Method and device for predicting preparation process of biaxially oriented polytetrafluoroethylene fiber membrane

Abstract

The invention discloses a method and a device for predicting a preparation process of a biaxially-oriented polytetrafluoroethylene fiber membrane, and relates to the field of process prediction. In the method, a data set of multiple groups of fiber membrane samples prepared through a preset orthogonal experiment is obtained, and the data set comprises a two-way stretching process parameter combination, pore structure data and performance indexes for preparing a target group of fiber membrane samples; constructing an initial machine learning model, taking the two-way stretching process parameter combination in the data set as an input parameter, taking the pore structure data and the performance index as output parameters, training the initial machine learning model to obtain an intermediate model, and adjusting the parameters of the intermediate model through cross validation to obtain a machine learning model; and according to the set target performance, recommending an optimal process parameter combination and a confidence interval through a machine learning model. By implementing the technical scheme provided by the invention, the purpose of efficiently predicting the preparation process of the biaxially oriented polytetrafluoroethylene fiber membrane is achieved.

Description

Technical Field

[0001] The present application relates to the technical field of process prediction, and in particular to a method and device for predicting a process for preparing a biaxially oriented polytetrafluoroethylene fiber membrane. Background Art

[0002] Polytetrafluoroethylene (PTFE) is widely used in filter materials, medical materials, electronic devices and environmental protection fields due to its excellent chemical stability, low friction coefficient, high temperature resistance and excellent dielectric properties. Among them, the biaxial stretching process is the core technology for preparing high-performance PTFE fiber membranes. Through synchronous or step-by-step longitudinal and transverse stretching, the macroscopic structure, pore structure and various properties of the fiber membrane can be adjusted to meet the needs of different application scenarios.

[0003] The preparation process of biaxially stretched PTFE membrane involves multi-parameter coupling, including longitudinal and transverse stretching temperature, stretching rate, stretching ratio, heat setting conditions and other process parameters. There is a complex nonlinear relationship between these parameters and the structure (such as porosity, pore size distribution, etc.) and performance (air permeability, filtration efficiency, breaking strength, etc.) of the final product. Therefore, it is a complex and difficult process to obtain a combination of specific structure and performance. The existing biaxially stretched polytetrafluoroethylene fiber membrane preparation process requires many (often more than hundreds of) experiments to determine the required multiple stretching preparation process parameters, which is not only time-consuming and labor-intensive, but also has a large experimental cost and resource waste, and cannot ensure that the prepared material has the best combination of specific structure and performance.

[0004] Therefore, a method is needed to predict the preparation process of biaxially oriented polytetrafluoroethylene fiber membrane according to target performance. Summary of the invention

[0005] The present application provides a method and device for predicting the preparation process of biaxially stretched polytetrafluoroethylene fiber membrane, which achieves the purpose of efficiently predicting the preparation process of biaxially stretched polytetrafluoroethylene fiber membrane, significantly improves the accuracy and stability of process parameter selection, and reduces the trial and error cost, providing reliable support for the development of high-performance fiber membranes.

[0006] In a first aspect of the present application, a method for predicting a preparation process of a biaxially oriented polytetrafluoroethylene fiber membrane is provided, which is applied to a preparation platform, and the method comprises: Acquire a data set of multiple groups of fiber membrane samples prepared by a preset orthogonal experiment, wherein the data set includes a combination of biaxial stretching process parameters, pore structure data, and performance indicators for preparing a target group of fiber membrane samples, wherein the target group of fiber membrane samples is any one group of the multiple groups of fiber membrane samples; Constructing an initial machine learning model, taking the combination of biaxial stretching process parameters in the data set as input parameters, taking the pore structure data and performance index as output parameters, training the initial machine learning model to obtain an intermediate model, and adjusting the parameters of the intermediate model through cross-validation to obtain a machine learning model; According to the set target performance, the optimal process parameter combination and confidence interval are recommended by the machine learning model.

[0007] Optionally, constructing the initial machine learning model includes embedding a physical constraint mechanism, specifically including: Calculating the critical speed of polytetrafluoroethylene according to preset viscoelastic parameters, setting the critical speed between the input layer and the hidden layer, and triggering parameter correction when the input stretching speed exceeds the critical speed; A porosity restriction function is set in the output layer. When the porosity predicted by the model exceeds the interval set by the porosity restriction function, the output result is corrected according to a preset adjustment strategy.

[0008] Optionally, the recommending the optimal process parameter combination and confidence interval by the machine learning model includes: Determining the range and step size of biaxial stretching process parameters, wherein the biaxial stretching process parameters include transverse stretching multiple, transverse speed, transverse temperature, longitudinal stretching multiple, longitudinal speed and longitudinal temperature; Determine a parameter space according to the range, and uniformly generate a plurality of parameter combinations in the parameter space using a Latin hypercube sampling method; The crossover probability, mutation probability, elite retention ratio and crowding degree calculation method are set, and the optimal solution is obtained from the plurality of parameter combinations through non-dominated sorting and crowding distance calculation.

[0009] Optionally, the recommending the optimal process parameter combination and confidence interval by the machine learning model further includes: Introducing random disturbances to each of the biaxial stretching process parameters, running the machine learning model multiple times, and obtaining performance index results under different disturbances; The coefficients of variation of air permeability and filtration efficiency are calculated according to the performance indicator results, and a process solution whose coefficients of variation of air permeability and filtration efficiency are both lower than a first preset threshold is screened out.

[0010] Optionally, the method further includes: The measured data of the samples prepared according to the optimal process parameter combination are subjected to deviation analysis with the predicted values. When the absolute error of the porosity or the error of the filtration efficiency exceeds the second preset threshold, the transfer learning strategy is used to freeze the underlying parameters related to the basic feature extraction in the machine learning model, and the preset fully connected layer is adjusted according to the measured data of the samples.

[0011] Optionally, the using of the transfer learning strategy to freeze the underlying parameters related to basic feature extraction in the machine learning model and adjusting the preset fully connected layer according to the sample measured data includes: Freeze the weight matrix and convolution kernel parameters of the feature extraction layer in the initial machine learning model, and when the loss function decrease rate of three consecutive rounds of training is less than a preset percentage, multiply the current learning rate of the fully connected layer by the attenuation coefficient.

[0012] Optionally, the step of obtaining a data set of multiple groups of fiber membrane samples prepared by a preset orthogonal experiment includes: The polytetrafluoroethylene dispersion resin and the hydrocarbon lubricating oil are mixed in a preset weight ratio, and the aging conditions are 30-70°C and 24-48 hours; the undeoiled base belt is formed by billet making, pushing extrusion and calendering; Control the longitudinal stretching multiple to 1-16 times, the longitudinal speed to 1-5 m / min, the longitudinal temperature to 120-330° C., the transverse expansion multiple to 3-25 times, the transverse speed to 1-16 m / min, and the transverse temperature to 210-320° C. to generate multiple groups of fiber membrane samples; The pore structures and properties of multiple groups of the fiber membrane samples are detected to obtain the data set.

[0013] In a second aspect of the present application, a system for predicting a process for preparing a biaxially oriented polytetrafluoroethylene fiber membrane is provided, comprising a data module, a model module and a prediction module, wherein: A data module is configured to obtain a data set of multiple groups of fiber membrane samples prepared by a preset orthogonal experiment, wherein the data set includes a combination of biaxial stretching process parameters, pore structure data, and performance indicators for preparing a target group of fiber membrane samples, wherein the target group of fiber membrane samples is any one of the multiple groups of fiber membrane samples; A model module is configured to construct an initial machine learning model, use the combination of biaxial stretching process parameters in the data set as input parameters, use the pore structure data and performance indicators as output parameters, train the initial machine learning model to obtain an intermediate model, and adjust the parameters of the intermediate model through cross-validation to obtain a machine learning model; The prediction module is configured to recommend an optimal process parameter combination and a confidence interval through the machine learning model according to a set target performance.

