A method and system for controlling the production of ceramic articles

Abstract

This invention discloses a method and system for controlling the production of ceramic products, relating to the field of industrial production control technology. A ceramic product production control system includes a data acquisition and classification module, a fluctuation impact assessment module, a feedforward parameter compensation module, a feedback parameter correction module, and a complexity assessment and decision-making module. This invention constructs a state deviation transmission model based on past production data of ceramic products. Based on the state deviation transmission model, it calculates three risk coefficients for ceramic raw materials and determines the priority of compensation strategies for these raw materials, thus obtaining an initial compensation strategy. This quantifies the impact of raw material fluctuations on the quality of ceramic products and identifies primary and secondary risks to avoid blindly adjusting all process parameters. This ensures the compensation effect of process parameter adjustments on raw material fluctuations and significantly improves the production quality of ceramic products, thereby increasing the yield of this ceramic product production control method and system.

Description

Technical Field

[0001] This invention relates to the field of industrial production control technology, specifically to a method and system for controlling the production of ceramic products. Background Technology

[0002] Ceramic products are a type of inorganic non-metallic material made primarily from natural minerals such as clay, quartz, and feldspar. These materials are crushed, mixed, and plasticized, then shaped through processes such as slip casting, pressing, or extrusion. After drying to remove moisture, they are finally sintered in a high-temperature kiln. The production process encompasses key steps including raw material pretreatment, slip preparation, precision molding, gradient drying, high-temperature firing, and surface glazing and decoration. The final products are characterized by high hardness, high-temperature resistance, corrosion resistance, and good insulation, and are widely used in daily utensils, building materials, industrial components, and artistic decoration.

[0003] Existing methods and systems for controlling the production of ceramic products typically apply corresponding processing parameters based on the raw material formula of the ceramic products. However, even if the raw material formula is consistent, the composition of raw materials in different batches or even the same batch will fluctuate and differ. If the processing parameters are not adjusted accordingly to offset the impact of raw material fluctuations in advance, it will lead to problems such as large dimensional deviations, high cracking and deformation rates, obvious batch color differences, and large fluctuations in yield of ceramic products, resulting in waste of raw materials and a significant increase in energy consumption costs.

[0004] Based on the above, this invention proposes a method and system for controlling the production of ceramic products with high yield. Summary of the Invention

[0005] To overcome the shortcomings of existing ceramic product production control methods and systems, which typically apply corresponding processing parameters based on the raw material formula of ceramic products during production, even if the raw material formula is consistent, the composition of raw materials in different batches or even the same batch will fluctuate and differ. If the processing parameters are not adjusted accordingly to offset the impact of raw material fluctuations in advance, it will lead to problems such as large dimensional deviations, high cracking and deformation rates, obvious batch color differences, and large fluctuations in yield of ceramic products, resulting in raw material waste and significantly increased energy consumption costs. This invention proposes a ceramic product production control method and system with high yield.

[0006] A method for controlling the production of ceramic products includes the following steps: XRF composition scanning and laser particle size detection were performed on the ceramic raw materials to obtain the state parameter data of the ceramic raw materials. Outlier processing, Z-score standardization and data vectorization were performed on the state parameter data of the ceramic raw materials to obtain preprocessed state parameter data. Principal component analysis was used to extract three principal components from the preprocessed state parameter data and to determine the state fluctuation type of the ceramic raw materials based on these components, thus obtaining the state fluctuation determination result of the ceramic raw materials. A state deviation transmission model is constructed based on the past production data of ceramic products. The three risk coefficients of ceramic raw materials are calculated according to the state deviation transmission model, and the priority of the compensation strategy for ceramic raw materials is determined accordingly to obtain the initial compensation strategy for ceramic raw materials. Based on the experimental production data of ceramic products, a response sensitivity matrix was constructed through multiple regression analysis. Based on the initial compensation strategy and the response sensitivity matrix, the front-end parameter adjustment scheme of ceramic raw materials was obtained through calculation. A digital twin model of ceramic product firing is constructed and the initial firing curve of the ceramic product is preset. Based on the preprocessed state parameter data and the front-end parameter adjustment scheme, the firing of the ceramic product is simulated using the digital twin model to obtain the simulated firing quality data of the ceramic product. Based on the deviation between the simulated firing quality data and the target firing quality data, the optimal firing curve of the ceramic product is obtained through particle swarm optimization algorithm. The complexity of the process adjustment of ceramic products is evaluated based on the front-end parameter adjustment scheme and the optimal firing curve to obtain the process adjustment complexity index of ceramic products. Based on the process adjustment complexity index, the production scheme of ceramic products is decided to obtain the final production scheme of ceramic products.

[0007] As a preferred aspect of the invention, the specific steps for extracting three principal components from the preprocessed state parameter data using principal component analysis and determining the state fluctuation type of the ceramic raw material based on these components to obtain the state fluctuation determination result of the ceramic raw material are as follows: Historical state parameter data of ceramic raw materials are obtained and the historical covariance matrix is ​​calculated. The historical covariance matrix is ​​then decomposed into eigenvalues ​​to obtain multiple eigenvalues ​​and their corresponding eigenvectors. The magnitude of the eigenvalue reflects the contribution of the data variance in the direction of the eigenvector. The three feature vectors with the highest contribution are selected as principal components and defined as the raw material composition dominant factor, raw material particle size dominant factor and raw material activity dominant factor, respectively. These three feature vectors are used to construct a projection matrix, and the preprocessed state parameter data is multiplied with the projection matrix to obtain the current three-dimensional principal component score of the ceramic raw material. The current three-dimensional principal component score is subtracted from the preset standard three-dimensional principal component score to obtain the current three-dimensional principal component fluctuation score of the ceramic raw material. The three principal components are classified into levels according to the principal component fluctuation score of each dimension. Based on the level classification result, the state fluctuation type of the ceramic raw material is determined as a single type of fluctuation or a composite type of fluctuation, and the state fluctuation determination result of the ceramic raw material is obtained.

