edge quality score and size deviation characteristics of the cut edge.

[0099] Specifically, edge detection is used to determine the actual cutting boundary of the material. Defect identification locates and marks existing defects on the detected edges. Dimensional measurement calculates the actual dimensions of the workpiece based on the detected edges. Through these steps, the method of this application can extract quantified edge quality information and dimensional information from the raw image data for calculating edge quality scores and dimensional deviation features. This processing flow provides a specific path from raw image data to the desired features, making the feature extraction process more explicit and controllable, thereby improving the accuracy and reliability of the obtained edge quality scores and dimensional deviation features, and providing more effective data input for subsequent adaptive parameter configuration models.

[0100] In some preferred embodiments, step S31 includes:

[0101] S311. Use the Canny operator to perform edge detection on the material surface image data to obtain an initial edge image;

[0102] S312. Use Gaussian filtering to smooth the initial edge image to obtain a smoothed edge image;

[0103] S313. For smooth edge images, fit the cutting edge to obtain the cutting edge curve;

[0104] S314. Calculate the curvature of the cutting edge curve, and determine the edge with curvature exceeding the threshold as a defect edge;

[0105] S315. Calculate the defect density by counting the length and number of defect edges, and calculate the edge quality score based on the defect density;

[0106] S316. Obtain dimensional deviation features based on the position and size of the cutting edge.

[0107] Specifically, step S311 uses the Canny operator to perform edge detection on the material surface image data, aiming to identify regions in the image with significant changes in pixel grayscale and generate an initial edge image containing potential edge information.

[0108] More specifically, step S312 uses Gaussian filtering to smooth the initial edge image in order to suppress noise, reduce false edges, and make subsequent processing more stable.

[0109] More specifically, step S313, for the smooth edge image, fits the edges to obtain the cutting edge curve representing the cutting edge, and determines the main shape and position of the cutting edge. More specifically,

[0110] More specifically, step S314 calculates the curvature of the cutting edge curve and quantifies and identifies defective edges that deviate from the ideal shape by comparing it with a preset threshold.

[0111] More specifically, step S315 counts the length and number of defective edges, calculates the defect density according to a preset mapping relationship or formula, and then converts the severity and distribution of edge defects into a numerical index, namely the edge quality score, according to the preset mapping relationship or formula. This score reflects the overall smoothness and integrity of the edge.

[0112] More specifically, step S316 obtains dimensional deviation characteristics based on the analysis of the cutting edge relative to the design position and size, reflecting the machining accuracy.

[0113] More specifically, these edge quality scores and dimensional deviation features, as quantitative indicators, can accurately and stably reflect the state and precision of the cut edges, providing reliable input data for subsequent adaptive adjustment of mold parameters, thereby solving the problem that traditional methods are difficult to accurately assess edge quality and dimensional deviation.

[0114] In some preferred embodiments, step S4 includes:

[0115] S41. Perform local change rate detection and abrupt change point detection on the thickness distribution data and density distribution data in sequence to obtain the local structural inhomogeneity characteristics of the material.

[0116] Specifically, local rate of change detection aims to identify the spatial trends and rates of change in material thickness or density. The inhomogeneity of material structures is often manifested in rapid changes in its thickness or density; the above steps, by calculating the rate of change, can quantify the degree of this inhomogeneity.

[0117] More specifically, mutation point detection further focuses on regions or points where the rate of change exceeds a certain threshold. These mutation points may correspond to hard spots, voids, lamination defects, or other structural anomalies within the material, which are key factors leading to uneven stress and decreased cutting quality during the die-cutting process.

[0118] More specifically, step S41 extracts key information reflecting the material's structural inhomogeneity from the raw data by detecting local rate of change and abrupt change points in the thickness and density distribution data. This information is integrated into a local structural inhomogeneity feature, which can more accurately describe the actual structural state of the material in the die-cutting area. This feature is then input into an adaptive parameter configuration model, which adjusts the die parameters according to the actual degree of material inhomogeneity, thereby improving die-cutting stability and product quality, and resolving die-cutting problems caused by local structural inhomogeneity.

[0119] In some preferred embodiments, step S41 includes:

[0120] S411. Perform Gaussian filtering on the thickness distribution data and density distribution data respectively to obtain smoothed thickness data and density distribution data;

[0121] S412. Calculate the first difference between the smoothed thickness data and density distribution data to obtain the thickness change rate and density change rate.