[0014] In the third aspect of the present application, an electronic device is provided, including a processor, a memory, a user interface and a network interface, the memory is used to store instructions, the user interface and the network interface are both used to communicate with other devices, and the processor is used to execute the instructions stored in the memory so that the electronic device executes any one of the methods described above.

[0015] In a fourth aspect of the present application, a computer-readable storage medium is provided, wherein the computer-readable storage medium stores instructions, and when the instructions are executed, any of the methods described above is executed.

[0016] In summary, one or more technical solutions provided in the embodiments of the present application have at least the following technical effects or advantages: 1. By obtaining data sets of multiple groups of fiber membrane samples, including information such as biaxial stretching process parameter combinations, pore structure data, and performance indicators, a wealth of training data is provided for the machine learning model. These data cover the preparation of fiber membranes under different process conditions, enabling the model to learn the complex relationship between process parameters and pore structure and performance indicators, thereby improving the accuracy of prediction; 2. Using the cross-validation method to adjust the parameters of the intermediate model can effectively avoid the problem of overfitting or underfitting of the model. By dividing the data set into multiple subsets and training and validating them in turn, the performance of the model can be more comprehensively evaluated, the optimal parameter combination can be found, and the prediction accuracy of the machine learning model can be further improved; 3. Based on the trained machine learning model, the optimal process parameter combination and confidence interval can be quickly recommended according to the set target performance. Compared with the traditional trial and error method or empirical method, this method can select the solution that best meets the target performance requirements from a large number of possible process parameter combinations in a short period of time, greatly shortening the time and cost of process optimization and improving production efficiency; 4. Through the prediction results of the model, the trend and degree of the influence of process parameters on the performance of fiber membranes can be clarified, providing strong guidance for subsequent experimental design. Researchers can conduct targeted experimental verification and further optimization based on the optimal process parameter combination and its confidence interval recommended by the model, avoiding the waste of resources and time delays caused by blind experiments; 5. This method can achieve precise control of the pore structure data and performance indicators of the fiber membrane by accurately predicting and optimizing the biaxial stretching process parameters. The stability and consistency of key performance indicators such as porosity, air permeability, and filtration efficiency are improved, thereby ensuring the quality and performance stability of the fiber membrane product and meeting the strict requirements of fiber membrane performance in different application scenarios; 6. In the actual production process, various factors may cause fluctuations in process parameters, thereby affecting product quality. This method uses a machine learning model to model and predict the relationship between process parameters and product performance, which can timely detect abnormal changes in process parameters and compensate for the impact of these changes on product quality by adjusting process parameters, thereby reducing the adverse effects of process fluctuations on product quality and improving the stability of product quality. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] Figure 1 It is a schematic flow chart of a method for predicting a process for preparing a biaxially oriented polytetrafluoroethylene fiber membrane disclosed in an embodiment of the present application; Figure 2 It is a schematic diagram of the structure of the core machine learning algorithm disclosed in the embodiment of the present application; Figure 3 It is a module schematic diagram of a system for predicting a process for preparing a biaxially oriented polytetrafluoroethylene fiber membrane disclosed in an embodiment of the present application; Figure 4 It is a structural schematic diagram of an electronic device disclosed in an embodiment of the present application.

[0018] Explanation of the reference numerals: 301, data module; 302, model module; 303, prediction module; 401, processor; 402, communication bus; 403, user interface; 404, network interface; 405, memory. DETAILED DESCRIPTION

[0019] In order to enable technicians in this field 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 application, not all of the embodiments.

[0020] In the description of the embodiments of the present application, words such as "for example" or "for example" are used to indicate examples, illustrations or explanations. Any embodiment or design described as "for example" or "for example" in the embodiments of the present application should not be interpreted as being more preferred or more advantageous than other embodiments or designs. Specifically, the use of words such as "for example" or "for example" is intended to present related concepts in a specific way.

[0021] In the description of the embodiments of the present application, the meaning of the term "multiple" refers to two or more. For example, multiple systems refer to two or more systems, and multiple screen terminals refer to two or more screen terminals. In addition, the terms "first" and "second" are used for descriptive purposes only and cannot be understood as indicating or implying relative importance or implicitly indicating the indicated technical features. Thus, the features defined as "first" and "second" may explicitly or implicitly include one or more of the features. The terms "include", "comprise", "have" and their variations all mean "including but not limited to", unless otherwise specifically emphasized.

[0022] This embodiment discloses a method for predicting the preparation process of biaxially stretched polytetrafluoroethylene fiber membrane, which is applied to the preparation platform. Figure 1 : is a schematic flow chart of a method for predicting a process for preparing a biaxially oriented polytetrafluoroethylene fiber membrane disclosed in an embodiment of the present application, such as Figure 1As shown, the method comprises the following steps: S101, obtaining a data set of multiple groups of fiber membrane samples prepared by a preset orthogonal experiment, wherein the data set includes a combination of biaxial stretching process parameters, pore structure data, and performance indicators for preparing a target group of fiber membrane samples, wherein the target group of fiber membrane samples is any group of the multiple groups of fiber membrane samples; S102, constructing an initial machine learning model, taking the combination of biaxial stretching process parameters in the data set as input parameters, taking the pore structure data and performance index as output parameters, training the initial machine learning model to obtain an intermediate model, and adjusting the parameters of the intermediate model through cross-validation to obtain a machine learning model; S103. According to the set target performance, the optimal process parameter combination and confidence interval are recommended by the machine learning model.

[0023] Multiple groups of fiber membrane samples are prepared by pre-setting orthogonal experiments. Orthogonal experiments are a statistical experimental design method that can reasonably arrange experiments and effectively analyze the impact of multiple factors on experimental results. In this step, the prepared fiber membrane sample data set contains the following important information: Biaxial stretching process parameter combination: This is a key factor affecting the performance of the fiber membrane, including transverse stretching multiple, transverse speed, transverse temperature, longitudinal stretching multiple, longitudinal speed and longitudinal temperature. Different parameter combinations will lead to differences in the orientation of the molecular chain and the formation of the pore structure of the fiber membrane during the stretching process, thus affecting its performance.

[0024] Pore ​​structure data: Pore structure is one of the important characteristics of fiber membranes, including porosity, pore size distribution, etc. Porosity reflects the proportion of pores in the fiber membrane, and pore size distribution describes the size range and number distribution of pores. These data can be measured by scanning electron microscopy (SEM), nitrogen adsorption-desorption isotherms, etc.

[0025] Performance indicators: Performance indicators are used to evaluate the quality and application performance of fiber membranes. Common performance indicators include air permeability, filtration efficiency, tensile strength, elongation at break, etc. Air permeability reflects the gas permeability of the fiber membrane, filtration efficiency indicates the filtration effect of the fiber membrane on particulate matter, and tensile strength and elongation at break reflect the mechanical properties of the fiber membrane.