[0008] As a preferred aspect of the invention, the specific steps of constructing a state deviation transmission model based on past production data of ceramic products, calculating three risk coefficients of ceramic raw materials according to the state deviation transmission model, and determining the priority of compensation strategies for ceramic raw materials to obtain the initial compensation strategy for ceramic raw materials are as follows: Obtain historical production data for ceramic products, including historical state parameter data of ceramic raw materials, historical three-dimensional principal component fluctuation scores, and historical quality deviation data of ceramic products. The historical quality deviation data includes shrinkage deviation, deformation deviation, and whiteness deviation of ceramic products, and the deviation refers to the difference between the measured value and the preset standard value. A multiple linear regression model was used as the initial model, and the relationship between the historical three-dimensional principal component fluctuation score of ceramic raw materials and various deviations in the historical quality deviation data was constructed. The coefficients of the relationship were solved by the least squares method to obtain the state deviation transmission model. The current three-dimensional principal component fluctuation score of the ceramic raw material is input into the state deviation transmission model to obtain the current predicted value of each deviation. The ratio of the current predicted value of each deviation to the preset standard value is calculated and used as the three risk coefficients of the ceramic raw material, including shrinkage risk coefficient, deformation risk coefficient and color difference risk coefficient. The priority of each risk is determined based on the three risk coefficients of the ceramic raw materials and their corresponding preset risk thresholds. Based on this, the parameter compensation mode is set as either single-parameter precise compensation or multi-parameter collaborative compensation. The initial compensation strategy for the ceramic raw materials is determined based on the set parameter compensation mode and the state fluctuation judgment results of the ceramic raw materials.

[0009] As a preferred aspect of the invention, the specific steps of constructing a response sensitivity matrix based on experimental production data of ceramic products and through multiple regression analysis, and calculating the front-end parameter adjustment scheme of ceramic raw materials based on the initial compensation strategy and the response sensitivity matrix are as follows: The experimental production data of ceramic products are obtained, including experimental process parameter data of ceramic products and three-dimensional principal component fluctuation score change data of ceramic raw materials before and after process treatment. Multivariate regression analysis is performed on the experimental production data and a response sensitivity matrix is ​​constructed. Each column of the response sensitivity matrix represents a process parameter, each row represents a principal component in the three-dimensional principal component, and the matrix elements are the change in principal component score caused by a unit fluctuation of the process parameter. Based on the initial compensation strategy of ceramic raw materials, the type of process parameter to be adjusted is determined. Based on the current three-dimensional principal component fluctuation score and response sensitivity matrix of ceramic raw materials, the adjustment amount of the process parameter to be adjusted is obtained by calculation, and the front-end parameter adjustment scheme of ceramic raw materials is obtained.

[0010] As a preferred aspect of the invention, the specific steps of obtaining the optimal firing curve of the ceramic product based on the deviation between the pre-firing quality data and the target firing quality data using a particle swarm optimization algorithm are as follows: When the deviation between the pre-firing quality data and the target firing quality data is greater than the preset threshold, the particle swarm algorithm is activated and multiple particles are randomly generated. Each particle represents a candidate firing curve and is represented as a vector in four-dimensional space. The meanings of each dimension of this vector are heating rate, holding temperature, holding time and cooling rate, respectively, and the value range of each dimension is based on process specification constraints. Each particle is evaluated and a digital twin is used for rapid simulation. The objective function value is calculated based on the simulation results. The objective function is set based on the deformation deviation and color difference deviation after firing. The particles are iteratively optimized and updated to minimize the objective function until the objective function value is less than the preset threshold or the maximum number of iterations is reached, thus obtaining the optimal firing curve of the ceramic product.

[0011] As a preferred aspect of the invention, the specific steps of evaluating the process adjustment complexity of ceramic products based on the front-end parameter adjustment scheme and the optimal firing curve to obtain a process adjustment complexity index of ceramic products, and making a decision on the production scheme of ceramic products based on the process adjustment complexity index to obtain the final production scheme of ceramic products are as follows: The process adjustment complexity of ceramic products is evaluated based on the deviation between the front-end parameter adjustment scheme and the optimal firing curve and the preset standard front-end parameter scheme and standard firing curve, and the process adjustment complexity index of ceramic products is obtained. When the process adjustment complexity index is less than the preset low threshold, the final production plan adopts both the front-end parameter adjustment plan and the optimal firing curve of the ceramic product. When the process adjustment complexity index is greater than the preset high threshold, the production plan of the ceramic product is not adjusted. When the process adjustment complexity index is between the preset high threshold and the preset low threshold, only the front-end parameter adjustment plan is adopted without adjusting the firing curve.

[0012] A ceramic product production control system includes: The data acquisition and classification module includes a data acquisition and processing unit and a raw material fluctuation classification unit. The data acquisition and processing unit is used to perform XRF composition scanning and laser particle size detection on ceramic raw materials to obtain state parameter data of ceramic raw materials. The state parameter data of ceramic raw materials is processed for outliers, Z-score standardization and data vectorization to obtain preprocessed state parameter data. The raw material fluctuation classification unit is used to extract three principal components from the preprocessed state parameter data using principal component analysis and to determine the state fluctuation type of ceramic raw materials based on these components, thus obtaining the state fluctuation determination result of ceramic raw materials. The fluctuation impact assessment module is used to construct a state deviation transmission model based on the past production data of ceramic products. Based on the state deviation transmission model, it calculates three risk coefficients of ceramic raw materials and determines the priority of compensation strategies for ceramic raw materials, thereby obtaining the initial compensation strategy for ceramic raw materials. The feedforward parameter compensation module is used to construct a response sensitivity matrix based on experimental production data of ceramic products and multiple regression analysis. Based on the initial compensation strategy and the response sensitivity matrix, the front-end parameter adjustment scheme of ceramic raw materials is obtained through calculation. The feedback parameter correction module is used to construct a digital twin model of ceramic product firing and preset the initial firing curve of ceramic product. Based on the preprocessed state parameter data and front-end parameter adjustment scheme, the digital twin model is used to perform a pre-firing simulation of ceramic product to obtain the pre-firing quality data of ceramic product. Based on the deviation between the pre-firing quality data and the target firing quality data, the optimal firing curve of ceramic product is obtained through particle swarm optimization algorithm. The complexity assessment and decision-making module includes an adjustment complexity assessment unit and a production scheme decision-making unit. The adjustment complexity assessment unit is used to assess the process adjustment complexity of ceramic products based on the front-end parameter adjustment scheme and the optimal firing curve to obtain the process adjustment complexity index of ceramic products. The production scheme decision-making unit is used to make a decision on the production scheme of ceramic products based on the process adjustment complexity index to obtain the final production scheme of ceramic products.

