Carbon fiber heat treatment optimization method and system based on multi-physical field coupling

Abstract

The application provides a carbon fiber heat treatment optimization method and system based on multi-physical field coupling. The carbon fiber heat treatment optimization method based on multi-physical field coupling of the application condenses key characteristic quantities reflecting microstructure evolution through multi-physical field coupling simulation, and embeds them as interaction items into a machine learning model, greatly improving the physical interpretability of the model and the prediction accuracy and generalization ability under small sample conditions. The parameterized reduced-order modeling technology is used to simplify the complex multi-physical field model, greatly reducing the calculation cost on the premise of keeping the key physical information, so that it is possible to use simulation data to support machine learning model training, solving the efficiency bottleneck of directly using high-fidelity simulation for optimization. Through the digital twin system, real-time adjustment and closed-loop control of process parameters in the carbon fiber heat treatment process are realized.

Description

Technical Field

[0001] This invention relates to the technical field of carbon fiber processing, and more specifically, to a method and system for optimizing carbon fiber heat treatment based on multi-physics coupling. Background Technology

[0002] As a strategic key material, the preparation process of carbon fiber involves strong nonlinear coupling between temperature field, tension field, atmosphere and chemical reaction field in the multi-stage heat treatment of precursor fibers, such as pre-oxidation and carbonization, which jointly determine the final microstructure and mechanical properties.

[0003] Currently, process optimization mainly relies on trial and error and experience-based adjustments, which suffers from low efficiency and high cost. In recent years, machine learning techniques such as BP neural networks, support vector machines, and hybrid deep learning models have been applied to establish predictive relationships between process parameters and performance. However, existing methods still have the following problems when applied to carbon fiber heat treatment: First, they have poor interpretability, making it difficult to reveal the intrinsic structure-property relationship between "process-structure-performance". Second, they are highly dependent on data, and are prone to overfitting under conditions of high dimension, small sample size, and lack of microstructure data, which limits prediction accuracy and generalization ability. Third, they ignore multi-physics coupling mechanisms and do not introduce physical constraints such as heat conduction, reaction exothermics, and stress evolution, which may lead to prediction results that violate actual laws and restrict their effective guidance in production.

[0004] Therefore, there is an urgent need for an optimization method for carbon fiber heat treatment based on multi-physics coupling. Summary of the Invention

[0005] In view of the above problems, the purpose of this invention is to provide a carbon fiber heat treatment optimization method and system based on multi-physics field coupling, so as to solve at least one problem existing in the prior art.

[0006] In a first aspect, the present invention provides a carbon fiber heat treatment optimization method based on multiphysics coupling, applied to electronic devices, comprising: acquiring process parameters in the carbon fiber heat treatment process; wherein the process parameters include temperature, tension, fiber linear velocity, and furnace atmosphere concentration in each temperature zone; performing simulation based on the process parameters using a multiphysics coupling digital twin model, and obtaining key feature quantities in the carbon fiber heat treatment process based on the simulation results; using the process parameters and the key feature quantities as input features, inputting them into a preset multiphysics coupling-based carbon fiber heat treatment model to predict the performance indicators of the carbon fiber; using a multi-objective optimization algorithm to perform synergistic optimization of the process parameters with the optimization objectives of maximizing fiber tensile strength, maximizing elastic modulus, and minimizing energy consumption efficiency, to obtain a Pareto optimal process parameter set; and controlling the carbon fiber heat treatment process according to the Pareto optimal process parameter set.

[0007] In addition, an optional technical solution includes collecting heat treatment operation data and carbon fiber product performance data adjusted according to the Pareto optimal process parameter set; and updating the multi-physics coupled digital twin model and the carbon fiber heat treatment model based on the heat treatment operation data and carbon fiber product performance data.

[0008] In addition, an optional technical solution is that the method for obtaining the multi-physics coupled carbon fiber heat treatment model includes: constructing a multi-physics coupled digital twin model, wherein the physical fields include a temperature field, a stress field, and a chemical reaction field; wherein a parameterized order reduction modeling method is used to reduce the order of the multi-physics coupled model; based on the multi-physics coupled digital twin model, a carbon fiber heat treatment feature vector is obtained through numerical simulation and symbolic regression analysis; the carbon fiber heat treatment feature vector includes the fiber radial temperature gradient, the stress difference between the skin and core, the cyclization reaction conversion rate, the microcrystal size evolution rate, and feature parameters reflecting the differences in skin and core structures; constructing a machine learning model embedded with physical constraints, and training the machine learning model using a carbon fiber heat treatment dataset to obtain a trained multi-physics coupled carbon fiber heat treatment model; wherein the process parameter set and the carbon fiber heat treatment feature vector are used as input features of the multi-physics coupled carbon fiber heat treatment model, and the carbon fiber performance indicators are used as output features of the multi-physics coupled carbon fiber heat treatment model; and optimizing the trained multi-physics coupled carbon fiber heat treatment model using an improved dung beetle algorithm.

[0009] In addition, an optional technical solution is that the digital twin model of the multi-physics coupling is a neural network architecture, which includes a physical constraint subnet; the physical constraint subnet is constructed based on the energy conservation equation, the chemical reaction kinetic equation and the linear momentum conservation equation.