[0026] Among these fiber membrane samples, any group of fiber membrane samples can be used as the target group, and the corresponding biaxial stretching process parameter combination, pore structure data and performance index will be used as the basic data for the subsequent construction of the machine learning model. Select a suitable machine learning algorithm to build the model, such as neural network, support vector machine, decision tree, etc. The biaxial stretching process parameter combination is used as the input parameter. These parameters are controllable factors affecting the performance of the fiber membrane. By adjusting them, the preparation process of the fiber membrane can be changed. The pore structure data and performance index are used as output parameters. They are the final performance of the fiber membrane and are the targets to be optimized by adjusting the process parameters. The initial machine learning model is trained using the acquired data set. During the training process, the model will learn the mapping relationship between the corresponding pore structure data and performance index according to the input process parameters. By continuously adjusting the internal parameters of the model, the prediction results of the model are made as close as possible to the actual data, thereby obtaining an intermediate model. In order to evaluate the performance of the model and avoid overfitting or underfitting, the cross-validation method is used. Common cross-validation methods include k-fold cross-validation (k is a positive integer greater than or equal to 2), which is to divide the data set into k subsets, select one of the subsets as the validation set each time, and the remaining k-1 subsets as the training set, and repeat the training and validation process k times. According to the performance indicators on the validation set (such as mean square error, accuracy, etc.), adjust the parameters of the intermediate model, such as the number of layers of the neural network, the number of neurons, the learning rate, etc., to improve the generalization ability and prediction accuracy of the model, and finally obtain the optimized machine learning model. According to the set target performance, such as requiring the fiber membrane to have high air permeability and filtration efficiency, or within a certain range of mechanical properties, these target performances are input into the trained machine learning model. The model will predict the optimal combination of process parameters that can achieve the target performance based on the relationship between the process parameters and performance indicators learned previously. At the same time, in order to evaluate the reliability of the prediction results, the corresponding confidence interval will also be given. The confidence interval indicates the range of values ​​of the performance indicators corresponding to the predicted optimal process parameter combination under a certain confidence level. It provides a reference for process control in actual production and helps operators to reasonably adjust the process parameters within the confidence interval to ensure that the performance of the fiber membrane meets the set requirements.

[0027] Optionally, the step of obtaining a data set of multiple groups of fiber membrane samples prepared by a preset orthogonal experiment includes: The polytetrafluoroethylene dispersion resin and the hydrocarbon lubricating oil are mixed in a preset weight ratio, and the aging conditions are 30-70°C and 24-48 hours; the undeoiled base belt is formed by billet making, pushing extrusion and calendering; Control the longitudinal stretching multiple to 1-16 times, the longitudinal speed to 1-5 m / min, the longitudinal temperature to 120-330° C., the transverse expansion multiple to 3-25 times, the transverse speed to 1-16 m / min, and the transverse temperature to 210-320° C. to generate multiple groups of fiber membrane samples; The pore structures and properties of multiple groups of the fiber membrane samples are detected to obtain the data set.

[0028] The polytetrafluoroethylene dispersion resin (crystallinity ≥ 97%, number average molecular weight 2 million to 20 million) is mixed with hydrocarbon lubricating oil in a preset weight ratio of 1: (0.10 to 0.35). This mixing ratio is carefully designed to make the polytetrafluoroethylene dispersion resin have good fluidity and plasticity in the subsequent processing, and the hydrocarbon lubricating oil can play a role in lubrication and improving processing performance. The mixed material needs to be aged at 30 to 70 ° C for 24 to 48 hours. The aging process is to allow the components in the mixture to fully interact with each other to form a uniform and stable system, so as to prepare for the subsequent billet making, push extrusion and calendering. The matured material is successively subjected to billet making, push extrusion and calendering to finally form an undeoiled base belt. The billet making process is carried out at a temperature of 30 to 50 ° C, and the pressure needs to be maintained for 3 to 10 minutes. The purpose of this step is to form a billet with a certain shape and density, which provides a basis for the subsequent push extrusion. The push extrusion process is carried out at a temperature of 30 to 60 ° C, and the compression ratio is controlled between 50 and 300. The push extrusion can further homogenize the material and form a continuous green body with a certain size and shape. The final calendering treatment is to make the thickness of the green body more uniform and form an undeoiled base belt that meets the subsequent biaxial stretching requirements. In the orthogonal experimental design of biaxial stretching, the longitudinal stretching process parameters are set as follows: the longitudinal stretching multiple is set in the range of 1 to 16 times, the stretching speed is controlled at 1 to 5 m / min, and the stretching temperature is adjusted between 120 and 330 ° C. Different combinations of stretching multiples, speeds and temperatures will have a significant effect on the longitudinal properties of polytetrafluoroethylene fiber membranes. For example, the greater the stretching multiple, the higher the degree of orientation of the fiber membrane in the longitudinal direction, which may lead to an increase in longitudinal strength, but it may also affect other properties such as its flexibility; changes in stretching speed and temperature will affect the movement and rearrangement of the molecular chain, thereby affecting the microstructure and macroscopic properties of the fiber membrane. The transverse expansion process parameters are set as follows: transverse expansion multiples of 3 to 25 times, transverse speed of 1 to 16 m / min, and transverse temperature of 210 to 320 °C. The transverse expansion multiple determines the degree of dimensional change of the fiber membrane in the transverse direction, and together with the longitudinal stretching multiple, affects the overall pore structure and performance of the fiber membrane; the transverse speed and temperature also have an important impact on the transverse performance of the fiber membrane, and cooperate with the longitudinal stretching process parameters. Through the cross-combination of transverse and longitudinal parameters, polytetrafluoroethylene fiber membrane samples with different pore structures and performances can be prepared. According to the parameter range in the above orthogonal experimental design, multiple groups of fiber membrane samples are generated by controlling the longitudinal stretching multiple, longitudinal speed, longitudinal temperature, transverse expansion multiple, transverse speed and transverse temperature. In the process of preparing each group of fiber membrane samples, the preset process parameters are strictly followed to ensure that the sample preparation conditions are accurate.For the prepared groups of fiber membrane samples, advanced testing methods are used to test their pore structure (including porosity, pore size, etc.) and performance (air permeability, filtration efficiency, etc.). The porosity test can be determined by methods such as nitrogen adsorption-desorption isotherms, and the pore size measurement can be achieved using scanning electron microscopy (SEM) and other technologies; the air permeability test is usually carried out using a special air permeability tester in accordance with standard test methods; the filtration efficiency test requires the selection of appropriate particles or gases for filtration experiments based on the application scenario of the fiber membrane, and the filtration efficiency is measured using professional testing equipment.

[0029] Mixing polytetrafluoroethylene dispersion resin and hydrocarbon lubricating oil in a preset weight ratio and aging them at 30-70°C for 24-48 hours can fully blend the two to form a uniform and stable mixed system. This helps to ensure the consistency of the fluidity and plasticity of the material during the subsequent billet making, push extrusion and calendering process, thereby forming an undeoiled base belt with uniform quality and stable performance, laying a good foundation for the biaxial stretching process. By controlling the billet making temperature at 30-50°C, the holding time at 3-10 minutes, the push extrusion temperature at 30-60°C, and the compression ratio at 50-300, the formation process of the undeoiled base belt can be accurately controlled. This is conducive to improving the density and uniformity of the base belt, reducing internal defects and stress concentration, so that it can better withstand the tensile force during the subsequent biaxial stretching process, and avoiding stretching failure or unstable fiber membrane performance due to base belt quality problems. The parameter ranges of longitudinal stretching multiples of 1 to 16 times, longitudinal speeds of 1 to 5 m / min, longitudinal temperatures of 120 to 330 °C, and transverse expansion multiples of 3 to 25 times, transverse speeds of 1 to 16 m / min, and transverse temperatures of 210 to 320 °C are set, providing a wealth of process options for the preparation of fiber membranes. Different combinations of stretching multiples, speeds, and temperatures will have a significant impact on the pore structure and properties of polytetrafluoroethylene fiber membranes, and can meet the diverse requirements of fiber membrane performance in different application scenarios. Due to the orthogonal experimental design and a wide range of stretching parameters, the generated data sets have good diversity and representativeness. This enables the machine learning model to effectively learn and train under different process conditions, avoiding the problem of overfitting or underfitting of the model due to single or limited data, ensuring that the model can accurately predict the performance of the fiber membrane under various combinations of process parameters, and providing a reliable basis for process optimization and product quality control in actual production.

[0030] Optionally, constructing the initial machine learning model includes embedding a physical constraint mechanism, specifically including: Calculating the critical speed of polytetrafluoroethylene according to preset viscoelastic parameters, setting the critical speed between the input layer and the hidden layer, and triggering parameter correction when the input stretching speed exceeds the critical speed; A porosity restriction function is set in the output layer. When the porosity predicted by the model exceeds the interval set by the porosity restriction function, the output result is corrected according to a preset adjustment strategy.

