Digital batching method for production of anti-degumming EVA (Ethylene Vinyl Acetate) outsole

Abstract

The invention provides a digital batching method for anti-degumming EVA outsole production, which comprises the following steps: judging whether predicted values of foaming performance, mechanical performance and bonding performance of an EVA foaming material exceed a preset performance threshold range or not, if so, triggering a formula optimization program, and if not, executing the formula optimization program; the use amount of a foaming agent, the use amount of a cross-linking agent and the proportion of thermosensitive components in the EVA foaming material formula are optimized and adjusted; and transmitting the combined data of the optimal foaming agent dosage, the cross-linking agent dosage and the thermosensitive component proportion obtained by the genetic algorithm search to a batching control system of the EVA foaming material production line, and adjusting the adding proportion of the foaming agent, the cross-linking agent and the thermosensitive component to realize the online optimization control of the formula.

Description

Technical Field

[0001] The invention relates to the field of information technology, and in particular to a digital batching method for producing an anti-debonding EVA outsole. Background Art

[0002] The core technical problem faced in the production process of EVA outsoles is how to achieve precise control of formulas and processes in a complex and changeable production environment to ensure the stability of product quality. Specifically, the performance of EVA materials is highly sensitive to temperature. Temperature fluctuations during the production process can cause material performance to deviate from expectations, thereby affecting the quality of the finished product. At the same time, there are complex interactions between the components of the formula, and changes in a single component may cause a chain reaction of multiple performance indicators. How to achieve precise control of material properties, foaming uniformity and dimensional stability in such a system is a difficult problem that needs to be solved urgently. Traditional fixed formulas and empirical adjustment methods are difficult to adapt to this complexity and cannot respond to various fluctuations in the production process in a timely manner. Therefore, there is an urgent need for an intelligent solution that can perceive changes in the production environment in real time, quickly optimize formulas and process parameters, and accurately control the molding process. This not only involves formula optimization and process control, but also requires the organic combination of multiple links such as data acquisition, analytical modeling, optimization algorithms, and process control to form a closed-loop intelligent manufacturing system. Summary of the invention

[0003] The present invention provides a digital batching method for producing anti-debonding EVA outsole, which mainly comprises:

[0004] By installing temperature sensors on the production line, the temperature data of the EVA foam material is collected in real time, and the temperature data includes the foaming temperature and the cooling rate. According to the different formula components of the EVA foam material, an orthogonal experimental scheme is designed. By changing the proportion of the formula components, the performance test of the EVA foam material is carried out under different foaming temperatures and cooling rates, and the foaming performance, mechanical properties and bonding performance index data are obtained, and an experimental data set including temperature, formula components and performance indicators is established;

[0005] The support vector machine regression algorithm is used to train the EVA foam material performance prediction model with foaming temperature, cooling rate and formula ingredients as input and performance indicators as output. The real-time collected temperature data is input into the trained performance prediction model to obtain the predicted values ​​of the foaming performance, mechanical properties and bonding properties of the EVA foam material under the current temperature conditions.

[0006] Determine whether the predicted values ​​of the foaming performance, mechanical properties and adhesive properties of the EVA foam material exceed the preset performance threshold range. If they exceed the threshold range, trigger the formula optimization program to optimize and adjust the amount of foaming agent, the amount of cross-linking agent and the proportion of heat-sensitive components in the EVA foam material formula;

[0007] In the formula optimization program, a genetic algorithm is used to optimize the EVA foam material formula, with the predicted values ​​of foaming performance, mechanical properties and adhesive properties as the fitness function. At the same time, the formula cost and process adaptability are combined, and the optimal combination of foaming agent dosage, crosslinking agent dosage and heat-sensitive component ratio is iteratively searched through selection, crossover and mutation genetic operations;

[0008] The optimal amount of foaming agent, cross-linking agent and heat-sensitive component ratio combination data obtained by genetic algorithm search is transmitted to the batching control system of the EVA foam material production line, and the addition ratio of foaming agent, cross-linking agent and heat-sensitive component is adjusted to achieve online optimization control of the formula;

[0009] The optimized and adjusted EVA foam material formula data is transmitted to the production line control system, and the process parameters of the production line are automatically adjusted. The EVA foam material is processed based on the production line after the process parameters are adjusted to obtain an EVA insole, and the EVA insole is heated and laminated with a pre-prepared composite rubber outsole to obtain an EVA outsole;

[0010] Through data visualization technology, the temperature data, performance prediction values, and formula optimization results of EVA foaming materials are graphically displayed, and a three-dimensional response surface diagram of foaming temperature, cooling rate, and performance indicators is drawn to intuitively present the nonlinear relationship between formula parameters and product performance, so that production managers can grasp the production status in real time.

[0011] The technical solution provided by the embodiment of the present invention may have the following beneficial effects:

[0012] The present invention discloses a digital batching method for the production of anti-degumming EVA outsole. By real-time acquisition of temperature data and combining with a pre-built model, the performance change trend of the EVA foam material is predicted. When the predicted performance exceeds the threshold, the formula optimization program is triggered. The program uses a temperature-sensitive database to extract matching data, uses a genetic algorithm to search for the optimal combination of thermosensitive component dosages, and performs iterative optimization with the goal of meeting preset quality requirements. The optimized formula data is transmitted to the production line control system to automatically adjust the process parameters. The present invention realizes real-time monitoring and intelligent optimization of the performance of the EVA foam material, ensures the stability of the material performance under different temperature conditions, and improves production efficiency and product quality. At the same time, the method can also be applied to the preparation of the EVA insole and the heating and bonding with the composite rubber outsole, realizing the intelligent control of the entire process of EVA outsole production. BRIEF DESCRIPTION OF THE DRAWINGS

[0013] Figure 1 The present invention is a flow chart of a digital batching method for producing anti-debonding EVA outsole.

[0014] Figure 2 The present invention is a schematic diagram of a digital batching method for producing an anti-debonding EVA outsole.

[0015] Figure 3 It is another schematic diagram of a digital batching method for producing anti-debonding EVA outsole of the present invention. DETAILED DESCRIPTION

[0016] In order to enable those skilled in the art to better understand the technical solutions in this specification, the technical solutions in the embodiments of this specification will be clearly and completely described below in conjunction with the drawings in the embodiments of this specification. Obviously, the described embodiments are only part of the embodiments of this specification, not all of the embodiments. Based on the embodiments in this specification, all other embodiments obtained by ordinary technicians in this field without creative work should fall within the scope of protection of this specification.

[0017] like Figure 1-3 In this embodiment, a digital batching method for producing anti-debonding EVA outsole may specifically include:

[0018] S101. Temperature data of EVA foaming materials are collected in real time by installing temperature sensors on the production line. The temperature data includes foaming temperature and cooling rate. According to the different formula components of EVA foaming materials, an orthogonal experimental scheme is designed. By changing the proportion of formula components, the performance of EVA foaming materials is tested under different foaming temperatures and cooling rates. The foaming performance, mechanical properties and bonding performance index data are obtained, and an experimental data set including temperature, formula components and performance indicators is established.

[0019] The temperature data of EVA foaming material is collected from the temperature sensor array, and the cooling rate value is obtained through the temperature gradient calculation formula according to the temperature data; the compressive strength, compression permanent deformation rate and shear strength of the foaming material are measured by a universal testing machine according to the raw material ratio, and the bubble morphology and distribution data of the foaming material are obtained through a scanning electron microscope; the temperature data and the performance index data are correlated with each other using a multivariate linear regression tool, and the mapping relationship between the foaming temperature and the performance index is obtained according to the correlation calculation result; the formula components and temperature parameters are trained using a fuzzy neural network tool, the temperature parameter threshold is judged according to the training result, and the heating device and the cooling device are adjusted according to the temperature parameter threshold.