[0013] The present invention has the following advantages: 1. This invention constructs a state deviation transmission model based on past production data of ceramic products. According to the state deviation transmission model, it calculates three risk coefficients of ceramic raw materials and determines the priority of compensation strategies for ceramic raw materials, thereby obtaining an initial compensation strategy for ceramic raw materials. This can accurately quantify the specific impact of raw material fluctuations on the final quality of ceramic products and identify primary and secondary risks to avoid blindly adjusting all process parameters. This effectively ensures the compensation effect of process parameter adjustments on raw material fluctuations and significantly improves the production quality of ceramic products, thereby improving the production yield of this ceramic product production control method and system.

[0014] 2. This invention constructs a response sensitivity matrix based on experimental production data of ceramic products and multiple regression analysis. Based on the initial compensation strategy and the response sensitivity matrix, it calculates the adjustment scheme for the front-end parameters of ceramic raw materials. This not only directly maps the raw material fluctuation score to the adjustment amount of various front-end process parameters, but also proactively eliminates most of the impact of raw material fluctuations at a low cost before defects occur, preventing problems from being transmitted to the firing stage and causing irreversible losses. Furthermore, it ensures that only key parameters are adjusted through priority screening, thereby eliminating new defects induced by overcompensation. This improves the production yield and practicality of this ceramic product production control method and system.

[0015] 3. This invention evaluates the process adjustment complexity of ceramic products based on the front-end parameter adjustment scheme and the optimal firing curve, obtaining a process adjustment complexity index for ceramic products. Based on this index, the production scheme for ceramic products is decided. This not only avoids production interruptions, secondary quality defects, and efficiency losses caused by excessively complex parameter adjustments, but also achieves a dynamic balance between quality improvement needs and production stability, prioritizing the execution of simple and efficient compensation paths. This improves the overall operating efficiency of the ceramic product production line and the actual feasibility of the compensation strategy, thus enhancing the practicality of this ceramic product production control method and system. Attached Figure Description

[0016] Figure 1 This is a schematic flowchart of a ceramic product production control method used in an embodiment of the present invention.

[0017] Figure 2 This is a schematic diagram of the structure of a ceramic product production control system used in an embodiment of the present invention. Detailed Implementation

[0018] To enable those skilled in the art to better understand the technical solutions of this invention, the technical solutions of this invention will be clearly and completely described below with reference to the accompanying drawings in the embodiments of this invention.

[0019] Example 1: A method for controlling the production of ceramic products, such as... Figure 1 As shown, it includes the following steps: XRF composition scanning and laser particle size detection were performed on the ceramic raw materials to obtain the state parameter data of the ceramic raw materials. Outlier processing, Z-score standardization and data vectorization were performed on the state parameter data of the ceramic raw materials to obtain preprocessed state parameter data. Principal component analysis was used to extract three principal components from the preprocessed state parameter data and to determine the state fluctuation type of the ceramic raw materials based on these components, thus obtaining the state fluctuation determination result of the ceramic raw materials. A state deviation transmission model is constructed based on the past production data of ceramic products. The three risk coefficients of ceramic raw materials are calculated according to the state deviation transmission model, and the priority of the compensation strategy for ceramic raw materials is determined accordingly to obtain the initial compensation strategy for ceramic raw materials. Based on the experimental production data of ceramic products, a response sensitivity matrix was constructed through multiple regression analysis. Based on the initial compensation strategy and the response sensitivity matrix, the front-end parameter adjustment scheme of ceramic raw materials was obtained through calculation. A digital twin model of ceramic product firing is constructed and the initial firing curve of the ceramic product is preset. Based on the preprocessed state parameter data and the front-end parameter adjustment scheme, the firing of the ceramic product is simulated using the digital twin model to obtain the simulated firing quality data of the ceramic product. Based on the deviation between the simulated firing quality data and the target firing quality data, the optimal firing curve of the ceramic product is obtained through particle swarm optimization algorithm. The complexity of the process adjustment of ceramic products is evaluated based on the front-end parameter adjustment scheme and the optimal firing curve to obtain the process adjustment complexity index of ceramic products. Based on the process adjustment complexity index, the production scheme of ceramic products is decided to obtain the final production scheme of ceramic products.

[0020] It should be noted that when solving the formulas for ceramic products and their glazes, traditional Seger glaze calculations can be transformed into matrix operations and specifically calculated using Matlab. The specific implementation is as follows: First, the chemical composition of the raw materials (Al2O3, SiO2, K2O, etc.) is detected and converted into molar mass. Then, a system of multivariate nonhomogeneous linear equations is established, constructing a coefficient matrix A (molar mass of raw material components), an objective vector b (formula design parameters), and an unknown vector X (parts of raw materials). Matlab is then used to solve for the mass fractions and percentages of each raw material. This program supports flexible parameter modification to adapt to different raw materials and formula designs. Compared to traditional manual calculations, this method significantly improves computational efficiency and accuracy, ensures product quality stability, and achieves a combination of computer science and materials science.

[0021] It should be noted that the above digital twin model adopts a hybrid modeling strategy: the mechanistic layer constructs the temperature field based on the Fourier heat conduction equation, and its governing equation is as follows:

[0022] in For temperature, For time, The thermal diffusivity (obtained in this embodiment by actual measurement using a laser thermal conductivity meter) is the coefficient of thermal diffusivity. ), For the Laplace operator, As an internal heat source, The stress field is modeled using a thermo-elastoplastic constitutive model to represent the rate of change of the internal heat source over time, with an elastic modulus E = 65 GPa and a Poisson's ratio ν = 0.22. The data-driven layer employs a backpropagation neural network to fit the glaze flowability and crystal phase transformation dynamics, which are difficult to model mechanistically. The inputs are the holding temperature and time, and the output is the liquid phase quantity. The training sample consists of 120 sets of DSC experimental data. Accuracy verification shows that the average relative error between the predicted shrinkage rate and the measured value of the twin is <3%, and the predicted deformation error is <0.15 mm, meeting the requirements for online pre-simulation.