[0031] The critical speed of polytetrafluoroethylene is calculated according to the preset viscoelastic parameters. As a viscoelastic polymer material, the movement and deformation behavior of its molecular chain during the stretching process is restricted by the viscoelastic properties. Through the preset viscoelastic parameters, such as relaxation time, modulus, etc., the critical speed of polytetrafluoroethylene under specific process conditions can be accurately calculated. The critical speed refers to the maximum speed at which the material can maintain stable deformation without damage or abnormal flow during the stretching process. The calculated critical speed is used as an important reference value and set between the input layer and the hidden layer of the machine learning model. In this way, when the input stretching speed data is passed to this layer, the model can compare and judge it with the critical speed. When the input stretching speed exceeds the set critical speed, the model triggers the parameter correction mechanism. This is because in the actual biaxial stretching process, if the stretching speed is too fast and exceeds the critical speed of polytetrafluoroethylene, the material may have problems such as uneven deformation, molecular chain breakage or structural damage, which will lead to the performance degradation of the fiber membrane. By triggering parameter correction, the model can adjust the input stretching speed or other related parameters to keep it within a reasonable range to ensure the accuracy and reliability of the prediction results. For example, the stretching speed can be appropriately reduced or other process parameters can be optimized and adjusted to simulate the measures taken in actual production to avoid exceeding the critical speed. A porosity restriction function is set in the output layer of the machine learning model. Porosity is one of the important structural parameters of polytetrafluoroethylene fiber membrane, which directly affects the air permeability, filtration efficiency and other properties of the fiber membrane. According to the application requirements and actual production experience of the fiber membrane, a reasonable porosity range can be determined and set as the interval of the porosity restriction function. For example, for fiber membranes used in certain filtration applications, the porosity may need to be controlled between 30% and 70% to ensure good filtration performance and mechanical strength. When the model predicts the porosity of the fiber membrane, the predicted result is compared with the interval set by the porosity restriction function. If the predicted porosity exceeds the interval, it means that the prediction result of the model may not meet the actual production requirements or physical laws. Once the predicted porosity exceeds the set interval, the model will correct the output result according to the preset adjustment strategy. The preset adjustment strategy can be in many ways. For example, the predicted porosity can be adjusted to the boundary value of the interval, or it can be reduced or enlarged according to a certain proportion according to the degree of excess to return it to a reasonable range. This can ensure that the porosity results output by the model have physical meaning and practical application value, and avoid incorrect evaluation of fiber membrane performance due to unreasonable porosity prediction. At the same time, by correcting the output results, the model can also better learn the reasonable relationship between process parameters and porosity in the subsequent training process, thereby improving the prediction accuracy and stability of the model.

[0032] The critical speed of polytetrafluoroethylene is calculated according to the preset viscoelastic parameters, and the critical speed is set between the input layer and the hidden layer, so that the physical properties of polytetrafluoroethylene materials can be integrated into the machine learning model. When the input stretching speed exceeds the critical speed, the parameter correction is triggered, which can avoid the model predicting results that do not conform to the physical properties of the material, so that the model prediction reflects the actual situation more accurately. The porosity restriction function is set in the output layer. When the porosity predicted by the model exceeds the set range, the output result is corrected according to the preset adjustment strategy, which further ensures that the porosity data predicted by the model meets the actual process requirements and material properties. Porosity is an important structural parameter of the fiber membrane and has a significant impact on its performance. In actual production, due to the limitations of process conditions and material properties, the porosity of the fiber membrane is usually within a specific range. Through the porosity restriction function and output correction, the model can avoid predicting unreasonable porosity values ​​and improve the reliability of the prediction results. Embedding the physical constraint mechanism can effectively avoid abnormal prediction results of the model. In the absence of physical constraints, the machine learning model may make predictions based on the rules in the training data, but these rules may not fully meet the physical properties of the material and the actual process requirements. The introduction of physical constraint mechanism helps to improve the generalization ability of the model. When faced with different process parameters and material properties, the model can make reasonable predictions based on physical constraints, rather than relying solely on specific rules in the training data. This enables the model to better adapt and predict the performance of the fiber membrane when dealing with unseen data or new process conditions, improving the applicability and reliability of the model.

[0033] Optionally, the recommending the optimal process parameter combination and confidence interval by the machine learning model includes: Determining the range and step size of biaxial stretching process parameters, wherein the biaxial stretching process parameters include transverse stretching multiple, transverse speed, transverse temperature, longitudinal stretching multiple, longitudinal speed and longitudinal temperature; Determine a parameter space according to the range, and uniformly generate a plurality of parameter combinations in the parameter space using a Latin hypercube sampling method; The crossover probability, mutation probability, elite retention ratio and crowding degree calculation method are set, and the optimal solution is obtained from the plurality of parameter combinations through non-dominated sorting and crowding distance calculation.

[0034] The biaxial stretching process parameters include transverse stretching multiple, transverse speed, transverse temperature, longitudinal stretching multiple, longitudinal speed and longitudinal temperature. The range of these parameters is determined based on actual production experience and material properties. For example, the range of transverse stretching multiple may be 3 to 25 times, the range of longitudinal stretching multiple may be 1 to 16 times, the range of transverse speed may be 1 to 16 m / min, the range of longitudinal speed may be 1 to 5 m / min, the range of transverse temperature may be 210 to 320 ° C, and the range of longitudinal temperature may be 120 to 330 ° C. Determining these parameter ranges can ensure that the parameter combinations generated during the optimization process are within the practical range, avoiding parameter values ​​that exceed the equipment capacity or the material tolerance range. The step size refers to the interval between two adjacent parameter values ​​within the parameter range. Reasonable step size settings can ensure the sampling density of the parameter space, so that the parameter space can be fully explored during the optimization process. For example, for the transverse stretching multiple, the step size can be set to 1 times; for the transverse speed, the step size can be set to 1m / min; for the transverse temperature, the step size can be set to 10 ° C. The size of the step size will affect the accuracy and efficiency of optimization. A small step size will lead to too many parameter combinations and increase the amount of calculation; a large step size may cause some important parameter combinations to be missed. Latin Hypercube Sampling (LHS) is a statistical method for sampling in multidimensional parameter space. The basic idea is to evenly divide the value range of each parameter into several sub-intervals, and then randomly select a sample point in each sub-interval. For each parameter, these sample points are evenly distributed within the value range of the parameter, and the sample points of different parameters are independent of each other. In this way, a better coverage of the parameter space can be achieved with fewer samples, and the efficiency and representativeness of sampling can be improved. According to the determined biaxial stretching process parameter range and step size, the dimension of the parameter space and the value range of each dimension are determined. Then, the Latin hypercube sampling method is used to evenly generate multiple parameter combinations in the parameter space. For example, if the dimension of the parameter space is 6 (i.e., transverse stretching multiple, transverse speed, transverse temperature, longitudinal stretching multiple, longitudinal speed, and longitudinal temperature), the value range of each dimension is [3, 25], [1, 16], [210, 320], [1, 16], [1, 5], and [120, 330], and the step sizes are 1, 1, 10, 1, 0.5, and 10, respectively. Then, multiple parameter combinations can be generated, such as (3, 1, 210, 1, 1, 120), (4, 2, 220, 2, 1.5, 130), (5, 3, 230, 3, 2, 140), etc. These parameter combinations will be used as inputs for subsequent optimization algorithms to find the optimal process parameter combinations. In order to find the optimal solution from the generated multiple parameter combinations, the parameters of the optimization algorithm need to be set. Common optimization algorithm parameters include crossover probability, mutation probability, elite retention ratio, and crowding calculation method.The crossover probability and mutation probability are used to control the crossover and mutation operations in the genetic algorithm. The crossover operation allows different parameter combinations to exchange information with each other and generate new parameter combinations; the mutation operation can introduce randomness in the parameter combination and increase the diversity of solutions. The elite retention ratio is used to retain a certain proportion of excellent individuals in each generation of evolution to ensure the convergence speed of the optimization algorithm and the quality of the solution. The crowding degree calculation method is used to measure the degree of crowding between individuals and is used for crowding distance calculation in the non-dominated sorting genetic algorithm II (NSGA-II) to maintain the diversity of solutions. Through non-dominated sorting and crowding distance calculation, the optimal solution is obtained from multiple parameter combinations. Non-dominated sorting is to sort the parameter combinations according to the quality of their objective function values. The parameter combination with the better objective function value is ranked higher. The crowding distance calculation is used to measure the degree of crowding between individuals after non-dominated sorting. The individuals with larger crowding distances are more dispersed and more likely to be retained. Through non-dominated sorting and crowding distance calculation, a set of optimal solutions can be obtained. These optimal solutions have achieved a balance between the objective function value and the diversity of solutions, and can be used as the recommended optimal process parameter combination.