[0020] Specifically, an infrared temperature sensor array is arranged at intervals of 2 meters on the production line, and the sampling frequency is set to 100 samples per second to obtain temperature data. The foaming temperature data of the EVA foam material is collected from the temperature sensor, and the cooling rate value is obtained by the temperature gradient calculation formula, and the temperature change curve is recorded in the data acquisition unit. According to the orthogonal experimental table, the EVA resin content, the foaming agent content, and the cross-linking agent content are selected as experimental factors, and the raw material ratio is carried out according to the L9 orthogonal experimental table. A high-precision balance is used for weighing, and the bubble morphology and distribution of the obtained foam material are observed by a scanning electron microscope. The foaming ratio, bubble diameter, and closed cell rate are statistically analyzed, and the compressive strength, compression permanent deformation rate, and shear strength values ​​of the obtained foam material are measured by a universal testing machine. The temperature data and the performance index data are associated and calculated using a multivariate linear regression tool, and the performance index includes the foaming ratio, bubble diameter, closed cell rate, compressive strength, compression permanent deformation rate, and shear strength. The parameter weight is obtained from the regression tool, and the mapping relationship between the foaming temperature, cooling rate, and performance index is established according to the parameter weight, and the key temperature control parameters are obtained from the mapping relationship. Fuzzy neural network tools are used to train and predict formula components, temperature parameters, and performance indicators. The upper and lower thresholds of temperature parameters are determined based on the output results of the neural network. Real-time temperature data is obtained from the temperature sensor. The heating device and cooling device are adjusted through the proportional integral differential controller. The controller parameters are determined online based on the temperature control parameters. In the layout of the infrared temperature sensor array in the EVA foaming production line, the sensor spacing distance is selected to be 2 meters because there is a gradient change in the temperature distribution of the foaming material during the flow deposition. Too large spacing will cause a blind area for temperature measurement, and too small spacing will cause data redundancy. It has been verified that the 2-meter spacing ensures the complete measurement of the temperature field distribution and avoids data redundancy. The sampling frequency of 100 times / second is based on the temperature change rate during the EVA foaming process. Usually, the cooling process of the foaming temperature from 160°C to 80°C lasts for 300 seconds. The temperature gradient is calculated by dividing the temperature difference between adjacent measuring points by the distance. In the orthogonal experimental design, the EVA resin content is selected at three levels of 70%, 75%, and 80%, the foaming agent content is selected at three levels of 1%, 2%, and 3%, and the cross-linking agent content is selected at three levels of 0.5%, 1.0%, and 1.5%. The L9 orthogonal experimental table contains 9 groups of formula combinations. In the bubble morphology observation, the scanning electron microscope magnification is set to 500 times. At this magnification, the bubble diameter distribution range is 50-200 microns, the closed cell rate is in the range of 85%-95%, and the foaming ratio is in the range of 15-25 times. The compressive strength test uses a 100×100×25 mm specimen, a compression rate of 2 mm / min, and records the 50% compression stress value.When multiple linear regression is used to establish the relationship between temperature and performance indicators, the foaming temperature range of 160-180℃ and the cooling rate of 0.5-2.0℃ / second are used as independent variables, and the foaming ratio, bubble diameter, closed cell rate, and compressive strength are used as dependent variables. The regression coefficient reflects the influence of temperature parameters on each performance index. The regression coefficient of foaming temperature and foaming ratio is 0.85, indicating that the two are strongly correlated. The input layer of the fuzzy neural network contains 5 nodes including formula component content, foaming temperature, and cooling rate. The hidden layer uses 10 nodes, the output layer is 6 performance indicator nodes, and the training sample data is 90 groups. In the proportional integral differential control, the temperature control accuracy is set to ±1℃, the proportional coefficient, integral time, and differential time are determined by the Ziegler-Nichols setting method, the heating power adjustment range is 0-100%, and the cooling water flow adjustment range is 0-2 cubic meters / hour. EVA foaming performance testing includes multiple dimensions. In the process of optimizing production process parameters, the foaming temperature is most closely related to the foaming ratio and bubble morphology. Too high a temperature causes bubble rupture, and too low a temperature causes insufficient foaming. Experimental data show that the bubble structure is most uniform when the foaming temperature is 170°C. The cooling rate is closely related to the mechanical properties of the material. The best mechanical properties are obtained at a cooling rate of 1.0°C / second, when the compressive strength reaches 0.8MPa and the compression permanent deformation rate is less than 5%.

[0021] S102. Using a support vector machine regression algorithm, with foaming temperature, cooling rate and formula ingredients as input and performance indicators as output, a performance prediction model for EVA foaming materials is trained, and the real-time collected temperature data is input into the trained performance prediction model to obtain the predicted values ​​of the foaming performance, mechanical properties and bonding properties of the EVA foaming material under the current temperature conditions.

[0022] According to the temperature sensor sampling data, a moving window is used to calculate the mean and standard deviation of the foaming temperature data and the cooling rate data, and spline interpolation is performed on the data in the temperature data that exceeds the standard deviation range to obtain the corrected temperature data; the maximum and minimum value normalization method is used to process the corrected temperature data, and the foaming performance index and the mechanical performance index are extracted from the normalized data to obtain a performance index data set; the kernel parameters and penalty parameters of the Gaussian kernel function of the support vector machine are traversed by a grid search tool, and the root mean square error value is calculated in cross validation to obtain the optimal parameter combination; the support vector machine regression predictor is constructed using the optimal parameter combination, the performance prediction of the validation set is performed and the root mean square error is calculated, the correction coefficient is obtained through the root mean square error, and the prediction result is calibrated using the correction coefficient.