[0023] The specific steps for extracting three principal components from the preprocessed state parameter data using principal component analysis and determining the state fluctuation type of the ceramic raw material based on these components to obtain the state fluctuation determination result of the ceramic raw material are as follows: Historical state parameter data of ceramic raw materials are obtained and the historical covariance matrix is ​​calculated. The historical covariance matrix is ​​then decomposed into eigenvalues ​​to obtain multiple eigenvalues ​​and their corresponding eigenvectors. The magnitude of the eigenvalue reflects the contribution of the data variance in the direction of the eigenvector. The three feature vectors with the highest contribution are selected as principal components and defined as the raw material composition dominant factor, raw material particle size dominant factor and raw material activity dominant factor, respectively. These three feature vectors are used to construct a projection matrix, and the preprocessed state parameter data is multiplied with the projection matrix to obtain the current three-dimensional principal component score of the ceramic raw material. The current three-dimensional principal component score is subtracted from the preset standard three-dimensional principal component score to obtain the current three-dimensional principal component fluctuation score of the ceramic raw material. The three principal components are classified into levels according to the principal component fluctuation score of each dimension. Based on the level classification result, the state fluctuation type of the ceramic raw material is determined as a single type of fluctuation or a composite type of fluctuation, and the state fluctuation determination result of the ceramic raw material is obtained.

[0024] It should be noted that the historical covariance matrix is ​​constructed based on nearly 2000 batches of data from this production line over the past few years, and the original state parameters include: XRF composition data (12 dimensions): SiO2 (65-72%), Al2O3 (18-24%), Fe2O3 (0.1-0.8%), CaO (0.2-1.0%), MgO (0.1-0.5%), K2O (1.5-3.5%), Na2O (1.0-2.5%), etc. Particle size distribution data (30 dimensions): D10 (2-8μm), D50 (8-18μm), D90 (20-35μm), uniformity coefficient (D60 / D10) is 5-8, specific surface area is 0.8-1.5m² / g; Process status data (38 dimensions): moisture content, plasticity index, fluidity, etc.

[0025] It should be noted that, in this embodiment, after eigenvalue decomposition, the cumulative contribution rate of the first three principal components reaches 94.7%, of which: PC1 (the dominant factor of raw material composition) contributed 68.3%, with eigenvector loadings of Al2O3 (0.85), SiO2 (-0.72), and Fe2O3 (-0.61), reflecting deviations in the chemical composition of the raw materials. PC2 (the dominant factor for raw material particle size) contributed 18.9%, with eigenvector loadings of D50 (0.78) and specific surface area (0.69), reflecting fineness and activity. PC3 (the dominant active factor of raw materials) contributed 7.5%, with eigenvector loadings of plasticity index (0.81) and loss on ignition (0.58).

[0026] The specific steps for constructing a state deviation transmission model based on past production data of ceramic products, calculating three risk coefficients of ceramic raw materials based on the state deviation transmission model, and determining the priority of compensation strategies for ceramic raw materials to obtain the initial compensation strategy for ceramic raw materials are as follows: Obtain historical production data for ceramic products, including historical state parameter data of ceramic raw materials, historical three-dimensional principal component fluctuation scores, and historical quality deviation data of ceramic products. The historical quality deviation data includes shrinkage deviation, deformation deviation, and whiteness deviation of ceramic products, and the deviation refers to the difference between the measured value and the preset standard value. A multiple linear regression model was used as the initial model, and the relationship between the historical three-dimensional principal component fluctuation score of ceramic raw materials and various deviations in the historical quality deviation data was constructed. The coefficients of the relationship were solved by the least squares method to obtain the state deviation transmission model. The current three-dimensional principal component fluctuation score of the ceramic raw material is input into the state deviation transmission model to obtain the current predicted value of each deviation. The ratio of the current predicted value of each deviation to the preset standard value is calculated and used as the three risk coefficients of the ceramic raw material, including shrinkage risk coefficient, deformation risk coefficient and color difference risk coefficient. The priority of each risk is determined based on the three risk coefficients of the ceramic raw materials and their corresponding preset risk thresholds. Based on this, the parameter compensation mode is set as either single-parameter precise compensation or multi-parameter collaborative compensation. The initial compensation strategy for the ceramic raw materials is determined based on the set parameter compensation mode and the state fluctuation judgment results of the ceramic raw materials.

[0027] It should be noted that the quality deviation data was obtained through the following method: Shrinkage deviation: According to GB / T3810.2 standard, take 10 100mm×100mm samples, measure the diagonal length before and after firing with digital vernier calipers (accuracy 0.01mm), calculate the linear shrinkage rate, deviation = measured average value - standard value (the standard value in this embodiment is 8.0±0.3%). Deformation deviation: A laser flatness tester (accuracy 0.05mm) is used to measure the height difference between the center point and the four corners of the product. The maximum value is taken as the deformation. Deviation = measured value - standard value (superior grade standard ≤ 2.0mm). Whiteness deviation: The average value of 5 points on the surface of the product is measured using a whiteness meter (D65 light source, wavelength 457nm). Deviation = measured whiteness - standard whiteness (standard whiteness ≥ 70 degrees in this embodiment).

[0028] For example, taking nearly 1200 batches of data from the past two years as an example, a typical sample is as follows: Batch A: Shrinkage deviation +0.28%, deformation 2.3mm, whiteness 68.5 degrees; Batch B: Shrinkage deviation -0.18%, deformation 1.9mm, whiteness 71.2 degrees; Batch C: Shrinkage deviation +0.45%, deformation 2.8mm, whiteness 66.1 degrees.

[0029] It should be noted that the specific form of the multiple linear regression is: shrinkage deviation / deformation deviation / whiteness deviation = a*composition fluctuation score + b*particle size fluctuation score + c*activity fluctuation score. The model is constructed using ridge regression with cross-validation, and the regularization coefficient λ=0.1 to prevent overfitting. The model has R²=0.89 and RMSE=0.12% on the validation set (300 batches), which meets the requirements for online prediction accuracy. The past production data used to train the transmission model must meet the following conditions: raw material batch span ≥30 different mining sites; sample size of each batch ≥5 independent tests; process parameter fluctuation range <±3% (ensuring that the raw material has a dominant influence).