[0035] The range and step size of the biaxial stretching process parameters, including transverse stretching multiple, transverse speed, transverse temperature, longitudinal stretching multiple, longitudinal speed and longitudinal temperature, are clarified, which provides clear boundaries and search accuracy for the subsequent parameter combination generation. This helps to conduct a comprehensive and effective search in the parameter space, avoiding insufficient optimization or invalid calculation caused by too wide or too narrow parameter range. The Latin hypercube sampling method is used to uniformly generate multiple parameter combinations in the parameter space, which can ensure the uniform distribution of parameters in each dimension and improve the representativeness and diversity of the sampling points. Compared with the traditional random sampling method, Latin hypercube sampling can better cover the parameter space with a smaller number of samples, reduce sampling bias, and improve the global search ability and efficiency of the optimization algorithm. The optimization algorithm is finely controlled by setting parameters such as crossover probability, mutation probability, elite retention ratio and crowding calculation method. The reasonable setting of crossover probability and mutation probability can balance the global search and local search capabilities of the algorithm, avoid premature convergence or fall into the local optimal solution; the setting of elite retention ratio can retain the excellent individuals in the current population and accelerate the convergence speed; the crowding calculation method helps to maintain the diversity of solutions in the population. The non-dominated sorting and crowding distance calculation are used to obtain the optimal solution from multiple parameter combinations, which can simultaneously consider the optimization of multiple objective functions, such as the porosity, air permeability, filtration efficiency and other performance indicators of the fiber membrane. Non-dominated sorting can hierarchically sort the parameter combinations according to their performance on each objective function, and the crowding distance calculation can sort the solutions in the same layer according to the distribution density, so as to select the parameter combinations that have excellent performance on the objective function and are evenly distributed, thereby enhancing the diversity and quality of the optimization results. The optimal process parameter combination obtained by the above method can better meet the performance requirements of the fiber membrane and improve product quality and production efficiency. Using these optimized process parameter combinations for the training and verification of the machine learning model can enable the model to learn a more accurate relationship between process parameters and fiber membrane performance, thereby improving the prediction accuracy of the model.

[0036] Optionally, the recommending the optimal process parameter combination and confidence interval by the machine learning model further includes: Introducing random disturbances to each of the biaxial stretching process parameters, running the machine learning model multiple times, and obtaining performance index results under different disturbances; The coefficients of variation of air permeability and filtration efficiency are calculated according to the performance indicator results, and a process solution whose coefficients of variation of air permeability and filtration efficiency are both lower than a first preset threshold is screened out.

[0037] The random perturbation is introduced for each biaxial stretching process parameter in order to simulate the possible fluctuation and uncertainty of the process parameters in the actual production process. Such fluctuation may be caused by factors such as equipment accuracy, environmental changes, and raw material differences. By introducing random perturbations, the performance index changes of the model under different perturbations can be evaluated, so as to screen out process solutions that are insensitive to process parameter fluctuations and have stable performance. When the machine learning model is run multiple times, random perturbations are introduced for each biaxial stretching process parameter (such as transverse stretching multiple, transverse speed, transverse temperature, longitudinal stretching multiple, longitudinal speed, and longitudinal temperature). The amplitude of the random perturbation can be determined based on actual production experience or the allowable fluctuation range of the process parameters. Usually, a random number generation method such as normal distribution and uniform distribution can be used to generate perturbation values, and then add them to the original process parameters to obtain the perturbed process parameter combination. After the random perturbation is introduced, the machine learning model is run multiple times, and each run uses the perturbed process parameter combination as input to obtain the corresponding performance index results (such as air permeability, filtration efficiency, etc.). Through multiple runs, a set of performance index results under different perturbations can be obtained, which reflects the performance changes of the fiber membrane under the condition of process parameter fluctuations. The performance index results obtained from each run are collected and recorded to form a data set containing multiple performance index values. These data will be used for subsequent coefficient of variation calculation and process solution screening. The coefficient of variation (CV) is an important indicator for measuring the degree of data dispersion. It is equal to the ratio of the standard deviation to the mean value, usually expressed as a percentage. The coefficient of variation can eliminate the influence of the data dimension and the size of the mean value, and more accurately reflect the relative degree of dispersion of the data. For the two performance indicators of air permeability and filtration efficiency, their coefficients of variation under different disturbances are calculated respectively. The specific calculation steps are as follows: Calculate the average values ​​of air permeability and filtration efficiency in multiple runs, recorded as Q and E respectively. Calculate the standard deviations of air permeability and filtration efficiency in multiple runs, recorded as SDQ and SDE respectively. Calculate the coefficient of variation CVQ of air permeability: CVQ=Q / SDQ×100%. Calculate the coefficient of variation CVE of filtration efficiency: CVE=E / SDE×100%. According to the set first preset threshold, the coefficients of variation of air permeability and filtration efficiency are screened. Only when the coefficient of variation of air permeability and filtration efficiency are both lower than the threshold, the corresponding process solution is considered to be a stable and reliable process solution. The first preset threshold can be determined according to the requirements of actual production for the stability of fiber membrane performance, and can usually be set through experimental verification or empirical judgment.

[0038] Introducing random disturbances for each biaxial stretching process parameter and running the machine learning model multiple times can simulate the fluctuations of process parameters caused by various factors in the actual production process. This helps to evaluate the prediction performance of the model under different disturbances and gain a more comprehensive understanding of the stability and reliability of the model. The coefficient of variation of air permeability and filtration efficiency is calculated based on the performance index results under different disturbances. The smaller the coefficient of variation, the less the performance index is affected by the fluctuation of process parameters, and the more stable the prediction results of the model. Screening out process solutions with coefficients of variation lower than the first preset threshold can ensure that the recommended process parameter combination has high stability and reliability in actual production and reduce the risk of quality fluctuations in the production process. By calculating the coefficient of variation of air permeability and filtration efficiency, the performance stability of different process solutions is quantitatively evaluated. This not only takes into account the performance of the process solution under ideal conditions, but also comprehensively considers its performance stability under fluctuations that may be encountered in actual production, thereby screening out process solutions that better meet actual production needs.

[0039] Optionally, the method further includes: The measured data of the samples prepared according to the optimal process parameter combination are subjected to deviation analysis with the predicted values. When the absolute error of the porosity or the error of the filtration efficiency exceeds the second preset threshold, the transfer learning strategy is used to freeze the underlying parameters related to the basic feature extraction in the machine learning model, and the preset fully connected layer is adjusted according to the measured data of the samples.