[0023] Specifically, when the temperature sensor sampling period is set to 10 milliseconds, a 60-second moving window is used to calculate the mean and standard deviation of the foaming temperature, cooling rate, and formula composition data from the sampled data, and cubic spline interpolation is performed on the data that exceeds the range of 3 times the standard deviation. The maximum and minimum normalization is used to map the data to the range of 0 to 1, and the foaming performance indicators including foaming ratio, bubble diameter, and closed cell rate are extracted from the normalized data. The mechanical performance indicators include compressive strength and compression permanent deformation rate, and the bonding performance indicators include shear strength data sets. The obtained data set is divided into training set, validation set, and test set in a ratio of 7:2:1. The kernel parameter search range is set from 0.01 to 100 in the Gaussian kernel function of the support vector machine, and the penalty parameter search range is set from 0.1 to 1000. The grid search tool is used to traverse the parameters, and the root mean square error value corresponding to each group of parameters is calculated in the five-fold cross validation. The parameter combination corresponding to the minimum root mean square error is selected based on the obtained error value. The Gaussian kernel function and the optimal parameter combination are used to construct the support vector machine regression predictor. The performance prediction is performed on the validation set. The root mean square error between the predicted value and the true value is calculated. The learning rate is set to 0.01, the number of iterations is set to 1000, and the root mean square error is used to correct the prediction deviation. The correction coefficient is obtained from the deviation correction process, and the correction coefficient is used to calibrate the prediction results of the test set. In the real-time data collection process, the temperature sensor collected data is normalized according to the 60-second data cache interval, and the normalized data is input into the support vector machine regression predictor to obtain the performance prediction results. The prediction results are corrected using the calibration correction coefficient, and the corrected performance prediction results are stored in the specified data table, which contains the acquisition timestamp, temperature data, and performance prediction value fields. In the temperature sensor data collection process, the sampling period is selected to be 10 milliseconds due to the temperature change characteristics of the EVA foaming material. During the foaming process, the temperature drops from 180°C to 80°C, and the entire cooling process lasts for 300 seconds. If the sampling period is too long, the temperature mutation point will be missed, and if it is too short, data redundancy will increase. The moving window is set to 60 seconds corresponding to the time constant of the thermal conductivity characteristics of the foaming material. The calculation of data standard deviation shows that the standard deviation of foaming temperature is 2.5℃, the standard deviation of cooling rate is 0.3℃ / second, and the proportion of outliers exceeding 3 times the standard deviation is 0.5%. These outliers mostly appear in the start-up and stop stages of the production line. The data set is divided into 7:2:1 ratio instead of the traditional 8:1:1 ratio. The reason is that the temperature change in the EVA foaming process is regular, and increasing the proportion of the validation set helps to improve the parameter optimization effect. In the optimization of the kernel function parameters of the support vector machine, the lower limit of the kernel parameter 0.01 corresponds to the minimum change scale of the temperature data, and the upper limit 100 corresponds to the maximum change range of the temperature data. The penalty parameter range is determined based on the tolerance of the foaming material performance indicators. In the cross-validation, the root mean square error gradually decreases from 0.15 to 0.08. In the prediction deviation correction, the learning rate is set to 0.01. Considering the change range of the performance indicators of the EVA foam material, a large learning rate will lead to correction oscillation, and a small learning rate will converge slowly.The prediction results show that the prediction error of the foaming ratio is within 5%, the prediction error of the bubble diameter is within 10 microns, the prediction error of the closed porosity is within 2%, and the prediction error of the compressive strength in the mechanical performance index is within 0.05MPa. The real-time data processing adopts a 60-second cache interval, which cooperates with the temperature sensor sampling cycle to accumulate 6,000 temperature data points in each cache cycle and input the predictor after normalization. In the prediction result data table, the prediction value range of the foaming ratio is 15-25, the prediction value range of the bubble diameter is 50-200 microns, the prediction value range of the closed porosity is 85%-95%, the prediction value range of the compressive strength is 0.6-1.0MPa, the prediction value range of the compression permanent deformation rate is 3%-8%, and the prediction value range of the shear strength is 0.3-0.6MPa. These prediction intervals are consistent with the actual production process control indicators. The data table is stored as a data file per hour for subsequent query and analysis.

[0024] S103, determining whether the predicted values ​​of the foaming properties, mechanical properties and adhesive properties of the EVA foam material exceed a preset performance threshold range; if so, triggering a formula optimization program to optimize and adjust the amount of the foaming agent, the amount of the cross-linking agent and the ratio of the heat-sensitive component in the EVA foam material formula.

[0025] A sliding window numerical filter is used to process the performance data, and an identification bit is generated for performance indicators that exceed a preset threshold range, and the identification bit is a positive deviation value or a negative deviation value; recipe data is obtained from a historical database according to the performance indicator deviation value, and the correlation between recipe components and performance indicators is obtained through the Pearson correlation coefficient, and the sensitivity of the recipe components to the performance indicators is obtained by multivariate linear regression; genetic algorithm parameters are set according to the sensitivity of recipe components, and the recipe parameters are optimized and calculated, and the optimized recipe with the highest fitness is obtained through preset threshold verification; the optimized recipe parameters are written into the batching device database by an industrial control unit, and real-time feeding data is collected through a mass sensor. If the deviation between the feeding data and the set value exceeds the threshold, the flow data is obtained from the proportional control valve to adjust the flow controller output.

[0026] Specifically, in the process of processing the predicted value of the performance of the EVA foam material, a sliding window numerical filter is used to process the performance data to determine whether the foaming ratio is within the range of 15 to 25 times, whether the bubble diameter is within the range of 50 to 200 microns, whether the closed cell rate is within the range of 85% to 95%, whether the compressive strength in the mechanical properties is within the range of 0.6 to 1.0 MPa, whether the compression permanent deformation rate is within the range of 3% to 8%, whether the shear strength is within the range of 0.3 to 0.6 MPa, and the deviation value of the performance indicators that exceed the preset threshold range is calculated and an identification bit is generated, and the value of the identification bit is a positive deviation of 1 or a negative deviation of -1. For the performance index deviation values ​​that exceed the preset threshold range, the formula data of the foaming agent dosage range of 4% to 8%, the cross-linking agent dosage range of 1% to 3%, and the thermosensitive component ratio range of 10% to 20% are extracted from the historical database of the last 180 days. The Pearson correlation coefficient is used to calculate the correlation between the formula components and the performance indicators, and the standardized regression coefficient is obtained from the multivariate linear regression tool. The standardized regression coefficient is used as the sensitivity of the influence of the formula components on the performance indicators. The genetic algorithm parameters are set according to the sensitivity of the formula components, the population size is set to 100, the crossover probability is set to 0.8, and the mutation probability is set to 0.1. The formula parameters of the foaming agent dosage, the cross-linking agent dosage, and the thermosensitive component ratio are optimized, and the formula parameter change step is set to 0.1%. The top 10 formula combinations with the highest fitness are selected in each iteration, and the feasibility of the formula combination is verified by the preset threshold. The formula with the highest fitness is selected from the qualified formulas as the optimization result. The optimized formula parameters are written into the batching device database by the industrial control unit. According to the fixed values ​​of the amount of foaming agent, the amount of cross-linking agent, and the ratio of thermosensitive components in the formula parameter setting, the sampling period is set to 100 milliseconds during the feeding process. The real-time feeding data is collected through the quality sensor. When the feeding data deviates from the set value by more than 0.5%, the real-time flow data is obtained from the proportional control valve, and the flow controller output is adjusted to correct the ratio deviation. The performance threshold of EVA foaming material is set based on process experience and product standards. For example, the foaming ratio threshold is set in the range of 15 to 25 times. If it is lower than 15 times, it indicates that the foaming is insufficient. If it is higher than 25 times, the bubble structure is easy to collapse. The bubble diameter is controlled in the range of 50 to 200 microns. Too small will lead to excessive number of bubbles and uneven foaming. Too large will easily form connected bubbles and reduce the closed cell rate. The closed cell rate is controlled in the range of 85% to 95% to reflect the degree of closure of the bubble structure. When the performance index exceeds the threshold range, the deviation direction is recorded through the identification bit. For example, the identification bit is -1 when the foaming ratio is 14 times, and 1 when it is 26 times. The sensitivity analysis of formula components shows that the amount of foaming agent has the greatest impact on the foaming ratio, with a standardized regression coefficient of 0.85, followed by the bubble diameter, with a coefficient of 0.65, and a smaller impact on the closed cell rate, with a coefficient of 0.35. The amount of cross-linking agent mainly affects the mechanical properties, with a standardized regression coefficient of 0.75 for compressive strength and a coefficient of 0.68 for compression permanent deformation rate.The proportion of thermosensitive components affects both foaming performance and mechanical properties, with a foaming ratio coefficient of 0.45 and a compressive strength coefficient of 0.55. In the process of formula optimization, the genetic algorithm population size is set to 100, which is the result of weighing the amount of calculation and the optimization effect. It is difficult to fully search the solution space if it is too small, and it increases the calculation burden if it is too large. The settings of the crossover probability of 0.8 and the mutation probability of 0.1 enable the algorithm to maintain both population diversity and convergence. Ten optimal formulas are retained in each round of iteration to ensure high-quality gene inheritance. The step size of the formula parameter change of 0.1% takes into account the control accuracy of the feeding equipment. Changes less than 0.1% are difficult to accurately control in actual production. The batching process uses a quality sensor to monitor the feeding amount in real time. The sampling period of 100 milliseconds corresponds to the response time of the feeding valve to achieve rapid feedback control. When the feeding deviation exceeds 0.5%, the correction is triggered. For example, when the target addition amount of the foaming agent is 5.0% and the actual addition amount reaches 5.03%, the flow rate is reduced by adjusting the valve opening. The flow rate regulation adopts proportional integral control, with an integral time constant of 2 seconds and a proportional coefficient of 0.8 to avoid oscillation caused by too fast regulation. Industrial practice shows that the above parameter settings have good dynamic response characteristics while ensuring control accuracy.