[0030] It should be noted that the risk priority classification above adopts a 5-level classification method, in which low (0-1.0), low-medium (1.0-2.0), medium (2.0-3.5), medium-high (3.5-5.0), and high (>5.0) are selected first for single-parameter compensation of the item with the highest risk coefficient. The multi-parameter collaborative mode corresponding to the two or three items with the highest risk coefficient is only triggered when the ratio of other risk coefficients to the highest risk coefficient is >0.8.

[0031] It should be noted that when the shrinkage risk coefficient is the highest, the pressing pressure is mainly adjusted; when the deformation risk coefficient is the highest, the ball mill speed is mainly adjusted; and when the color difference risk coefficient is the highest, the raw material iron removal rate is mainly adjusted. When performing precise compensation for a single parameter, only the main process parameter corresponding to the highest risk coefficient is adjusted. When performing synergistic compensation for multiple parameters, a time-sequential step adjustment method is required to adjust multiple parameters. For example, when both the shrinkage risk coefficient and the deformation risk coefficient are high, and the shrinkage risk coefficient is the highest, the pressing pressure should first be adjusted to 70% of the target value and the quality improvement trend should be observed. When the quality improvement trend meets expectations, the pressing pressure should be adjusted to 100% of the target value, and the ball mill speed should be adjusted to 50% of the target value. If other risk coefficients do not worsen, the ball mill speed should also be adjusted to 100% of the target value.

[0032] The above steps construct a state deviation transmission model based on past production data of ceramic products. The model is used to calculate three risk coefficients of ceramic raw materials and determine the priority of compensation strategies for ceramic raw materials. This yields an initial compensation strategy for ceramic raw materials, which can accurately quantify the specific impact of raw material fluctuations on the final quality of ceramic products. It also identifies primary and secondary risks to avoid blindly adjusting all process parameters, thereby effectively ensuring the compensation effect of process parameter adjustments on raw material fluctuations and significantly improving the production quality of ceramic products. This enhances the production yield of this ceramic product production control method and system.

[0033] The specific steps for constructing a response sensitivity matrix based on experimental production data of ceramic products and using multiple regression analysis, and for calculating the adjustment scheme of front-end parameters of ceramic raw materials based on the initial compensation strategy and the response sensitivity matrix are as follows: The experimental production data of ceramic products are obtained, including experimental process parameter data of ceramic products and three-dimensional principal component fluctuation score change data of ceramic raw materials before and after process treatment. Multivariate regression analysis is performed on the experimental production data and a response sensitivity matrix is ​​constructed. Each column of the response sensitivity matrix represents a process parameter, each row represents a principal component in the three-dimensional principal component, and the matrix elements are the change in principal component score caused by a unit fluctuation of the process parameter. Based on the initial compensation strategy of ceramic raw materials, the type of process parameter to be adjusted is determined. Based on the current three-dimensional principal component fluctuation score and response sensitivity matrix of ceramic raw materials, the adjustment amount of the process parameter to be adjusted is obtained by calculation, and the front-end parameter adjustment scheme of ceramic raw materials is obtained.

[0034] It should be noted that the response sensitivity matrix is ​​constructed based on a single-factor fluctuation experiment, the experimental design of which is as follows: Sample size: 20 groups of raw materials were selected using the Latin hypercube sampling method. It was necessary to ensure that the selected 20 groups of raw materials covered the edge and center of the principal component space. Each group was repeated 5 times. All experiments were carried out in a constant temperature and humidity laboratory. The same batch of raw materials was fed in 5 batches. After each experiment, the equipment was idle for 30 minutes to restore the baseline state. Parameter adjustment gradient: Set 5 levels for each process parameter (e.g., ball mill speed: 40, 42.5, 45, 47.5, 50 Hz), adjust only one parameter at a time, and record the change ΔPC of each principal component score; Matrix element calculation: ,in Indicates when the first When the process parameter changes by a unit, the first... The change in the scores of each principal component. Indicates the first The unit adjustment amount for each process parameter This indicates that only the first [item] is adjusted. When the process parameter is 1, the 2nd process parameter is 3 The measured change in principal component scores, for example, if PC2 (particle size factor) decreases by 0.08 for every 1 Hz increase in ball mill speed, then R[PC2][speed] = -0.08; Matrix validation: Leave-one-out cross-validation was used to construct a matrix with 19 sets of raw material data and predict the change of principal components of the 20th raw material. The average prediction error was 12.7%, which meets the accuracy requirements for engineering applications (<15%).

[0035] It should be noted that the single-factor experimental method used above is based on the principle of engineering approximation. When the raw material fluctuation range is normal, that is, within the mean ±2σ, the parameter coupling effect can be approximated as linear superposition. To verify this hypothesis, a two-factor interactive experiment was designed: for example, while keeping other parameters constant, the pressing pressure was adjusted by +1MPa and the ball mill speed by -2Hz. The actual change in PC2 was observed to be 0.81, while the predicted value of the single-factor experimental matrix was +0.66 due to the pressure effect and +0.16 due to the speed effect. The superimposed predicted value was 0.82. The interaction error was |0.81-0.82| / 0.81≈1.2% < 5% of the engineering tolerance limit. Therefore, the single-factor matrix is ​​usable within this fluctuation range.

[0036] The above steps construct a response sensitivity matrix based on experimental production data of ceramic products and through multiple regression analysis. Based on the initial compensation strategy and the response sensitivity matrix, the adjustment scheme of the front-end parameters of ceramic raw materials is calculated. This not only directly maps the raw material fluctuation score to the adjustment amount of various front-end process parameters, but also proactively eliminates most of the impact of raw material fluctuations at a low cost before defects occur, so as to avoid the problem from being transmitted to the firing stage and causing irreversible losses. Furthermore, it can ensure that only key parameters are adjusted through priority screening, thereby eliminating new defects induced by overcompensation. This improves the production yield and practicality of this ceramic product production control method and system.

[0037] The specific steps for obtaining the optimal firing curve of the ceramic product based on the deviation between the pre-firing quality data and the target firing quality data using the particle swarm optimization algorithm are as follows: When the deviation between the pre-firing quality data and the target firing quality data is greater than the preset threshold, the particle swarm algorithm is activated and multiple particles are randomly generated. Each particle represents a candidate firing curve and is represented as a vector in four-dimensional space. The meanings of each dimension of this vector are heating rate, holding temperature, holding time and cooling rate, respectively, and the value range of each dimension is based on process specification constraints. Each particle is evaluated and a digital twin is used for rapid simulation. The objective function value is calculated based on the simulation results. The objective function is set based on the deformation deviation and color difference deviation after firing. The particles are iteratively optimized and updated to minimize the objective function until the objective function value is less than the preset threshold or the maximum number of iterations is reached, thus obtaining the optimal firing curve of the ceramic product.