[0040] Deviation analysis is performed on the measured data of the samples prepared according to the optimal process parameter combination and the predicted values, which is a way to evaluate the predictive performance of the machine learning model. By comparing the differences between the measured data and the predicted values, the accuracy and reliability of the model in practical applications can be understood. Deviation analysis usually includes calculating indicators such as the absolute error of porosity and the error of filtration efficiency, which can quantify the gap between the model prediction results and the actual situation. When the absolute error of porosity or the error of filtration efficiency exceeds the second preset threshold, it means that there is a large deviation between the prediction results of the model and the actual situation, and the model needs to be adjusted and optimized. At this time, the transfer learning strategy is used to freeze the underlying parameters related to the basic feature extraction in the machine learning model, and adjust the preset fully connected layer according to the measured data of the sample. The underlying parameters are usually related to the extraction of basic features, and these features have a certain degree of universality in different tasks. Freezing the underlying parameters can retain the basic feature representations learned by the model in the original task, avoiding the loss or excessive changes of these features due to retraining. The fully connected layer is usually used to map the extracted features to specific output results, and may need to be adjusted according to the specific situation in different tasks. Adjusting the preset fully connected layer according to the actual sample data can make the model better adapt to the current task and improve the model's prediction accuracy and generalization ability.

[0041] By analyzing the deviation between the measured data and predicted values ​​of the samples prepared according to the optimal process parameter combination, the difference between the model prediction results and the actual situation can be discovered in time. When the absolute error of porosity or the error of filtration efficiency exceeds the second preset threshold, it means that the prediction accuracy of the model within the current process parameter range needs to be improved and the model needs to be adjusted. By using the transfer learning strategy, the underlying parameters related to the basic feature extraction in the machine learning model are frozen, and the preset fully connected layer is adjusted according to the measured data of the sample. On the basis of retaining the original feature extraction capability of the model, the model can be better adapted to new data and process conditions. This helps to improve the prediction accuracy of the model within a specific process parameter range and enhance the adaptability of the model. Adjusting the preset fully connected layer according to the measured data of the sample can make the model better adapt to the data distribution within a specific process parameter range. The fully connected layer is usually used to map the extracted features to the final output result. By adjusting the parameters of the fully connected layer, the prediction results of the model within a specific process parameter range can be made more accurate, thereby improving the generalization ability of the model.

[0042] Optionally, the using of the transfer learning strategy to freeze the underlying parameters related to basic feature extraction in the machine learning model and adjusting the preset fully connected layer according to the sample measured data includes: Freeze the weight matrix and convolution kernel parameters of the feature extraction layer in the initial machine learning model, and when the loss function decrease rate of three consecutive rounds of training is less than a preset percentage, multiply the current learning rate of the fully connected layer by the attenuation coefficient.

[0043] In the initial machine learning model, the weight matrix and convolution kernel parameters of the feature extraction layer have learned the basic feature representation through a large amount of data training. These features are of great significance for the general patterns and laws in the preparation process of biaxially stretched polytetrafluoroethylene fiber membrane. Freezing these parameters means that these parameters will no longer be updated in the subsequent transfer learning process, thereby retaining the model's ability to extract basic features and avoiding the model's loss of recognition ability for general features due to over-adjustment. The preset fully connected layer is the part of the model responsible for mapping features to output results (such as performance indicators such as porosity and filtration efficiency). In this step, the parameters of the fully connected layer will be adjusted according to the measured data of the samples prepared according to the optimal process parameter combination, so that it can better adapt to the new data distribution and process requirements, and improve the prediction accuracy of the model within a specific process parameter range. Monitor the loss function decline rate of three consecutive rounds of training. The loss function is an indicator that measures the difference between the model's predicted value and the actual value. If the loss function decline rate is less than the preset percentage, it indicates that the effect of continuing to train the fully connected layer at the current learning rate is not obvious, and it may be trapped in a local optimum or the learning process has become slow. When the rate of decrease of the loss function is detected to be too low, the current learning rate of the fully connected layer is multiplied by the decay coefficient. The learning rate is an important hyperparameter for controlling the step size of model parameter updates. Reducing the learning rate can make the model's exploration in the parameter space more refined, which helps the fully connected layer to further optimize the parameters and improve the performance and stability of the model.

[0044] Example 1 of the present application: A polytetrafluoroethylene dispersion resin (number average molecular weight 20 million, crystallinity ≥ 97%) and an isoparaffin lubricant are mixed in a weight ratio of 1:0.25, and the aging conditions are 50°C and 36 hours; a non-deoiled base belt with a width of 300 mm is formed through blanking (40°C, holding pressure for 5 minutes), push extrusion (50°C, compression ratio 150) and calendering treatment; an orthogonal experiment of biaxial stretching is designed, with a longitudinal stretching multiple of 1 to 16 times, a speed of 1 to 5 m / min, and a temperature of 120 to 330°C; a lateral expansion multiple of 3 to 25 times, a speed of 1 to 16 m / min, temperature 210-320℃; prepare 600 groups of PTFE fiber membrane samples, record the process parameters of each group, and test their average pore size, porosity, air permeability and filtration efficiency; use random forest algorithm, input parameters include longitudinal / transverse stretching temperature, speed, multiple and heat setting temperature, output targets are pore size, porosity, air permeability and filtration efficiency and other performance, the data set is divided into training set and test set at 7:3, and hyperparameters (such as tree depth, number of leaf nodes) are optimized by grid search; RMSE (root mean square error) is The cross-validation model with a value of 1.8 and R²=0.92 (R² is the coefficient of determination) has a strong predictive ability; the target performance is set as follows: porosity ≥85%, filtration efficiency ≥99.9%, and the model reversely recommends a combination of process parameters: longitudinal stretching temperature 210°C, speed 3m / min, multiple 8 times, transverse expansion temperature 280°C, speed 10m / min, multiple 15 times, and heat setting temperature 360°C; the sample is prepared according to the optimal parameters given by the model, and the measured porosity is 86.2%, and the filtration efficiency is 99.95%, achieving the target performance.

[0045] Example 2 of the present application: A polytetrafluoroethylene dispersion resin (number average molecular weight 20 million, crystallinity ≥ 97%) and an isoparaffin lubricant are mixed in a weight ratio of 1:0.25, and the aging conditions are 50°C and 36 hours; a non-deoiled base belt with a width of 300 mm is formed through blanking (40°C, holding pressure for 5 minutes), push extrusion (50°C, compression ratio 150) and calendering treatment; an orthogonal experiment of biaxial stretching is designed, with a longitudinal stretching multiple of 1 to 16 times, a speed of 1 to 5 m / min, and a temperature of 120 to 330°C; a transverse expansion multiple of 3 to 25 times, a speed of 1 to 16 m / min, and a temperature of 210 to 320°C; 600 groups of PTFE fiber membrane samples are prepared, the process parameters of each group are recorded, and the average pore size, porosity, air permeability and filtration efficiency thereof are tested; Figure 2 is a schematic diagram of the structure of the core machine learning algorithm disclosed in the embodiment of this application, such as Figure 2As shown, an artificial neural network algorithm is used to construct a multi-layer perceptron. The input layer is the longitudinal / transverse temperature, speed, and multiple (a total of 6 input nodes), the hidden layer contains 2 fully connected layers (the number of nodes is 8 and 10 respectively), and the output layer is the performance of pore size, porosity, air permeability and filtration efficiency (12 output nodes). The data set is divided into training set, validation set and test set according to 7:2:1. The Adam optimizer is used, the initial learning rate is 0.001, the loss function is the mean square error (MSE), and the training cycle is 300000 times; the prediction ability of the model is strong through loss curve diagnosis and RMSE of 1.5 and R²=0.984 cross-validation; set the target performance: porosity ≥88%, air permeability ≥25 L / (m²·s), filtration efficiency ≥99.95%, and the model reversely recommends the process parameter combination: longitudinal stretching temperature 230℃, speed 4 m / min, multiple 10 times, lateral expansion temperature 290℃, speed 12m / min, multiple 16 times; samples were prepared according to the optimal parameters given by the model, with the measured porosity 89.3%, air permeability 26.8L / (m²·s), and filtration efficiency 99.98%.