[0027] S104. In the formula optimization program, a genetic algorithm is used to optimize the EVA foaming material formula, with the predicted values ​​of foaming performance, mechanical properties and adhesive properties as the fitness function. At the same time, the formula cost and process adaptability are combined, and the optimal combination of foaming agent dosage, crosslinking agent dosage and thermosensitive component ratio is iteratively searched through selection, crossover and mutation genetic operations.

[0028] The weighted summation method is used to construct the recipe fitness function, and the recipe score is calculated according to the performance indicator weights and the performance test values; the component dosage range is obtained from the process database, and the recipe parameter gene sequence is generated through binary coding; the selection probability is calculated according to the recipe score, and the roulette selection operator is used to obtain the recipe individual, and the single-point crossover operation and non-uniform mutation operation are performed on the recipe individual to obtain a new generation of recipe population; if the continuous change amplitude of the fitness value of the best individual in the group is less than the preset threshold, the best individual is decoded to obtain the recipe parameters, and it is judged whether the recipe parameters meet the process requirements through process verification. If the process requirements are met, it is determined to be the optimal recipe.

[0029] Specifically, the weighted summation method is used to construct the formula fitness function, and the weight of the foaming ratio, the weight of the bubble diameter, the weight of the closed cell ratio, the weight of the compressive strength, the weight of the compression permanent deformation rate, and the weight of the shear strength are set to 0.15. The unit price of the foaming agent, the unit price of the cross-linking agent, and the unit price of the thermosensitive component are read from the raw material database. The total cost of the formula is calculated according to the product of the raw material dosage and the unit price. The process adaptability score is evaluated according to the three indicators of temperature range, pressure range, and mixing uniformity. The performance indicators, cost indicators, and process indicators are normalized by the maximum and minimum values ​​to obtain the fitness value. From the process database, the dosage range of foaming agent is 4% to 8%, the dosage range of cross-linking agent is 1% to 3%, and the ratio range of heat-sensitive components is 10% to 20%. The formula components are encoded using binary coding, and 100 initial individuals containing the genetic coding of formula parameters are generated. The crossover probability is set to 0.8 and the mutation probability is set to 0.1 in the genetic operation. The selection probability is calculated according to the individual fitness value, and the formula individual is selected according to the selection probability using the roulette selection operator. A single-point crossover operation is performed on the selected formula individual, and the crossover position is randomly selected in the formula gene coding. If the formula component exceeds the process parameter range after the crossover, repair is performed. The gene position is mutated using a non-uniform mutation operator with an adaptive mutation range. The variable asynchronous length gradually decreases with the number of iterations. The mutated formula is decoded to obtain the group distribution ratio, and the formula performance prediction value is obtained from the support vector machine regression tool, and the fitness value corresponding to the prediction value is calculated. When the change in the fitness value of the best individual in the population for 20 consecutive generations is less than 0.01, it is determined that the convergence condition has been reached, and the formula parameters are decoded for the best individual. The process verification is carried out according to the temperature range, pressure range, and mixing uniformity requirements in the process parameter table. If all indicators meet the requirements, they are recorded as the optimal formula. If there are indicators that do not meet the requirements, the optimization calculation is triggered again. The optimization of the EVA foaming material formula involves the trade-off of multiple performance indicators. The weight setting of the performance indicators reflects the emphasis on product performance. The higher weights of the foaming ratio and compressive strength reflect the lightweight and load-bearing requirements of the product. The weights of the indicators related to the bubble structure are second, reflecting the importance of the internal structure. In the raw material cost calculation, the unit price of the foaming agent is 200 yuan / kg, the unit price of the cross-linking agent is 180 yuan / kg, and the unit price of the thermosensitive component is 150 yuan / kg. The cost is calculated according to the actual addition amount. In the process adaptability score, the temperature range of 160-180℃ gets full marks, the pressure range of 15-20 MPa gets full marks, and the mixing uniformity is evaluated by observing the bubble distribution through a scanning electron microscope. The formula parameter coding adopts binary mode. The 4%-8% interval of foaming agent dosage is divided into 64 equal parts, represented by 6 bits of binary. The 1%-3% interval of cross-linking agent dosage is divided into 32 equal parts, represented by 5 bits of binary. The 10%-20% interval of heat-sensitive component ratio is divided into 64 equal parts, represented by 6 bits of binary. The coding accuracy meets the actual production requirements. For example, the coding accuracy of foaming agent dosage is 0.0625%. The selection probability calculation adopts fitness value normalization. Individuals with a fitness value of 0.8 have a selection probability of 0.1.The crossover operation randomly selects the crossover position in the 17-bit gene code. If the result of 8.2% of the foaming agent dosage is out of range, it is corrected to 8% through boundary processing. The initial variation range of the mutation operator is 10% of the parameter variation interval, which decreases linearly with the increase of the number of iterations. The variation range is reduced to 1% after 500 generations of iterations. The performance indicators of the formula after mutation are predicted by the support vector machine. The predicted values ​​of foaming ratio are 20 times, bubble diameter is 120 microns, and closed cell rate is 90%, corresponding to the fitness value of 0.85. The threshold value of the fitness value change amplitude of 0.01 in the convergence criterion is determined based on the tolerance of the actual production parameters. The fluctuation of the optimal individual in the population for 20 consecutive generations is less than the threshold, indicating that the optimization has reached a stable state. The process verification process checks whether the temperature control accuracy of ±1℃, pressure fluctuation of ±0.5 MPa, and mixing uniformity deviation of ±5% meet the requirements. Practice shows that after about 300 generations of iterations, the formula optimization results of 6.2% of foaming agent dosage, 1.8% of cross-linking agent dosage, and 15.5% of heat-sensitive component ratio meet both performance requirements and process feasibility.

[0030] S105, transmitting the optimal foaming agent dosage, cross-linking agent dosage and heat-sensitive component ratio combination data obtained by the genetic algorithm search to the batching control system of the EVA foam material production line, adjusting the addition ratio of the foaming agent, cross-linking agent and heat-sensitive component, and realizing online optimization control of the formula.

[0031] Material weight data is collected according to the optimized formula parameters received by the industrial controller, the material weight data is obtained by a balance sensor and filtered proportion data is obtained through a cumulative averaging method; the filtered proportion data is used to control an electromagnetic flowmeter to measure material flow, and the servo motor is driven by the material flow data to adjust the speed of the screw feeder; the mixed slurry output from the screw feeder is detected by a density meter and a spectrometer, and the mixing state is judged by the temperature data collected by the mixer wall temperature sensor; a proportional-integral control signal is output to an electric control valve according to the mixing state data, and the electric control valve adjusts the opening according to the control signal, and the material formula is adjusted online through the opening data.