[0038] For example, the firing curve parameter constraints for porcelain tiles (water absorption ≤ 0.5%) are as follows: Heating rate: 20-35℃ / h for oxidation section (500-700℃), 30-50℃ / h for reduction section (700-1000℃), and 25-40℃ / h for high temperature section (1000-1180℃). The upper limit of 50℃ / h is prone to causing cracking of the billet, and the lower limit of 20℃ / h affects the production capacity. Insulation temperature: Based on the differences in glaze formulation, it is set at 1180-1220℃, 1180-1200℃ for frit glaze and 1200-1220℃ for raw material glaze. A temperature deviation of ±5℃ will result in a shrinkage rate fluctuation of 0.15%. Holding time: 15-40 minutes, adjusted by the frequency of the roller kiln drive. For every 5 minutes the time is extended, the grain size increases by 0.5μm. Cooling rate: 80-150℃ / h for rapid cooling section (1220-700℃), 45-65℃ / h for slow cooling section (700-500℃). When the rate of slow cooling section exceeds 65℃ / h, the cracking rate of the product increases by 12%.

[0039] It should be noted that the key parameter settings for the particle swarm optimization algorithm are as follows: Population size: N=20, balancing search breadth and computation speed (1.5 seconds per particle simulation, total duration <30 seconds); Inertia weight: w=0.7 (linear decreasing strategy, initial 0.9, final 0.5), enhancing global exploration in the early stage and accelerating local convergence in the later stage; Learning factors: c1=c2=1.5, determined according to the Clerc shrinkage factor method to ensure convergence stability; Termination conditions: objective function value < 0.01 or number of iterations > 30, average convergence number of iterations 18.

[0040] For example, the specific optimization process is as follows: When the Fe2O3 content of a batch of raw materials increases by 0.05%, causing the predicted whiteness value to decrease by 2.5 degrees, the particle swarm optimization algorithm initially uses 20 particles, with an inertia weight w=0.7 and learning factors c1=c2=1.5. After 18 iterations (taking 9 seconds), the optimal particles are [28℃ / h, 1195℃, 22min, 52℃ / h]. This curve reduces the predicted deformation from 2.6mm to 1.9mm and the color difference from 2.8 to 1.3, meeting the superior product standard.

[0041] The specific steps for evaluating the process adjustment complexity of ceramic products based on the front-end parameter adjustment scheme and the optimal firing curve to obtain the process adjustment complexity index of ceramic products, and then making a decision on the production scheme of ceramic products based on the process adjustment complexity index to obtain the final production scheme of ceramic products are as follows: The process adjustment complexity of ceramic products is evaluated based on the deviation between the front-end parameter adjustment scheme and the optimal firing curve and the preset standard front-end parameter scheme and standard firing curve, and the process adjustment complexity index of ceramic products is obtained. When the process adjustment complexity index is less than the preset low threshold, the final production plan adopts both the front-end parameter adjustment plan and the optimal firing curve of the ceramic product. When the process adjustment complexity index is greater than the preset high threshold, the production plan of the ceramic product is not adjusted. When the process adjustment complexity index is between the preset high threshold and the preset low threshold, only the front-end parameter adjustment plan is adopted without adjusting the firing curve.

[0042] It should be noted that the formula for calculating the Process Adjustment Complexity Index (PCI) is as follows:

[0043] in For the first The adjustment amount of each parameter (the deviation from the standard value, i.e., the degree of deviation). This parameter represents the upper limit of the allowed physical adjustments. To adjust the number of parameters. For example, pressure adjustment -1.5MPa (upper limit ±3MPa), speed adjustment +1.2Hz (upper limit ±5Hz), then... ; The preset thresholds are as follows: Low threshold: PCI < 0.3, allowing full parameter adjustment; High threshold: PCI > 0.5, prohibiting adjustment and triggering production conversion; Middle range: 0.3 ≤ PCI ≤ 0.5, allowing only front-end parameter adjustment. The high and low thresholds are determined through historical successful batch statistics. Analysis of the past 500 successful compensation cases revealed that when PCI < 0.3, the compensation success rate is > 92%; when PCI > 0.5, the success rate is < 60% and the secondary defect rate is > 15%. Therefore, the thresholds are determined by data, not subjective experience.

[0044] It should be noted that the front-end parameter adjustment scheme mainly includes adjusting parameters such as ball mill speed, pressing pressure, and aging time. Compared with the subsequent firing curve adjustment, the front-end parameter adjustment scheme can eliminate 70-80% of the impact of raw material fluctuations. Furthermore, when adjusting the production scheme for ceramic products, it is necessary to ensure that the adjustment amounts of each parameter are within the physical limits of the equipment adjustment, and that the adjusted parameters do not violate any prohibitive rules of the ceramic product production process. To verify the effectiveness of the front-end parameter compensation, this embodiment includes a comparative experiment: 50 batches of fluctuating raw materials were taken; 25 batches underwent only front-end compensation (ball milling + pressing + drying), and 25 batches were not compensated. The results show: Compensation group: average shrinkage deviation 0.12%, deformation 1.95mm, color difference ΔE=1.2, overall pass rate 94.3%; Uncompensated group: average shrinkage deviation 0.58%, deformation 2.8mm, color difference ΔE=2.6, overall pass rate 68.7%; Therefore, front-end compensation reduced the quality deviation by 76.5%, proving that it can eliminate the dominant influence.

[0045] The above steps assess the process adjustment complexity of ceramic products based on the front-end parameter adjustment scheme and the optimal firing curve, obtaining the process adjustment complexity index of ceramic products. Based on the process adjustment complexity index, decisions are made on the production scheme of ceramic products. This not only avoids production interruptions, secondary quality defects, and efficiency losses caused by excessively complex parameter adjustments, but also achieves a dynamic balance between quality improvement needs and production stability, prioritizing the execution of simple and efficient compensation paths. This improves the overall operating efficiency of the ceramic product production line and the actual feasibility of the compensation strategy, thus enhancing the practicality of this ceramic product production control method and system.