[0046] Pore ​​size test (average pore size, most probable pore size): Use PMI (gas permeation pore size analyzer) equipment for testing. Cut the sample to be tested into a circle or other shape with the diameter specified by the equipment. Note that the inside of the test sample cannot be cut. After the sample is soaked in the surfactant required by the PMI equipment test for 30 minutes, put the sample into the test slot of the instrument, and then tighten the cover of the slot to conduct the test. The test result is directly read by the equipment and converted into the distribution of the proportion (%) of each pore size (μm) of the filter material to the entire material (including average pore size, maximum pore size μm, minimum pore size μm, pore size distribution), and three positions (A, B, C) are evenly selected along the transverse direction of the polytetrafluoroethylene film for sampling and testing, and three points (1, 2, 3) are selected at each position.

[0047] Air filtration efficiency: Tested in accordance with GB 19083-2010, the sodium chloride (NaCl) aerosol particle size distribution used in the specified test conditions should be a particle number median diameter (CMD) of 0.075μm±0.020μm, a geometric standard deviation of no more than 1.86 (equivalent to an aerodynamic mass median diameter (MMAD) of 0.24μm±0.06μm), and a gas flow rate of 85L / min. The air filtration efficiency test is carried out at the same sampling point as the aperture test above.

[0048] Air permeability: Based on GB / T5453-1997 standard, the air permeability of polytetrafluoroethylene fiber membrane is tested. The test point is a circle with an area of ​​100cm2, the pressure is 200Pa, the unit is m3 / m2*min, and there are 10 test points with equal spacing in the width direction, and the average value is taken.

[0049] This embodiment also discloses a system for predicting the preparation process of biaxially stretched polytetrafluoroethylene fiber membrane. Figure 3 Schematic diagram of a module of a system for predicting a process for preparing a biaxially oriented polytetrafluoroethylene fiber membrane disclosed in an embodiment of the present application, such as Figure 3 As shown, the system includes a data module 301, a model module 302 and a prediction module 303, wherein: A data module 301 is configured to obtain a data set of multiple groups of fiber membrane samples prepared by a preset orthogonal experiment, wherein the data set includes a combination of biaxial stretching process parameters, pore structure data, and performance indicators for preparing a target group of fiber membrane samples, wherein the target group of fiber membrane samples is any one of the multiple groups of fiber membrane samples; A model module 302 is configured to construct an initial machine learning model, use the combination of biaxial stretching process parameters in the data set as input parameters, use the pore structure data and performance indicators as output parameters, train the initial machine learning model to obtain an intermediate model, and adjust the parameters of the intermediate model through cross-validation to obtain a machine learning model; The prediction module 303 is configured to recommend an optimal process parameter combination and a confidence interval through the machine learning model according to the set target performance.

[0050] Optionally, the model module 302 is further configured to: Calculating the critical speed of polytetrafluoroethylene according to preset viscoelastic parameters, setting the critical speed between the input layer and the hidden layer, and triggering parameter correction when the input stretching speed exceeds the critical speed; A porosity restriction function is set in the output layer. When the porosity predicted by the model exceeds the interval set by the porosity restriction function, the output result is corrected according to a preset adjustment strategy.

[0051] Optionally, the prediction module 303 is configured to: Determining the range and step size of biaxial stretching process parameters, wherein the biaxial stretching process parameters include transverse stretching multiple, transverse speed, transverse temperature, longitudinal stretching multiple, longitudinal speed and longitudinal temperature; Determine a parameter space according to the range, and uniformly generate a plurality of parameter combinations in the parameter space using a Latin hypercube sampling method; The crossover probability, mutation probability, elite retention ratio and crowding degree calculation method are set, and the optimal solution is obtained from the plurality of parameter combinations through non-dominated sorting and crowding distance calculation.

[0052] Optionally, the prediction module 303 is configured to: Introducing random disturbances to each of the biaxial stretching process parameters, running the machine learning model multiple times, and obtaining performance index results under different disturbances; The coefficients of variation of air permeability and filtration efficiency are calculated according to the performance indicator results, and a process solution whose coefficients of variation of air permeability and filtration efficiency are both lower than a first preset threshold is screened out.

[0053] Optionally, the system further comprises an adjustment module configured to: The measured data of the samples prepared according to the optimal process parameter combination are subjected to deviation analysis with the predicted values. When the absolute error of the porosity or the error of the filtration efficiency exceeds the second preset threshold, the transfer learning strategy is used to freeze the underlying parameters related to the basic feature extraction in the machine learning model, and the preset fully connected layer is adjusted according to the measured data of the samples.

[0054] Optionally, the adjustment module is configured to: Freeze the weight matrix and convolution kernel parameters of the feature extraction layer in the initial machine learning model, and when the loss function decrease rate of three consecutive rounds of training is less than a preset percentage, multiply the current learning rate of the fully connected layer by the attenuation coefficient.

[0055] Optionally, the data module 301 is configured to: The polytetrafluoroethylene dispersion resin and the hydrocarbon lubricating oil are mixed in a preset weight ratio, and the aging conditions are 30-70°C and 24-48 hours; the undeoiled base belt is formed by billet making, pushing extrusion and calendering; Control the longitudinal stretching multiple to 1-16 times, the longitudinal speed to 1-5 m / min, the longitudinal temperature to 120-330° C., the transverse expansion multiple to 3-25 times, the transverse speed to 1-16 m / min, and the transverse temperature to 210-320° C. to generate multiple groups of fiber membrane samples; The pore structures and properties of multiple groups of the fiber membrane samples are detected to obtain the data set.

[0056] It should be noted that: when the device provided in the above embodiment realizes its function, only the division of the above functional modules is used as an example. In actual application, the above functions can be assigned to different functional modules as needed, that is, the internal structure of the device is divided into different functional modules to complete all or part of the functions described above. In addition, the device and method embodiments provided in the above embodiment belong to the same concept, and the specific implementation process is detailed in the method embodiment, which will not be repeated here.

[0057] This embodiment also discloses an electronic device, referring to Figure 4The electronic device may include: at least one processor 401 , at least one communication bus 402 , a user interface 403 , a network interface 404 , and at least one memory 405 .

[0058] The communication bus 402 is used to realize the connection and communication between these components.

[0059] The user interface 403 may include a display screen (Display) and a camera (Camera). Optionally, the user interface 403 may also include a standard wired interface and a wireless interface.

[0060] The network interface 404 may optionally include a standard wired interface or a wireless interface (such as a WI-FI interface).

[0061] Among them, the processor 401 may include one or more processing cores. The processor 401 uses various interfaces and lines to connect various parts in the entire server, and executes various functions of the server and processes data by running or executing instructions, programs, code sets or instruction sets stored in the memory 405, and calling data stored in the memory 405. Optionally, the processor 401 can be implemented in at least one hardware form of digital signal processing (Digital Signal Processing, DSP), field programmable gate array (Field-Programmable Gate Array, FPGA), and programmable logic array (Programmable Logic Array, PLA). The processor 401 can integrate one or a combination of a central processing unit (Central Processing Unit, CPU), a graphics processing unit (Graphics Processing Unit, GPU) and a modem. Among them, the CPU mainly processes the operating system, user interface and application programs; the GPU is responsible for rendering and drawing the content to be displayed on the display screen; the modem is used to process wireless communications. It can be understood that the above-mentioned modem may not be integrated into the processor 401, and it can be implemented separately through a chip.

[0062] Among them, the memory 405 may include a random access memory (Random Access Memory, RAM) and may also include a read-only memory (Read-Only Memory). Optionally, the memory 405 includes a non-transitory computer-readable storage medium. The memory 405 can be used to store instructions, programs, codes, code sets or instruction sets. The memory 405 may include a program storage area and a data storage area, wherein the program storage area may store instructions for implementing an operating system, instructions for at least one function (such as a touch function, a sound playback function, an image playback function, etc.), instructions for implementing the above-mentioned various method embodiments, etc.; the data storage area may store data involved in the above-mentioned various method embodiments, etc. The memory 405 may also be optionally at least one storage device located away from the aforementioned processor 401. As Figure 4 As shown, the memory 405 as a computer storage medium may include an operating system, a network communication module, a user interface module, and an application program for a method for predicting a process for preparing a biaxially oriented polytetrafluoroethylene fiber membrane.