[0032] Specifically, an industrial controller is used to receive optimized formula parameters with a sampling period of 100 milliseconds, and the amount of foaming agent, the amount of cross-linking agent, and the proportion of thermosensitive components are converted into digital quantities. The material weight data is collected through a balance sensor with an accuracy of 0.01 kg. The actual value of the ratio is calculated based on the weight data, and the ratio data is filtered using the cumulative average method. The ratio error value is calculated from the filtered data. If the ratio error exceeds 0.5%, the feeding control signal is triggered. The material addition amount is set according to the formula parameters, and the material flow rate is measured using an electromagnetic flowmeter with a range of 0 to 100 kg / hour. The screw feeder is driven by a high-precision servo motor with a speed range of 0 to 1200 rpm. The proportional integral controller is used to output the speed control signal, and the instantaneous flow feedback signal is collected from the electromagnetic flowmeter. The feeding rate is corrected according to the flow feedback data, and the feeder speed is adjusted when the flow fluctuation exceeds the set value by 5%. The density data of the mixed slurry is collected from the online density meter, and the sampling cycle is 1 second. The mixing uniformity is detected in real time by the infrared spectrometer. The sampling position is located at the outlet of the mixer. The temperature sensor is arranged at 4 detection points on the wall of the mixer. The temperature data is collected to judge the mixing state. When the temperature fluctuates by more than 2°C, the stirring time is increased, and the stirring speed is controlled within the range of 300 to 600 rpm. The actual proportion data, mixing uniformity data, and temperature data are written into the industrial controller data register. The feeding equipment is controlled in real time through the adjustment loop. The proportional integral controller is used to output a 0 to 100% control signal with a control accuracy of 0.1%. The opening of the electric control valve is adjusted. The control valve action time is less than 1 second. 4 to 20 mA current signals are used for communication to achieve online adjustment of the formula parameters. The batching control of the EVA foaming material production line has multiple challenges. The sampling cycle of the industrial controller is set to 100 milliseconds based on the material characteristics. The foaming agent and cross-linking agent are liquid at room temperature, and the heat-sensitive component is a solid powder. The material conveying characteristics are significantly different. The accuracy of the balance sensor is 0.01 kg, which meets the requirements of trace addition. In actual production, when the single batch dosage of foaming agent is 50 kg, the cross-linking agent is 15 kg, and the heat-sensitive component is 25 kg, the ratio error is controlled within 0.5%, that is, the foaming agent error is less than 0.25 kg. Different feeding equipment is used in the material addition process. Liquid materials are measured by electromagnetic flowmeters, and solid materials are quantitatively fed by screw feeders. The electromagnetic flowmeter range of 0-100 kg / hour is adapted to different material addition rates. The foaming agent addition rate is 80 kg / hour and the cross-linking agent is 50 kg / hour. The screw feeder speed range of 0-1200 rpm meets the requirements of heat-sensitive component addition. When the addition amount is 60 kg / hour, the speed is maintained at 800 rpm, and precise control is achieved through speed feedback. The density detection of the mixed slurry reflects the degree of mixing uniformity. Under normal working conditions, the density is maintained at 1.2 g / cm3, and the fluctuation range is less than 0.05 g / cm3.The infrared spectrometer detects the mixing uniformity, and the content of each component is judged by the absorption peak intensity. The characteristic peak of the foaming agent is at 1720 wavenumbers, the cross-linking agent is at 1680 wavenumbers, and the thermosensitive component is at 1650 wavenumbers. The temperature sensor is evenly arranged at 4 detection points on the wall of the mixer, and the temperature of the mixing process is controlled within the range of 25±2℃ to avoid pre-reaction of the material. The actuator control adopts a 4-20 mA standard current signal, which has the advantages of strong anti-interference ability and long transmission distance. The electric control valve adopts a linear characteristic, and the opening 0-100% corresponds to a linear relationship of flow 0-100 kg / h. The response time is less than 1 second to meet the needs of rapid adjustment. The proportional integral controller has an output resolution of 0.1%. Under the condition of a foaming agent flow of 80 kg / h, the control accuracy is better than 0.08 kg / h, which meets the process requirements. The actual operation data shows that the material ratio fluctuation is less than 0.3% during the online adjustment of the formula parameters, the mixing uniformity is stable at more than 98%, and the temperature fluctuation is controlled within the range of ±1℃.

[0033] S106, transmitting the optimized and adjusted EVA foam material formula data to the production line control system, automatically adjusting the process parameters of the production line, processing the EVA foam material based on the production line after the process parameters are adjusted to obtain an EVA insole, and heating and bonding the EVA insole and the pre-prepared composite rubber outsole to obtain an EVA outsole.

[0034] The temperature signal of the thermocouple measuring point on the surface of the mold cavity and the position signal of the displacement sensor are obtained, and the corresponding relationship between the foaming ratio and the mold gap is calculated according to the temperature signal and the position signal; the infrared temperature measurement array is used to obtain the foaming molding temperature data, and the pressure sensor is used to obtain the cavity pressure data, and the heating power control signal is output through the proportional integral differential controller according to the temperature data and the pressure data; the preheating temperature data collected by the temperature sensor is received, and the heating plate power is adjusted through the proportional integral controller according to the preheating temperature data, and the distance sensor collects the bonding gap data; according to the bonding process data collected by the pressure sensor array and the temperature sensor array, and the gap data collected by the laser distance sensor, the bonding pressure, temperature and gap are closed-loop controlled through the proportional integral controller.