[0122] S413. Count the number of local mutation points within a unit area. Local mutation points are locations where the thickness change rate is greater than a preset thickness change rate threshold and / or the density change rate is greater than a preset density change rate threshold.

[0123] S414. Calculate the proportion of local abrupt change points and convert the proportion into the characteristics of local structural inhomogeneity of the material.

[0124] Specifically, step S411 performs Gaussian filtering on the thickness distribution data and density distribution data respectively. The purpose is to smooth the data, reduce the impact of measurement noise on subsequent analysis, and make the data better reflect the structural trend of the material itself.

[0125] More specifically, step S412 detects the rate or gradient of change of thickness and density in space by calculating the first difference between the smoothed thickness data and density distribution data, thereby identifying the location where the material structure undergoes significant changes.

[0126] More specifically, local mutation points indicate the presence of local structural anomalies or uneven regions within the material. Step S413, combined with step S414, counts the number of these local mutation points within the die-cutting area and calculates their proportion of the total number of points within the die-cutting area. This provides an objective and quantitative indicator to describe the degree of unevenness in the local structure of the material. This quantitative feature can more fully reflect the actual structural characteristics of the material, enabling subsequent mold parameter configuration to more accurately consider the microstructural differences of the material itself, thereby optimizing the die-cutting process and improving product consistency.

[0127] In some preferred embodiments, local abrupt change points include thickness abrupt change points and density abrupt change points. Thickness abrupt change points are locations where the thickness change rate is greater than a preset thickness change rate threshold, and density abrupt change points are locations where the density change rate is greater than a preset density change rate threshold.

[0128] Step S414 includes:

[0129] The proportions of thickness abrupt change points and density abrupt change points are calculated separately. The proportions of thickness abrupt change points and density abrupt change points are weighted based on preset weighting coefficients to obtain the characteristics of local structural inhomogeneity of the material.

[0130] Specifically, the above processing method clearly distinguishes local abrupt change points into thickness abrupt change points and density abrupt change points. Thickness abrupt change points are locations where the material thickness change rate exceeds a preset threshold, and density abrupt change points are locations where the material density change rate exceeds a preset threshold. When calculating the local structural inhomogeneity characteristics of the material, instead of simply calculating the total proportion of all abrupt change points, the proportions of thickness abrupt change points and density abrupt change points within the die-cutting region are calculated separately. Then, these two independent proportions are weighted and summed using preset weighting coefficients to obtain the final local structural inhomogeneity characteristics of the material.

[0131] More specifically, this differentiated and weighted approach can more precisely reflect the inhomogeneity of the local structure of the material, distinguishing the contributions of thickness and density variations to the overall inhomogeneity, and overcoming the inaccurate evaluation problem caused by simply combining the two. By adjusting the weighting coefficients, the inhomogeneity of the local structure of the material can be more accurately quantified based on the differences in sensitivity of different materials or different die-cutting processes to thickness and density inhomogeneities, thus providing a more discriminative input for subsequent adaptive configuration of mold parameters.

[0132] In some preferred embodiments, step S5 includes:

[0133] S51. Obtain the cumulative die-cutting length, average die-cutting pressure, and overload frequency of the mold based on historical mold usage statistics;

[0134] S52. Based on a preset wear model or wear mapping table, calculate the fatigue value of the mold according to the cumulative die-cutting length, average die-cutting pressure and overload frequency, and use it as the wear state characteristic of the mold.

[0135] Specifically, step S51 refers to performing statistical analysis based on the historical usage data of the mold to obtain key indicators reflecting the working load of the mold. This can be achieved by recording the length of each die-cutting operation, the instantaneous pressure data collected by the pressure sensor, and setting an overload threshold and recording the number of times the threshold is exceeded in the die-cutting machine control system. These statistical data provide a quantitative basis for subsequent assessment of the mold wear condition.

[0136] More specifically, step S52 refers to calculating the fatigue value of the die using statistically obtained indicators combined with a pre-established wear model or mapping table. Specifically, a mathematical model based on material fatigue theory can be used, which takes cumulative die-cutting length, average die-cutting pressure, and overload frequency as input and outputs a value representing the degree of fatigue damage; or a multi-dimensional lookup table can be used, which directly corresponds to a fatigue value based on different ranges of cumulative die-cutting length, average die-cutting pressure, and overload frequency. The calculated fatigue value is used as a characteristic of the die's wear state, which can comprehensively reflect the cumulative damage of the die under complex stress history.