[0046] Example 2: A ceramic product production control system, such as... Figure 2 As shown, it includes: The data acquisition and classification module includes a data acquisition and processing unit and a raw material fluctuation classification unit. The data acquisition and processing unit is used to perform XRF composition scanning and laser particle size detection on ceramic raw materials to obtain state parameter data of ceramic raw materials. The state parameter data of ceramic raw materials is processed for outliers, Z-score standardization and data vectorization to obtain preprocessed state parameter data. The raw material fluctuation classification unit is used to extract three principal components from the preprocessed state parameter data using principal component analysis and to determine the state fluctuation type of ceramic raw materials based on these components, thus obtaining the state fluctuation determination result of ceramic raw materials. The fluctuation impact assessment module is used to construct a state deviation transmission model based on the past production data of ceramic products. Based on the state deviation transmission model, it calculates three risk coefficients of ceramic raw materials and determines the priority of compensation strategies for ceramic raw materials, thereby obtaining the initial compensation strategy for ceramic raw materials. The feedforward parameter compensation module is used to construct a response sensitivity matrix based on experimental production data of ceramic products and multiple regression analysis. Based on the initial compensation strategy and the response sensitivity matrix, the front-end parameter adjustment scheme of ceramic raw materials is obtained through calculation. The feedback parameter correction module is used to construct a digital twin model of ceramic product firing and preset the initial firing curve of ceramic product. Based on the preprocessed state parameter data and front-end parameter adjustment scheme, the digital twin model is used to perform a pre-firing simulation of ceramic product to obtain the pre-firing quality data of ceramic product. Based on the deviation between the pre-firing quality data and the target firing quality data, the optimal firing curve of ceramic product is obtained through particle swarm optimization algorithm. The complexity assessment and decision-making module includes an adjustment complexity assessment unit and a production scheme decision-making unit. The adjustment complexity assessment unit is used to assess the process adjustment complexity of ceramic products based on the front-end parameter adjustment scheme and the optimal firing curve to obtain the process adjustment complexity index of ceramic products. The production scheme decision-making unit is used to make a decision on the production scheme of ceramic products based on the process adjustment complexity index to obtain the final production scheme of ceramic products.

[0047] It should be understood that those skilled in the art can make improvements or modifications based on the above description, and all such improvements and modifications should fall within the protection scope of the appended claims. Parts not described in detail in this specification are prior art known to those skilled in the art.

Claims

1. A method for controlling the production of ceramic products, characterized in that, Includes the following steps: XRF composition scanning and laser particle size detection were performed on the ceramic raw materials to obtain the state parameter data of the ceramic raw materials. Outlier processing, Z-score standardization and data vectorization were performed on the state parameter data of the ceramic raw materials to obtain preprocessed state parameter data. Principal component analysis was used to extract three principal components from the preprocessed state parameter data and to determine the state fluctuation type of the ceramic raw materials based on these components, thus obtaining the state fluctuation determination result of the ceramic raw materials. A state deviation transmission model is constructed based on the past production data of ceramic products. The three risk coefficients of ceramic raw materials are calculated according to the state deviation transmission model, and the priority of the compensation strategy for ceramic raw materials is determined accordingly to obtain the initial compensation strategy for ceramic raw materials. Based on the experimental production data of ceramic products, a response sensitivity matrix was constructed through multiple regression analysis. Based on the initial compensation strategy and the response sensitivity matrix, the front-end parameter adjustment scheme of ceramic raw materials was obtained through calculation. A digital twin model of ceramic product firing is constructed and the initial firing curve of the ceramic product is preset. Based on the preprocessed state parameter data and the front-end parameter adjustment scheme, the firing of the ceramic product is simulated using the digital twin model to obtain the simulated firing quality data of the ceramic product. Based on the deviation between the simulated firing quality data and the target firing quality data, the optimal firing curve of the ceramic product is obtained through particle swarm optimization algorithm. The complexity of the process adjustment of ceramic products is evaluated based on the front-end parameter adjustment scheme and the optimal firing curve to obtain the process adjustment complexity index of ceramic products. Based on the process adjustment complexity index, the production scheme of ceramic products is decided to obtain the final production scheme of ceramic products.

2. The method for controlling the production of ceramic products according to claim 1, characterized in that, The specific steps for extracting three principal components from the preprocessed state parameter data using principal component analysis and determining the state fluctuation type of the ceramic raw material based on these components to obtain the state fluctuation determination result of the ceramic raw material are as follows: Historical state parameter data of ceramic raw materials are obtained and the historical covariance matrix is ​​calculated. The historical covariance matrix is ​​then decomposed into eigenvalues ​​to obtain multiple eigenvalues ​​and their corresponding eigenvectors. The magnitude of the eigenvalue reflects the contribution of the data variance in the direction of the eigenvector. The three feature vectors with the highest contribution are selected as principal components and defined as the raw material composition dominant factor, raw material particle size dominant factor and raw material activity dominant factor, respectively. These three feature vectors are used to construct a projection matrix, and the preprocessed state parameter data is multiplied with the projection matrix to obtain the current three-dimensional principal component score of the ceramic raw material. The current three-dimensional principal component score is subtracted from the preset standard three-dimensional principal component score to obtain the current three-dimensional principal component fluctuation score of the ceramic raw material. The three principal components are classified into levels according to the principal component fluctuation score of each dimension. Based on the level classification result, the state fluctuation type of the ceramic raw material is determined as a single type of fluctuation or a composite type of fluctuation, and the state fluctuation determination result of the ceramic raw material is obtained.