[0035] Specifically, an industrial controller is used to read the molding process parameters with a sampling period of 100 milliseconds. 12 thermocouple measuring points are arranged on the surface of the mold cavity at a spacing of 50 mm, with a sampling frequency of 10 Hz. The mold position signal is collected through a displacement sensor with an accuracy of 0.01 mm. The real-time foaming ratio data is obtained from the foaming ratio detection unit. The electric cylinder displacement control signal is output according to the corresponding relationship curve between the foaming ratio and the mold gap, and the mold gap is controlled within the range of 0.5 to 5 mm. During the EVA foam insole molding process, the infrared temperature measurement array is arranged with 16 temperature measurement points, with a temperature measurement range of 50 to 200 degrees Celsius and an accuracy of 0.1 degrees Celsius. Four pressure sensors are arranged at the bottom of the mold cavity, with a range of 0 to 2 MPa and a sampling frequency of 100 Hz. The heating power control signal is output through a proportional integral differential controller, with a heating power adjustment range of 0 to 100 kilowatts, a pressure control range of 0.5 to 1.5 MPa, and a holding time of 30 to 180 seconds. It should be noted that the anti-debonding EVA outsole includes an EVA insole and a composite rubber outsole. By weight percentage, the EVA insole includes the following components: 55%-70% ethylene acetate copolymer, 10%-15% PE polyethylene, 10%-20% POE, 4%-8% calcium carbonate, 1.7%-2.1% zinc oxide powder, 1.2%-1.8% flow aid, 2.2%-3% foaming agent, 2%-3% titanium dioxide, 0.5-1% bridging agent, and 7%-18% colorant; the composite rubber outsole includes the following components: 14%-20% butadiene rubber, 20%-25% styrene-butadiene rubber, 20%-23% butyl bromide, 3%-3.5% synthetic rubber, 17%-30% white carbon black, 1.5%-2% petroleum resin, 4.5%-6% active zinc, 2.5-4% active agent, 2%-3% anti-vomiting cream, and 0.5%-0.8% antioxidant. The preferred embodiment of the EVA inner sole includes the following components: 58% ethylene acetate copolymer, 11% PE polyethylene, 11% POE, 4.5% calcium carbonate, 1.8% zinc oxide powder, 1.3% flow aid, 2.3% foaming agent, 2.2% titanium dioxide, 0.9% bridging agent, and 7% colorant. The preferred embodiment of the composite rubber outer sole includes the following components: 18% cis-butadiene rubber, 22% styrene-butadiene rubber, 21.5% bromobutyl, 3.2% synthetic rubber, 22% white carbon black, 1.7% petroleum resin, 5.5% active zinc, 3% active agent, 2.6% anti-spitting cream agent, and 0.5% antioxidant. For the preheating process of the composite rubber outsole, the temperature sensor is arranged at 8 temperature measuring points with a sampling period of 50 milliseconds. The heating plate power is output through the proportional-integral controller to control the preheating temperature within the range of 120 to 160 degrees Celsius. The fitting gap data is collected from the distance measuring sensor with a measurement range of 0 to 10 mm. The servo motor is controlled to drive the screw feed speed within the range of 0.1 to 1 mm / s.During the bonding process of the EVA insole and the rubber outsole, the pressure sensor array is arranged with 9 measuring points at a spacing of 100 mm, the temperature sensor array is arranged with 9 measuring points at a spacing of 100 mm, and the laser distance sensor is arranged with 4 measuring points, with an accuracy of 0.05 mm and a sampling frequency of 50 Hz. The bonding pressure, temperature, and gap are closed-loop controlled by a proportional integral controller, and the bonding pressure is controlled to be 0.8 to 1.2 MPa, the bonding temperature is 140 to 180 degrees Celsius, and the bonding gap is 0.2 to 2 mm. Specifically, the preparation process of the anti-debonding EVA outsole includes the following steps: A1: EVA insole material preparation: different dosage ratios of each item are set according to different physical property requirements, and ethylene acetate copolymer, PE polyethylene, POE, calcium carbonate, zinc oxide powder, flow aid, foaming agent, titanium dioxide, bridging agent, and colorant are put into a mixing machine for mixing, and granulation and cooling are performed after mixing. A2: Foaming: After weighing the weight of each pair of cooled pellets, pour them into the foaming mold and heat them for foaming. After foaming and molding, remove them from the mold, take them out and put them on the cooling rack. After they are completely cooled, they are packaged and sent for surface treatment. A3: Surface treatment: After the foaming, peel the EVA insole and polish the surface cleanly, then blow off the soot and pack them for storage to complete the EVA insole processing. B1: Preparation of composite rubber outsole materials: According to different physical property requirements, different dosage ratios are set, and cis-butyl rubber, styrene-butadiene rubber, bromobutyl, synthetic rubber, white carbon black, petroleum resin, active zinc, activator, anti-spitting frosting agent, and antioxidant are put into the mixing machine for mixing. After mixing, the sheets are cooled. B2: Molding: After the mixed sheets are cut to the appropriate size, they are placed in the corresponding mold for heating and molding, and the excess burrs are corrected. B3: Glue brushing and curing: The molded composite rubber outsole is brushed with glue on the fitting surface, and then dried and cured in a dryer to complete the composite rubber outsole processing. C1: Heating and bonding: The molding operator puts the corresponding EVA insole and composite rubber outsole into the mold for heating and molding. The composite rubber outsole glue is cured for the second time and fits tightly with the EVA insole. C2: Quality inspection: After the finished product is ejected from the mold, it is packaged and put into storage after quality inspection, trimming, sorting, cleaning, and full inspection. The glue in B3 is preferably SA-790H series activated glue or A-790 activated glue. Among them, A-790 or SA-790H series activated glue is a high-performance glue produced using Japanese modified elastomer rubber and plastic raw materials and German synthetic rubber. It is used for the integrated molding of rubber soles and EVA small foams, which has the advantages of simple operation, convenience, labor saving, simplified production process, energy saving and emission reduction. The outsole produced by this process has a clean and beautiful appearance, can ensure good tensile adhesion, and has good heat resistance and aging resistance. SA-790H is a glue for brushing glue in rubber molds. After the rubber is heated and vulcanized, open the mold and directly brush the glue on the rubber sole. After trimming the edges of the brushed rubber sole, it can be bonded together with the EVA small foam.SA-790 is a glue used in conjunction with rubber treatment agent on the upper assembly line. First wash the rubber treatment agent and dry it in the oven before brushing SA-790 glue. After drying, trim the edges and it can be bonded together with EVA small foam. The above two glues are only used for single-sided rubber brushing, saving the EVA surface brushing treatment agent. The two process flows of glue have the advantages of simplifying the process flow, saving labor, energy saving and emission reduction, and reducing VOC emissions and environmental protection. The control of EVA foam material molding process parameters involves multiple dimensions of temperature, pressure, and displacement. Thermocouples are arranged at intervals of 50 mm on the surface of the mold cavity to form a detection array, covering a 300×400 mm molding area, and a total of 12 measuring points to realize temperature field distribution monitoring. The displacement sensor has an accuracy of 0.01 mm to meet the mold gap control requirements. When the foaming ratio reaches 15 times, the mold gap is set to 2 mm. When the foaming ratio is 25 times, the gap increases to 4 mm, realizing dynamic control of the foam material filling process. In the insole molding stage, the infrared temperature measurement array is arranged with 16 measuring points, using a 4×4 matrix arrangement, with a spacing of 80 mm between measuring points, covering an area of ​​320×320 mm. The pressure sensors are arranged at the four corners of the mold, with a range of 2 MPa to meet the pressure monitoring requirements of foam molding. In actual production, the molding pressure is controlled at 1.2 MPa, and the best foaming effect is obtained when the holding time is 120 seconds. The heating power of 100 kW is distributed in 4 heating zones, 25 kW in each zone, and the temperature control accuracy of ±0.5°C is achieved through proportional integral differential control. The preheating temperature control of the composite rubber outsole adopts an 8-point temperature measurement arrangement, which is evenly distributed along the periphery of the outsole, and the temperature measurement frequency is 20 Hz to capture rapid temperature changes. The power of the heating plate is controlled by partitions, with 40 kW in the center area and 20 kW in the edge area to achieve uniform temperature distribution. The fitting gap distance measurement sensors are arranged at the forefoot, mid-heel, and heel positions. When the feed speed is 0.5 mm / s, the fitting process is smooth to avoid bubble generation. The bonding process uses a 9-point pressure and temperature detection array to form a 3×3 matrix, with a measurement point spacing of 100 mm to ensure data representativeness. Laser rangefinders are arranged at the four corners of the bonding surface, and the sampling frequency of 50 Hz monitors the bonding gap in real time. The pressure control uses a proportional integral controller with a proportional coefficient of 0.8 and an integral time of 2 seconds. The control accuracy reaches ±0.05 MPa under a pressure of 1 MPa. When the bonding temperature is 160 degrees Celsius, the pressure is 1 MPa, and the gap is 1 mm, the bonding strength of the inner and outer soles is the highest, and the peel strength reaches 4 N / mm. From the analysis of production data, the uniformity of the mold temperature field affects the product quality most significantly, followed by the uniformity of the pressure distribution, and the impact of the bonding gap fluctuation is relatively small.

[0036] S107. Through data visualization technology, the temperature data, performance prediction values, and formula optimization results of the EVA foaming material are graphically displayed, and a three-dimensional response surface diagram of the foaming temperature, cooling rate, and performance indicators is drawn to intuitively present the nonlinear relationship between the formula parameters and product performance, so that production managers can grasp the production status in real time.

[0037] Temperature data, performance prediction values ​​and recipe parameters are obtained from a real-time database, and the data are extracted according to a preset sampling period; a three-dimensional grid surface is constructed according to the recipe parameters, and a support vector regression predictor is used to calculate the performance indicator values ​​at the grid nodes to obtain the node performance prediction values; the controller data is obtained using an industrial Ethernet communication protocol, and the controller data is normalized to the maximum and minimum values ​​to obtain normalized data; a three-dimensional response surface view is drawn on a graphics workstation according to the normalized data, and the performance indicator values ​​are marked with grayscale gradients, and the optimal recipe parameter positions are determined through the performance indicator values.