[0137] More specifically, steps S51 and S52 work together to transform the mold's usage history into a fatigue value feature that better represents its actual wear and fatigue state. This fatigue value feature, along with the blade sharpness and material hardness features obtained from transient force distribution data, the edge quality score and dimensional deviation features obtained from material surface image data, and the material local structural inhomogeneity features obtained from thickness and density distribution data, are fed as input into a pre-trained adaptive parameter configuration model. Compared to relying solely on simple usage history estimation, this approach, by introducing the concept of fatigue value, more deeply reflects the actual damage state of the mold, improves the accuracy of judging mold wear factors, and thus enhances the effectiveness of adaptive parameter adjustment.

[0138] In some preferred embodiments, the method further includes steps performed after step S7:

[0139] S8. Based on the visual inspection system or offline quality inspection results, obtain the actual product die-cutting quality data, and input it as a feedback signal into the adaptive parameter configuration model to iteratively optimize the adaptive parameter configuration model.

[0140] Specifically, actual product die-cutting quality data can be collected in real time through a vision inspection system. For example, an industrial camera can be used to capture images of the product edges after die-cutting, and image processing algorithms can be used to analyze features such as edge smoothness, burrs, and chipping, quantifying them into quality scores or the number of defects. Alternatively, actual product die-cutting quality data can be obtained through offline quality inspection. For example, after die-cutting, the product can be sent to specialized inspection equipment (such as an optical microscope or profilometer) for precise measurement to obtain data such as edge size deviation and breakage rate.

[0141] More specifically, in this step, the acquired actual product die-cutting quality data is used as a feedback signal to input the adaptive parameter configuration model, enabling the model to perform iterative optimization using the feedback signal. The iterative optimization process adjusts the model's internal parameters, such as the weights and biases of the neural network.

[0142] More specifically, through repeated processes of die-cutting, quality inspection, feedback, and model optimization, the adaptive parameter configuration model can learn the complex relationship between input features and actual die-cutting quality, and gradually correct its parameter output strategy. Therefore, the die adjustment parameters output by the adaptive parameter configuration model can more accurately adapt to various changes in the actual die-cutting process, such as die wear and material batch differences, thereby improving the stability and consistency of die-cutting quality.

[0143] In some preferred embodiments, the method further includes steps performed between steps S5 and S6:

[0144] SA normalizes the characteristics of blade sharpness, material hardness, edge quality score, dimensional deviation, material local structural inhomogeneity, and mold wear state to form a feature vector for inputting into the adaptive parameter model.

[0145] Specifically, normalization is the process of scaling features with different dimensions and numerical ranges to a uniform range. For example, the min-max scaling method can be used to linearly transform the feature values ​​to a preset interval, such as 0 to 1. Alternatively, the Z-score standardization method can be used to convert the feature values ​​into a distribution with a mean of 0 and a standard deviation of 1.

[0146] More specifically, step SA normalizes the previously acquired feature data, mapping them to a unified numerical space and eliminating dimensional differences. The adaptive parameter configuration model learns and predicts based on the normalized feature vectors, outputting die adjustment parameters. Normalization ensures the model treats each feature fairly, improving the efficiency and accuracy of model learning. This allows for more precise output of adjustment parameters adapted to the current die and material conditions, thus improving die-cutting quality.

[0147] In some preferred embodiments, the die adjustment parameters include: die-cutting pressure, die-cutting speed, and die-cutting depth fine-tuning.

[0148] Specifically, the die-cutting pressure can be controlled by the hydraulic system or servo motor of the die-cutting machine's pressure head, and is monitored in real time and controlled in a closed loop by a force sensor array installed in the die-cutting area. The die-cutting speed can be controlled by adjusting the speed of the drive motor of the material conveying mechanism or the die motion mechanism. The die-cutting depth can be finely adjusted by a fine-tuning mechanism driven by a high-precision Z-axis servo motor or stepper motor.

[0149] More specifically, addressing the quality fluctuations caused by individual die variations, localized material inhomogeneity, and die wear during die-cutting, this solution utilizes an adaptive parameter configuration model to output specific adjustments for die-cutting pressure, speed, and depth. Die-cutting pressure directly affects the cutting depth and force distribution of the blade. Adjusting the pressure can compensate for the effects of localized blade dulling or material hard spots, ensuring complete cut-through. Die-cutting speed influences the material's deformation behavior and the rate of force applied to the blade during cutting. Optimizing the speed can reduce burrs and tearing. The depth adjustment provides fine-tuning capabilities to compensate for die wear or minor fluctuations in material thickness, preventing damage to lower layers or incomplete cuts. These parameters are dynamically calculated by the adaptive parameter configuration model based on input characteristics such as blade sharpness, material hard spots, edge quality, dimensional deviations, localized material structural inhomogeneity, and die wear. These parameters are then sent to the die-cutting machine's actuators, such as pressure control valves, motor drivers, and depth adjustment mechanisms, thereby achieving real-time, adaptive control of the die-cutting process and improving the consistency of product die-cutting quality.