[0063] exist Figure 4 In the electronic device shown, the user interface 403 is mainly used to provide an input interface for the user and obtain data input by the user; and the processor 401 can be used to call the application program stored in the memory 405 for predicting the method for preparing the biaxially oriented polytetrafluoroethylene fiber membrane. When executed by one or more processors 401, the electronic device executes one or more methods in the above-mentioned embodiments.

[0064] It should be noted that, for the aforementioned method embodiments, for the sake of simplicity, they are all described as a series of action combinations, but those skilled in the art should be aware that the present application is not limited by the order of the actions described, because according to the present application, certain steps can be performed in other orders or simultaneously. Secondly, those skilled in the art should also be aware that the embodiments described in the specification are all preferred embodiments, and the actions and modules involved are not necessarily required for the present application.

[0065] In the above embodiments, the description of each embodiment has its own emphasis. For parts that are not described in detail in a certain embodiment, reference can be made to the relevant descriptions of other embodiments.

[0066] In the several embodiments provided in this application, it should be understood that the disclosed devices can be implemented in other ways. For example, the device embodiments described above are only schematic, such as the division of units, which is only a logical function division. There may be other division methods in actual implementation, such as multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some service interfaces, and the indirect coupling or communication connection of devices or units can be electrical or other forms.

[0067] The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed on multiple network units. Some or all of the units may be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0068] In addition, each functional unit in each embodiment of the present application may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit. The above-mentioned integrated unit may be implemented in the form of hardware or in the form of software functional units.

[0069] If the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable memory. Based on this understanding, the technical solution of the present application, or the part that contributes to the prior art, or all or part of the technical solution can be embodied in the form of a software product, and the computer software product is stored in a memory 405, including a number of instructions for a computer device (which can be a personal computer, server or network device, etc.) to perform all or part of the steps of the various embodiments of the present application. The aforementioned memory 405 includes: various media that can store program codes, such as a USB flash drive, a mobile hard disk, a magnetic disk or an optical disk.

[0070] The above is only an exemplary embodiment of the present disclosure and cannot be used to limit the scope of the present disclosure. That is, any equivalent changes and modifications made according to the teachings of the present disclosure are still within the scope of the present disclosure. After considering the disclosure of the specification, those skilled in the art will easily think of other embodiments of the present disclosure. This application is intended to cover any modification, use or adaptation of the present disclosure, which follows the general principles of the present disclosure and includes common knowledge or customary technical means in the technical field that are not recorded in the present disclosure. The description and examples are only regarded as exemplary, and the scope and spirit of the present disclosure are defined by the claims.

Claims

1. A method for predicting the preparation process of biaxially oriented polytetrafluoroethylene fiber membrane, characterized in that: Applied to the preparation platform, the method comprises: Acquire a data set of multiple groups of fiber membrane samples prepared by a preset orthogonal experiment, wherein the data set includes a combination of biaxial stretching process parameters, pore structure data, and performance indicators for preparing a target group of fiber membrane samples, wherein the target group of fiber membrane samples is any one group of the multiple groups of fiber membrane samples; Constructing an initial machine learning model, taking the combination of biaxial stretching process parameters in the data set as input parameters, taking the pore structure data and performance index as output parameters, training the initial machine learning model to obtain an intermediate model, and adjusting the parameters of the intermediate model through cross-validation to obtain a machine learning model; According to the set target performance, the machine learning model is used to recommend the optimal process parameter combination and confidence interval. The construction of the initial machine learning model includes embedding a physical constraint mechanism, specifically including: Calculating the critical speed of polytetrafluoroethylene according to preset viscoelastic parameters, setting the critical speed between the input layer and the hidden layer, and triggering parameter correction when the input stretching speed exceeds the critical speed; A porosity restriction function is set in the output layer. When the porosity predicted by the model exceeds the interval set by the porosity restriction function, the output result is corrected according to a preset adjustment strategy.

2. The method for predicting the preparation process of biaxially oriented polytetrafluoroethylene fiber membrane according to claim 1, characterized in that: The recommending of the optimal process parameter combination and confidence interval by the machine learning model includes: Determining the range and step size of biaxial stretching process parameters, wherein the biaxial stretching process parameters include transverse stretching multiple, transverse speed, transverse temperature, longitudinal stretching multiple, longitudinal speed and longitudinal temperature; Determine a parameter space according to the range, and uniformly generate a plurality of parameter combinations in the parameter space using a Latin hypercube sampling method; The crossover probability, mutation probability, elite retention ratio and crowding degree calculation method are set, and the optimal solution is obtained from the plurality of parameter combinations through non-dominated sorting and crowding distance calculation.

3. The method for predicting the preparation process of biaxially oriented polytetrafluoroethylene fiber membrane according to claim 2, characterized in that: The method of recommending the optimal process parameter combination and confidence interval by the machine learning model further includes: Introducing random disturbances to each of the biaxial stretching process parameters, running the machine learning model multiple times, and obtaining performance index results under different disturbances; The coefficients of variation of air permeability and filtration efficiency are calculated according to the performance indicator results, and a process solution whose coefficients of variation of air permeability and filtration efficiency are both lower than a first preset threshold is screened out.

4. The method for predicting the preparation process of biaxially oriented polytetrafluoroethylene fiber membrane according to claim 1, characterized in that: The method further comprises: The measured data of the samples prepared according to the optimal process parameter combination are subjected to deviation analysis with the predicted values. When the absolute error of the porosity or the error of the filtration efficiency exceeds the second preset threshold, the transfer learning strategy is used to freeze the underlying parameters related to the basic feature extraction in the machine learning model, and the preset fully connected layer is adjusted according to the measured data of the samples.

5. The method for predicting the preparation process of biaxially oriented polytetrafluoroethylene fiber membrane according to claim 4, characterized in that: The method of utilizing the transfer learning strategy to freeze the underlying parameters related to basic feature extraction in the machine learning model and adjusting the preset fully connected layer according to the sample measured data includes: Freeze the weight matrix and convolution kernel parameters of the feature extraction layer in the initial machine learning model, and when the loss function decrease rate of three consecutive rounds of training is less than a preset percentage, multiply the current learning rate of the fully connected layer by the attenuation coefficient.

6. A system for predicting the preparation process of biaxially oriented polytetrafluoroethylene fiber membrane, characterized in that: It includes data module, model module and prediction module, among which: A data module is configured to obtain a data set of multiple groups of fiber membrane samples prepared by a preset orthogonal experiment, wherein the data set includes a combination of biaxial stretching process parameters, pore structure data, and performance indicators for preparing a target group of fiber membrane samples, wherein the target group of fiber membrane samples is any one of the multiple groups of fiber membrane samples; A model module is configured to construct an initial machine learning model, use the combination of biaxial stretching process parameters in the data set as input parameters, use the pore structure data and performance indicators as output parameters, train the initial machine learning model to obtain an intermediate model, and adjust the parameters of the intermediate model through cross-validation to obtain a machine learning model; A prediction module is configured to recommend an optimal process parameter combination and a confidence interval through the machine learning model according to a set target performance, The model module is also configured to: Calculating the critical speed of polytetrafluoroethylene according to preset viscoelastic parameters, setting the critical speed between the input layer and the hidden layer, and triggering parameter correction when the input stretching speed exceeds the critical speed; A porosity restriction function is set in the output layer. When the porosity predicted by the model exceeds the interval set by the porosity restriction function, the output result is corrected according to a preset adjustment strategy.

7. An electronic device, characterized in that: It includes a processor, a memory, a user interface and a network interface, the memory is used to store instructions, the user interface and the network interface are both used to communicate with other devices, and the processor is used to execute the instructions stored in the memory so that the electronic device executes the method as described in any one of claims 1-5.

8. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores instructions, and when the instructions are executed, the method according to any one of claims 1 to 5 is performed.

Images