[0038] Specifically, the foaming temperature data, performance prediction values, and formula parameters are extracted from the real-time database according to a 1-second sampling cycle. The temperature field distribution in the range of 50 to 200 degrees Celsius is displayed using a red and blue temperature scale. The temperature change curve within 2 hours is drawn in the 600-pixel area on the left side of the 1920×1080 resolution display screen, and the curve data points are spaced 1 second apart. The foaming ratio, bubble diameter, and closed cell ratio numerical bar graphs are displayed in the 400-pixel area on the right side of the screen, and the bar graph refresh cycle is 1 second. According to the formula parameters in the range of 4% to 8% of the foaming agent dosage, 1% to 3% of the cross-linking agent dosage, and 10% to 20% of the thermosensitive component ratio, a 20×20×20 three-dimensional mesh surface is constructed, and the support vector regression predictor is used to generate the performance index values ​​at the mesh nodes. The performance index values ​​are marked with red, green, and blue gradient intervals in the three-dimensional surface diagram, and the gradient band division value is set to 256 levels. The optimal formula parameter corresponding to the surface position is extracted from the predicted data. The industrial Ethernet communication protocol is used to collect field controller data in real time, and the communication baud rate is set to 100 megabits / second.

[0039] Seconds, the collection cycle is 100 milliseconds. 4096 sets of temperature data, performance prediction values, and formula parameters are temporarily stored in the data buffer area. After the maximum and minimum values ​​are normalized and mapped to the range of 0 to 1, the data display refresh rate is set to 60 Hz, and the display delay is less than 16 milliseconds. A 800×600 pixel three-dimensional response surface view is drawn on the graphics workstation. The coordinate axis scale interval is set to 0.5% foaming agent, 0.2% cross-linking agent, and 1% thermosensitive component. The surface grid unit size is 0.1%×0.1%×0.1%. 256 levels of grayscale gradient are used to identify the performance index value. The graphical interface resolution is set to 2560×1600 pixels, and the graphic refresh rate is 60 Hz. In the EVA foaming data visualization design, the sampling cycle of 1 second matches the temperature change rate of the foaming process. The temperature data drops from 160°C to 80°C for 300 seconds. The data point interval of 1 second can fully record the temperature change characteristics. The red and blue temperature scale uses a gradient color band, with red in the high temperature area, blue in the low temperature area, and a gradient purple in the middle temperature, which intuitively reflects the temperature distribution. The 2-hour temperature curve contains 7,200 data points. In the 600-pixel width area, each data point occupies 0.083 pixels, and the curve is continuous and smooth. The formulation parameter response surface is constructed using a 20×20×20 grid division, forming a grid point with a 0.2% interval in the range of 4%-8% of the foaming agent dosage, a grid point with a 0.1% interval in the range of 1%-3% of the cross-linking agent dosage, and a grid point with a 0.5% interval in the range of 10%-20% of the thermosensitive component ratio. The support vector regression predictor predicts the performance of 8,000 grid nodes and generates a continuous and smooth response surface. The performance indicators are marked with red, green and blue colors, with red representing a foaming ratio of 15-25 times, green representing a closed cell ratio of 85%-95%, and blue representing a compressive strength of 0.6-1.0MPa. Data acquisition uses industrial Ethernet 100Mbps bandwidth to meet the real-time data transmission requirements under a 100ms sampling period. 4096 sets of data cache correspond to a 409.6-second production process. The temperature data is normalized to map 50-200℃ to the 0-1 interval, and the performance prediction value is normalized to map the foaming ratio 15-25 times to the 0-1 interval. The display delay per frame is 16.7ms at a 60Hz refresh rate, and the human eye observes that the data update is continuous, smooth and flicker-free. The three-dimensional response surface is displayed in an 800×600 pixel view, and the coordinate axis scale intervals are selected as 0.5% foaming agent, 0.2% cross-linking agent, and 1% heat-sensitive component for easy reading.

[0040] The 0.1%×0.1%×0.1% grid unit has a projection size of about 4×3 pixels in the view, which meets the needs of surface detail display. The 256-level grayscale gradient can clearly distinguish the 0.4% change in performance indicators, and the graphic details are clear on a 2560×1600 high-resolution display. Actual operation data shows that when the foaming temperature is 175°C, the amount of foaming agent is 6%, the amount of cross-linking agent is 2%, and the heat-sensitive component is 15%, the response surface presents a better performance area.

[0041] The above is only a specific implementation method of this specification. Those skilled in the art can clearly understand that for the convenience and simplicity of description, the specific working process of the system, module and unit described above can refer to the corresponding process in the aforementioned method embodiment, and will not be repeated here. It should be understood that the protection scope of this specification is not limited to this. Any technician familiar with the technical field can easily think of various equivalent modifications or replacements within the technical scope disclosed in this specification, and these modifications or replacements should be covered within the protection scope of this specification.

Claims

1. A digital batching method for the production of anti-degumming EVA outsole, characterized in that: The method comprises: By installing temperature sensors on the production line, the temperature data of the EVA foam material is collected in real time, and the temperature data includes the foaming temperature and the cooling rate. According to the different formula components of the EVA foam material, an orthogonal experimental scheme is designed. By changing the proportion of the formula components, the performance test of the EVA foam material is carried out under different foaming temperatures and cooling rates, and the foaming performance, mechanical properties and bonding performance index data are obtained, and an experimental data set including temperature, formula components and performance indicators is established; The support vector machine regression algorithm is used to train the EVA foam material performance prediction model with foaming temperature, cooling rate and formula ingredients as input and performance indicators as output. The real-time collected temperature data is input into the trained performance prediction model to obtain the predicted values ​​of the foaming performance, mechanical properties and bonding properties of the EVA foam material under the current temperature conditions. Determine whether the predicted values ​​of the foaming performance, mechanical properties and adhesive properties of the EVA foam material exceed the preset performance threshold range. If they exceed the threshold range, trigger the formula optimization program to optimize and adjust the amount of foaming agent, the amount of cross-linking agent and the proportion of heat-sensitive components in the EVA foam material formula; In the formula optimization program, a genetic algorithm is used to optimize the EVA foam material formula, with the predicted values ​​of foaming performance, mechanical properties and adhesive properties as the fitness function. At the same time, the formula cost and process adaptability are combined, and the optimal combination of foaming agent dosage, crosslinking agent dosage and heat-sensitive component ratio is iteratively searched through selection, crossover and mutation genetic operations; The optimal amount of foaming agent, cross-linking agent and heat-sensitive component ratio combination data obtained by genetic algorithm search is transmitted to the batching control system of the EVA foam material production line, and the addition ratio of foaming agent, cross-linking agent and heat-sensitive component is adjusted to achieve online optimization control of the formula; The optimized and adjusted EVA foam material formula data is transmitted to the production line control system, and the process parameters of the production line are automatically adjusted. The EVA foam material is processed based on the production line after the process parameters are adjusted to obtain an EVA insole, and the EVA insole is heated and laminated with a pre-prepared composite rubber outsole to obtain an EVA outsole; Through data visualization technology, the temperature data, performance prediction values, and formula optimization results of EVA foaming materials are graphically displayed, and a three-dimensional response surface diagram of foaming temperature, cooling rate, and performance indicators is drawn to intuitively present the nonlinear relationship between formula parameters and product performance, so that production managers can grasp the production status in real time.