[0150] In some preferred embodiments, in step S6, the adaptive parameter configuration model is a multilayer perceptron (MLP) or a convolutional neural network (CNN).

[0151] Specifically, a multilayer perceptron learns the nonlinear mapping relationship between input features and output parameters through multiple neuron layers. A convolutional neural network extracts patterns from input features through convolution and pooling operations, then maps the extracted patterns to output parameters through fully connected layers. Both models possess the ability to learn complex relationships from high-dimensional input data, handle feature data from different sources and with different properties, and predict the required fine-tuning amounts of die-cutting pressure, speed, and depth based on these features. Therefore, the adaptive parameter configuration model can dynamically adjust die parameters according to actual conditions, overcoming the limitations of traditional fixed-parameter or simple models that cannot accurately capture the interactions between complex factors, improving the accuracy and adaptability of parameter adjustment, and thus stabilizing die-cutting quality.

[0152] In some preferred embodiments, step S1 includes:

[0153] S11. Acquire transient pressure data during trial cutting of the die-cutting area edge using a force sensor array;

[0154] S12. Use a vision camera to obtain images of the material surface after die-cutting at the test-cut edge of the die-cutting area;

[0155] S13. Use an ultrasonic sensor to scan the die-cutting area to obtain ultrasonic reflection signals. Analyze the amplitude and delay of the reflection signals to calculate the thickness distribution data and density distribution data.

[0156] S14. Retrieve the mold usage history corresponding to the identifier of the currently used mold from the mold management database.

[0157] Specifically, transient pressure data is acquired through a force sensor array positioned below the die-cutting area. This array consists of multiple independent force sensors, which can capture the pressure values ​​at different locations in the die-cutting area over time at the moment the die-cutting blade contacts and cuts the material, thereby obtaining information on the force distribution during the die-cutting process.

[0158] More specifically, the material surface image data is acquired by a vision camera positioned downstream of the die-cutting station. After the trial cut, the camera takes pictures of the die-cut edge area, recording the visual characteristics of the material surface, including the shape of the cut edge and any potential defects.

[0159] More specifically, thickness and density distribution data are obtained by scanning the material surface using an ultrasonic sensor. The sensor emits ultrasonic pulses into the material, receives the reflected echoes, and by measuring the intensity and propagation time of the echoes, the thickness and density of the material at the scanning point can be calculated.

[0160] More specifically, historical data on mold usage is obtained by querying a pre-established mold management database. The database stores a unique identifier for each mold and its corresponding usage records, such as the cumulative number of die cuts and the type of material processed.

[0161] More specifically, these multi-source data acquisition methods work together to comprehensively capture information on key factors affecting die-cutting quality, providing the necessary data input for subsequent adaptive parameter adjustments based on this data. This enables the system to cope with dynamic changes in molds and materials and improve product quality stability.

[0162] Secondly, please refer to Figure 3 and Figure 4 Some embodiments of this application also provide a mold parameter adjustment system for use in a die-cutting machine, the system comprising:

[0163] The acquisition module 201 is used to acquire transient force distribution data, material surface image data, thickness distribution data, density distribution data, and mold usage history of the die-cutting area of ​​the workpiece.

[0164] The first analysis module 202 is used to obtain the blade sharpness characteristics and material hard point characteristics based on transient force distribution data;

[0165] The second analysis module 203 is used to obtain the edge quality score and dimensional deviation characteristics of the cut edge based on the material surface image data.

[0166] The third analysis module 204 is used to obtain the local structural inhomogeneity characteristics of the material based on the thickness distribution data and density distribution data;

[0167] The fourth analysis module 205 is used to estimate the wear characteristics of the mold based on the mold's usage history;

[0168] The parameter generation module 206 is used to input the blade sharpness features, material hard point features, edge quality score, size deviation features, material local structural non-uniformity features, and mold wear state features into a pre-trained adaptive parameter configuration model to output mold adjustment parameters.

[0169] The actuator control module 207 is used to control the actuator of the die-cutting machine to perform die cutting based on the mold adjustment parameters.