3. The method for controlling the production of ceramic products according to claim 2, characterized in that, The specific steps for constructing a state deviation transmission model based on past production data of ceramic products, calculating three risk coefficients of ceramic raw materials based on the state deviation transmission model, and determining the priority of compensation strategies for ceramic raw materials to obtain the initial compensation strategy for ceramic raw materials are as follows: Obtain historical production data for ceramic products, including historical state parameter data of ceramic raw materials, historical three-dimensional principal component fluctuation scores, and historical quality deviation data of ceramic products. The historical quality deviation data includes shrinkage deviation, deformation deviation, and whiteness deviation of ceramic products, and the deviation refers to the difference between the measured value and the preset standard value. A multiple linear regression model was used as the initial model, and the relationship between the historical three-dimensional principal component fluctuation score of ceramic raw materials and various deviations in the historical quality deviation data was constructed. The coefficients of the relationship were solved by the least squares method to obtain the state deviation transmission model. The current three-dimensional principal component fluctuation score of the ceramic raw material is input into the state deviation transmission model to obtain the current predicted value of each deviation. The ratio of the current predicted value of each deviation to the preset standard value is calculated and used as the three risk coefficients of the ceramic raw material, including shrinkage risk coefficient, deformation risk coefficient and color difference risk coefficient. The priority of each risk is determined based on the three risk coefficients of the ceramic raw materials and their corresponding preset risk thresholds. Based on this, the parameter compensation mode is set as either single-parameter precise compensation or multi-parameter collaborative compensation. The initial compensation strategy for the ceramic raw materials is determined based on the set parameter compensation mode and the state fluctuation judgment results of the ceramic raw materials.

4. The method for controlling the production of ceramic products according to claim 3, characterized in that, The specific steps for constructing a response sensitivity matrix based on experimental production data of ceramic products and using multiple regression analysis, and for calculating the adjustment scheme of front-end parameters of ceramic raw materials based on the initial compensation strategy and the response sensitivity matrix are as follows: The experimental production data of ceramic products are obtained, including experimental process parameter data of ceramic products and three-dimensional principal component fluctuation score change data of ceramic raw materials before and after process treatment. Multivariate regression analysis is performed on the experimental production data and a response sensitivity matrix is ​​constructed. Each column of the response sensitivity matrix represents a process parameter, each row represents a principal component in the three-dimensional principal component, and the matrix elements are the change in principal component score caused by a unit fluctuation of the process parameter. Based on the initial compensation strategy of ceramic raw materials, the type of process parameter to be adjusted is determined. Based on the current three-dimensional principal component fluctuation score and response sensitivity matrix of ceramic raw materials, the adjustment amount of the process parameter to be adjusted is obtained by calculation, and the front-end parameter adjustment scheme of ceramic raw materials is obtained.

5. The method for controlling the production of ceramic products according to claim 4, characterized in that, The specific steps for obtaining the optimal firing curve of the ceramic product based on the deviation between the pre-firing quality data and the target firing quality data using the particle swarm optimization algorithm are as follows: When the deviation between the pre-firing quality data and the target firing quality data is greater than the preset threshold, the particle swarm algorithm is activated and multiple particles are randomly generated. Each particle represents a candidate firing curve and is represented as a vector in four-dimensional space. The meanings of each dimension of this vector are heating rate, holding temperature, holding time and cooling rate, respectively, and the value range of each dimension is based on process specification constraints. Each particle is evaluated and a digital twin is used for rapid simulation. The objective function value is calculated based on the simulation results. The objective function is set based on the deformation deviation and color difference deviation after firing. The particles are iteratively optimized and updated to minimize the objective function until the objective function value is less than the preset threshold or the maximum number of iterations is reached, thus obtaining the optimal firing curve of the ceramic product.

6. The method for controlling the production of ceramic products according to claim 5, characterized in that, The specific steps for evaluating the process adjustment complexity of ceramic products based on the front-end parameter adjustment scheme and the optimal firing curve to obtain the process adjustment complexity index of ceramic products, and then making a decision on the production scheme of ceramic products based on the process adjustment complexity index to obtain the final production scheme of ceramic products are as follows: The process adjustment complexity of ceramic products is evaluated based on the deviation between the front-end parameter adjustment scheme and the optimal firing curve and the preset standard front-end parameter scheme and standard firing curve, and the process adjustment complexity index of ceramic products is obtained. When the process adjustment complexity index is less than the preset low threshold, the final production plan adopts both the front-end parameter adjustment plan and the optimal firing curve of the ceramic product. When the process adjustment complexity index is greater than the preset high threshold, the production plan of the ceramic product is not adjusted. When the process adjustment complexity index is between the preset high threshold and the preset low threshold, only the front-end parameter adjustment plan is adopted without adjusting the firing curve.

7. A ceramic product production control system, applied to the ceramic product production control method according to any one of claims 1-6, characterized in that, Including: The data acquisition and classification module includes a data acquisition and processing unit and a raw material fluctuation classification unit. The data acquisition and processing unit is used to perform XRF composition scanning and laser particle size detection on ceramic raw materials to obtain state parameter data of ceramic raw materials. The state parameter data of ceramic raw materials is processed for outliers, Z-score standardization and data vectorization to obtain preprocessed state parameter data. The raw material fluctuation classification unit is used to extract three principal components from the preprocessed state parameter data using principal component analysis and to determine the state fluctuation type of ceramic raw materials based on these components, thus obtaining the state fluctuation determination result of ceramic raw materials. The fluctuation impact assessment module is used to construct a state deviation transmission model based on the past production data of ceramic products. Based on the state deviation transmission model, it calculates three risk coefficients of ceramic raw materials and determines the priority of compensation strategies for ceramic raw materials, thereby obtaining the initial compensation strategy for ceramic raw materials. The feedforward parameter compensation module is used to construct a response sensitivity matrix based on experimental production data of ceramic products and multiple regression analysis. Based on the initial compensation strategy and the response sensitivity matrix, the front-end parameter adjustment scheme of ceramic raw materials is obtained through calculation. The feedback parameter correction module is used to construct a digital twin model of ceramic product firing and preset the initial firing curve of ceramic product. Based on the preprocessed state parameter data and front-end parameter adjustment scheme, the digital twin model is used to perform a pre-firing simulation of ceramic product to obtain the pre-firing quality data of ceramic product. Based on the deviation between the pre-firing quality data and the target firing quality data, the optimal firing curve of ceramic product is obtained through particle swarm optimization algorithm. The complexity assessment and decision-making module includes an adjustment complexity assessment unit and a production scheme decision-making unit. The adjustment complexity assessment unit is used to assess the process adjustment complexity of ceramic products based on the front-end parameter adjustment scheme and the optimal firing curve to obtain the process adjustment complexity index of ceramic products. The production scheme decision-making unit is used to make a decision on the production scheme of ceramic products based on the process adjustment complexity index to obtain the final production scheme of ceramic products.

Images