2. The method according to claim 1, characterized in that The temperature data of the EVA foam material is collected in real time by installing a temperature sensor on the production line. The temperature data includes the foaming temperature and the cooling rate. According to the different formula components of the EVA foam material, an orthogonal experimental scheme is designed. By changing the proportion of the formula components, the performance test of the EVA foam material is carried out under different foaming temperatures and cooling rates, and the foaming performance, mechanical properties and bonding performance index data are obtained. An experimental data set including temperature, formula components and performance indicators is established, including: Collecting temperature data of the EVA foam material from a temperature sensor array, and obtaining a cooling rate value according to the temperature data by using a temperature gradient calculation formula; According to the raw material ratio, the compressive strength, compression permanent deformation rate and shear strength of the foaming material are measured by a universal testing machine, and the bubble morphology and distribution data of the foaming material are obtained by scanning electron microscopy; A multivariate linear regression tool is used to perform correlation calculation on the temperature data and the performance index data, and a mapping relationship between the foaming temperature and the performance index is obtained according to the correlation calculation result; The recipe components and temperature parameters are trained using a fuzzy neural network tool, the temperature parameter threshold is determined based on the training results, and the heating device and the cooling device are adjusted based on the temperature parameter threshold.

3. The method according to claim 1, characterized in that The support vector machine regression algorithm is used, with foaming temperature, cooling rate and formula ingredients as input, and performance indicators as output, to train the EVA foam material performance prediction model, input the real-time collected temperature data into the trained performance prediction model, and obtain the predicted values ​​of the foaming performance, mechanical properties and bonding properties of the EVA foam material under the current temperature conditions, including: According to the temperature sensor sampling data, the mean and standard deviation of the foaming temperature data and the cooling rate data are calculated by using a moving window, and the data in the temperature data that exceeds the standard deviation range is subjected to spline interpolation to obtain the corrected temperature data; The corrected temperature data is processed by a maximum and minimum value normalization method, and foaming performance indexes and mechanical performance indexes are extracted from the normalized data to obtain a performance index data set; The kernel parameters and penalty parameters of the Gaussian kernel function of the support vector machine are traversed through the grid search tool, and the root mean square error value is calculated in the cross-validation to obtain the optimal parameter combination; The optimal parameter combination is used to construct a support vector machine regression predictor, performance prediction is performed on the validation set and a root mean square error is calculated, a correction coefficient is obtained through the root mean square error, and the correction coefficient is used to calibrate the prediction result.

4. The method according to claim 1, characterized in that: The method of determining whether the predicted values ​​of the foaming performance, mechanical properties and adhesive properties of the EVA foam material exceed a preset performance threshold range, and if so, triggering a formula optimization program to optimize and adjust the amount of the foaming agent, the amount of the cross-linking agent and the ratio of the heat-sensitive component in the EVA foam material formula, including: The performance data is processed using a sliding window numerical filter, and a flag is generated for performance indicators that exceed a preset threshold range, wherein the flag value is a positive deviation value or a negative deviation value; According to the deviation value of the performance index, the formula data is obtained from the historical database, the correlation between the formula components and the performance index is obtained by the Pearson correlation coefficient, and the sensitivity of the formula components to the performance index is obtained by multiple linear regression; The genetic algorithm parameters are set according to the sensitivity of the formula components, the formula parameters are optimized and calculated, and the optimized formula with the highest fitness is obtained through verification by preset thresholds; The industrial control unit is used to write the optimized formula parameters into the batching device database, and the real-time feeding data is collected through the quality sensor. If the deviation between the feeding data and the set value exceeds the threshold, the flow data is obtained from the proportional control valve to adjust the flow controller output.

5. The method according to claim 1, characterized in that In the formula optimization program, a genetic algorithm is used to optimize the EVA foam material formula, and the predicted values ​​of foaming performance, mechanical properties and adhesive properties are used as fitness functions. At the same time, the formula cost and process adaptability are combined, and the optimal combination of foaming agent dosage, crosslinking agent dosage and heat-sensitive component ratio is iteratively searched through selection, crossover and mutation genetic operations, including: The weighted summation method is used to construct the recipe fitness function, and the recipe score is calculated based on the performance index weights and performance test values; Obtain component dosage intervals from the process database and generate recipe parameter gene sequences through binary coding; Calculating the selection probability according to the recipe score, using a roulette wheel selection operator to obtain recipe individuals, and performing a single-point crossover operation and a non-uniform mutation operation on the recipe individuals to obtain a new generation recipe population; If the continuous change range of the fitness value of the best individual in the group is less than the preset threshold, the best individual is decoded to obtain the recipe parameters, and the recipe parameters are judged through process verification to determine whether they meet the process requirements. If they meet the process requirements, it is determined to be the optimal recipe.

6. The method according to claim 1, characterized in that The optimal amount of foaming agent, the amount of cross-linking agent and the ratio of heat-sensitive components obtained by searching with the genetic algorithm are transmitted to the batching control system of the EVA foam material production line, and the addition ratio of the foaming agent, the cross-linking agent and the heat-sensitive components is adjusted to realize the online optimization control of the formula, including: Collect material weight data according to the optimized formula parameters received by the industrial controller, the material weight data is obtained by a balance sensor and filtered proportion data is obtained by cumulative averaging method; The filtered proportioning data is used to control the electromagnetic flowmeter to measure the material flow rate, and the material flow rate data is used to drive the servo motor to adjust the speed of the screw feeder; The mixed slurry output from the screw feeder is detected by a density meter and a spectrometer, and the mixing state is determined by the temperature data collected by the mixer wall temperature sensor; A proportional-integral control signal is output to the electric regulating valve according to the mixing state data. The electric regulating valve adjusts the opening according to the control signal, and the material formula is adjusted online through the opening data.

7. The method according to claim 1, characterized in that The optimized and adjusted EVA foam material formula data is transmitted to a production line control system, the process parameters of the production line are automatically adjusted, the EVA foam material is processed based on the production line after the process parameters are adjusted to obtain an EVA insole, and the EVA insole is heated and laminated with a pre-prepared composite rubber outsole to obtain an EVA outsole, including: Acquire the temperature signal of the thermocouple measuring point on the surface of the mold cavity and the position signal of the displacement sensor, and calculate the corresponding relationship between the foaming ratio and the mold gap according to the temperature signal and the position signal; An infrared temperature measurement array is used to obtain foaming molding temperature data, and a pressure sensor is used to obtain cavity pressure data. A heating power control signal is output through a proportional-integral-differential controller according to the temperature data and pressure data. Receiving preheating temperature data collected by the temperature sensor, adjusting the power of the heating plate through a proportional integral controller according to the preheating temperature data, and collecting the laminating gap data through a distance measuring sensor; Based on the bonding process data collected by the pressure sensor array and the temperature sensor array, and the gap data collected by the laser ranging sensor, a proportional-integral controller is used to perform closed-loop control of the bonding pressure, temperature and gap.

8. The method according to claim 1, characterized in that The temperature data, performance prediction value, and formula optimization results of the EVA foaming material are graphically displayed through data visualization technology, and a three-dimensional response surface diagram of the foaming temperature, cooling rate, and performance index is drawn to intuitively present the nonlinear relationship between the formula parameters and the product performance, so that the production management personnel can grasp the production status in real time, including: Acquire temperature data, performance prediction values ​​and recipe parameters from a real-time database, wherein the data is extracted according to a preset sampling period; Constructing a three-dimensional mesh surface according to the recipe parameters, and using a support vector regression predictor to calculate the performance index values ​​at the mesh nodes to obtain the node performance prediction values; The controller data is acquired by adopting the industrial Ethernet communication protocol, and the maximum and minimum values ​​of the controller data are normalized to obtain normalized data; A three-dimensional response surface view is drawn on a graphic workstation according to the normalized data, the performance index value is marked with grayscale gradient, and the optimal recipe parameter position is determined by the performance index value.

Images