[0170] The die parameter adjustment system of this application comprehensively perceives various factors affecting die-cutting quality by collecting transient stress, material images, local physical properties of the material, and historical die data during the die-cutting process. It integrates multi-source data and generates die adjustment parameters based on an adaptive parameter configuration model to achieve adaptive adjustment control of the die-cutting machine. This means that die-cutting parameters are no longer fixed empirical values, but are dynamically adjusted according to the current die state, material characteristics, and preliminary cutting results. This adaptive adjustment mechanism can effectively cope with quality fluctuations caused by individual die differences, wear, and the superposition of local material inhomogeneities, improving the stability of die-cutting processing and the consistency of product quality. Through perception, analysis, and intelligent decision-making, the system of this application achieves closed-loop control and optimization of the die-cutting process.

[0171] In some preferred embodiments, the system further includes:

[0172] The model optimization module 208 is used to obtain actual product die-cutting quality data based on the visual inspection system or offline quality inspection results, and input it as a feedback signal into the adaptive parameter configuration model to iteratively optimize the adaptive parameter configuration model.

[0173] In some preferred embodiments, the system further includes:

[0174] The normalization module 209 is used to normalize the characteristics of blade sharpness, material hard points, edge quality score, dimensional deviation, material local structural inhomogeneity, and mold wear state to form a feature vector for input into the adaptive parameter model.

[0175] Furthermore, the units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0176] Furthermore, the functional modules in the various embodiments of this application can be integrated together to form an independent part, or each module can exist independently, or two or more modules can be integrated to form an independent part.

[0177] In this document, relational terms such as first and second are used only to distinguish one entity or operation from another entity or operation, without necessarily requiring or implying any such actual relationship or order between these entities or operations.

[0178] The above description is merely an embodiment of this application and is not intended to limit the scope of protection of this application. Various modifications and variations can be made to this application by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of this application should be included within the scope of protection of this application.

Claims

1. A method for adjusting mold parameters, applied in a die-cutting machine, characterized in that, The method includes the following steps: S1. Obtain transient force distribution data, material surface image data, thickness distribution data, density distribution data, and mold usage history of the die-cutting area of ​​the workpiece; S2. Obtain the blade sharpness characteristics and material hardness characteristics based on the transient force distribution data; S3. Obtain the edge quality score and dimensional deviation characteristics of the cut edge based on the material surface image data; S4. Obtain the characteristics of local structural inhomogeneity of the material based on thickness distribution data and density distribution data; S5. Estimate the wear characteristics of the mold based on the mold's usage history; S6. Input the blade sharpness feature, material hard point feature, edge quality score, size deviation feature, material local structural non-uniformity feature, and mold wear state feature into the pre-trained adaptive parameter configuration model to output mold adjustment parameters; S7. Based on the mold adjustment parameters, control the actuator of the die-cutting machine to perform die cutting; Step S2 includes: S21. Perform force waveform analysis, peak value analysis, and distribution non-uniformity analysis on the transient force distribution data in sequence to extract the blade sharpness characteristics and material hard point characteristics; Step S21 includes: S211. The transient force distribution data is analyzed by fast Fourier transform to extract the main frequency component and harmonic components and construct a spectrum diagram. S212. Calculate the spectral energy concentration based on the spectrum diagram, and convert the blade sharpness characteristics based on the spectral energy concentration. S213. Use a peak detection algorithm to identify peak points in transient force distribution data, and record the force value and location coordinates corresponding to each peak point; S214. Based on the force value and position coordinates of the peak point, calculate the force distribution uniformity coefficient of the die-cut area, and determine the hard point characteristics of the material based on the force distribution uniformity coefficient. Step S214 includes: S2141. Calculate the deviation between the force value at each peak point and the average force value in the die-cut area; S2142. Calculate the sum of squares of the deviation values ​​based on each deviation value, and divide the sum of squares of the deviation values ​​by the number of peak points to obtain the force variance value. S2143. Divide the average force by the force variance to obtain the force distribution uniformity coefficient. Determine the material hard point distribution uniformity based on the force distribution uniformity coefficient and the preset mapping table.

2. The mold parameter adjustment method according to claim 1, characterized in that, Step S3 includes: S31. The material surface image data is sequentially subjected to edge detection, defect identification, and size measurement to obtain the edge quality score and size deviation characteristics of the cut edge.

3. The mold parameter adjustment method according to claim 1, characterized in that, Step S4 includes: S41. Perform local change rate detection and abrupt change point detection on the thickness distribution data and density distribution data in sequence to obtain the local structural inhomogeneity characteristics of the material.

4. The mold parameter adjustment method according to claim 3, characterized in that, Step S41 includes: S411. Perform Gaussian filtering on the thickness distribution data and density distribution data respectively to obtain smoothed thickness data and density distribution data; S412. Calculate the first difference between the smoothed thickness data and density distribution data to obtain the thickness change rate and density change rate. S413. Count the number of local mutation points within a unit area, wherein the local mutation point is a location where the thickness change rate is greater than a preset thickness change rate threshold and / or the density change rate is greater than a preset density change rate threshold. S414. Calculate the proportion of the local abrupt change points and convert the proportion into the characteristics of local structural inhomogeneity of the material.

5. The mold parameter adjustment method according to claim 4, characterized in that, Local abrupt change points include thickness abrupt change points and density abrupt change points. Thickness abrupt change points are locations where the thickness change rate is greater than a preset thickness change rate threshold, and density abrupt change points are locations where the density change rate is greater than a preset density change rate threshold. Step S414 includes: The proportions of thickness abrupt change points and density abrupt change points are calculated separately, and the proportions of thickness abrupt change points and density abrupt change points are weighted based on preset weighting coefficients to obtain the local structural inhomogeneity characteristics of the material.

6. The mold parameter adjustment method according to claim 1, characterized in that, Step S5 includes: S51. Obtain the cumulative die-cutting length, average die-cutting pressure, and overload frequency of the mold based on the historical statistics of mold usage; S52. Based on a preset wear model or wear mapping table, calculate the fatigue value of the mold according to the cumulative die-cutting length, average die-cutting pressure and overload frequency, as a characteristic of the mold wear state.

7. The mold parameter adjustment method according to claim 1, characterized in that, The method further includes steps performed after step S7: S8. Based on the visual inspection system or offline quality inspection results, obtain the actual product die-cutting quality data, and input it as a feedback signal into the adaptive parameter configuration model to iteratively optimize the adaptive parameter configuration model.

8. A mold parameter adjustment system, used in a die-cutting machine, characterized in that, The system includes: The acquisition module is used to acquire transient force distribution data, material surface image data, thickness distribution data, density distribution data, and mold usage history of the die-cutting area of ​​the workpiece. The first analysis module is used to obtain the blade sharpness characteristics and material hard point characteristics based on the transient force distribution data. The second analysis module is used to obtain the edge quality score and dimensional deviation characteristics of the cut edge based on the material surface image data. The third analysis module is used to obtain the local structural inhomogeneity characteristics of the material based on the thickness distribution data and density distribution data; The fourth analysis module is used to estimate the wear characteristics of the mold based on the mold's usage history; The parameter generation module is used to input the blade sharpness feature, material hard point feature, edge quality score, size deviation feature, material local structural non-uniformity feature, and mold wear state feature into a pre-trained adaptive parameter configuration model to output mold adjustment parameters. An actuator control module is used to control the actuator of the die-cutting machine to perform die-cutting based on the mold adjustment parameters; The steps of obtaining the blade sharpness characteristics and material hardness characteristics based on the transient force distribution data include: S21. Perform force waveform analysis, peak value analysis, and distribution non-uniformity analysis on the transient force distribution data in sequence to extract the blade sharpness characteristics and material hard point characteristics; Step S21 includes: S211. The transient force distribution data is analyzed by fast Fourier transform to extract the main frequency component and harmonic components and construct a spectrum diagram. S212. Calculate the spectral energy concentration based on the spectrum diagram, and convert the blade sharpness characteristics based on the spectral energy concentration. S213. Use a peak detection algorithm to identify peak points in transient force distribution data, and record the force value and location coordinates corresponding to each peak point; S214. Based on the force value and position coordinates of the peak point, calculate the force distribution uniformity coefficient of the die-cut area, and determine the hard point characteristics of the material based on the force distribution uniformity coefficient. Step S214 includes: S2141. Calculate the deviation between the force value at each peak point and the average force value in the die-cut area; S2142. Calculate the sum of squares of the deviation values ​​based on each deviation value, and divide the sum of squares of the deviation values ​​by the number of peak points to obtain the force variance value. S2143. Divide the average force by the force variance to obtain the force distribution uniformity coefficient. Determine the material hard point distribution uniformity based on the force distribution uniformity coefficient and the preset mapping table.

Images