[0010] Furthermore, an optional technical solution is that the key characteristic quantities also include cross-scale energy state characteristics, evolution path discrimination characteristics, interface coupling stress characteristics, and structural anisotropic evolution characteristics; the cross-scale energy state characteristics include the power spectral density of the radial temperature field; wherein, the evolution path discrimination characteristics include the microcrystalline ordering rate and the by-product / main reaction rate ratio; the microcrystalline ordering rate is achieved by the following formula,

[0011] in, The rate of microcrystalline ordering; I G The intensity of the G peak for carbon materials; I D The intensity of the D peak of the carbon material; The byproduct / main reaction rate ratio is achieved by the following formula. , in, The instantaneous rate of the main reaction; The release rate of key byproducts; The interface coupling stress characteristics are realized by the following formula. , Where K is a constant related to the elastic modulus E, Poisson's ratio ν, and geometric dimensions of the two-layer material; and These are the average intrinsic strains of the core layer and the skin layer, respectively; The anisotropic evolution characteristics of the structure, including the orientation consistency index, are achieved through the following formula. , Among them, f H For Hermann orientation factor; The width of the orientation distribution; It is a constant.

[0012] Alternatively, an optional technical solution is to use the carbon fiber heat treatment dataset to train the machine learning model, where the loss function is implemented using the following formula. , in, L data For data loss; L pde Loss due to physical mechanisms; L bc Loss is for boundary / initial conditions; , , It is a constant; , , , in, N d This represents the total number of training data points. x i , t i , p i These are the spatial coordinates, time coordinates, and process parameter vector of the i-th data point, respectively. T ( x i , t i , p i ), These are the predicted values ​​of temperature and reaction conversion rate output by the neural network, respectively. T i , They were not the first i The true value of each data point; N p This represents the total number of residual points. ρ The density of the material; C p The specific heat capacity of the material; k The thermal conductivity of the material; This is the heat source term for the chemical reaction; N b This represents the total number of boundary points and initial condition points. k Index of the boundary / initial point; T (x k , t k , p k () represents the neural network's prediction at the boundary / initial point; T bc,k For the first k The known values ​​specified by the boundary / initial point.

[0013] In addition, an optional technical solution is that the carbon fiber heat treatment dataset includes simulation data and experimental data generated using a digital twin model coupled with multi-physics fields.

[0014] Secondly, the present invention also provides a carbon fiber heat treatment optimization system based on multi-physics coupling. The system includes: a detection data acquisition unit for acquiring process parameters during the carbon fiber heat treatment process; wherein the process parameters include temperature, tension, fiber linear velocity, and furnace atmosphere concentration in each temperature zone; a simulation based on the process parameters using a multi-physics coupling digital twin model, and obtaining key characteristic quantities in the carbon fiber heat treatment process based on the simulation results; a simulation optimization unit for using the process parameters and the key characteristic quantities as input features, inputting them into a pre-defined multi-physics coupling-based carbon fiber heat treatment model, and predicting the performance indicators of the carbon fiber; using a multi-objective optimization algorithm to perform collaborative optimization of the process parameters with the optimization objectives of maximizing fiber tensile strength, maximizing elastic modulus, and minimizing energy consumption efficiency, to obtain a Pareto optimal process parameter set; and an execution unit for controlling the carbon fiber heat treatment process according to the Pareto optimal process parameter set.

[0015] Thirdly, the present invention provides an electronic device, the electronic device including a memory, a processor, and a carbon fiber heat treatment optimization program based on multiphysics coupling stored in the memory and executable on the processor, wherein when the carbon fiber heat treatment optimization program based on multiphysics coupling is executed by the processor, it implements the carbon fiber heat treatment optimization method based on multiphysics coupling as described above.

[0016] Fourthly, the present invention provides a computer-readable storage medium storing a computer program, which, when executed by a processor, implements the carbon fiber heat treatment optimization method based on multiphysics coupling as described above.

[0017] This invention discloses a method, system, and storage medium for optimizing carbon fiber heat treatment based on multiphysics coupling. Key features reflecting the evolution of microstructure are extracted through multiphysics coupling simulation and embedded as interaction terms into a machine learning model, significantly improving the model's physical interpretability and prediction accuracy and generalization ability under small sample conditions. Parametric order reduction modeling techniques are employed to simplify the complex multiphysics model, greatly reducing computational costs while preserving key physical information. This makes it possible to use simulation data to support machine learning model training, overcoming the efficiency bottleneck of directly using high-fidelity simulations for optimization. A digital twin system enables real-time adjustment and closed-loop control of process parameters during carbon fiber heat treatment.

[0018] To achieve the foregoing and related objectives, one or more aspects of the invention include the features that will be described in detail below. The following description and accompanying drawings illustrate certain exemplary aspects of the invention. However, these aspects indicate only a few of the various ways in which the principles of the invention can be used. Furthermore, the invention is intended to encompass all such aspects and their equivalents. Attached Figure Description

[0019] Other objects and results of the invention will become more apparent and readily understood with reference to the following description taken in conjunction with the accompanying drawings. In the drawings: Figure 1 This is a schematic flowchart of a carbon fiber heat treatment optimization method based on multiphysics coupling according to an embodiment of the present invention. Figure 2 This is a schematic diagram illustrating the principle of the carbon fiber heat treatment optimization method based on multi-physics coupling according to an embodiment of the present invention. Figure 3 A schematic diagram of a carbon fiber heat treatment optimization system based on multi-physics field coupling provided according to an embodiment of the present invention; Figure 4This is a schematic diagram of the internal structure of an electronic device that implements a carbon fiber heat treatment optimization method based on multi-physics field coupling according to an embodiment of the present invention.

[0020] In all the accompanying drawings, the same reference numerals indicate similar or corresponding features or functions. Detailed Implementation

[0021] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0022] The technical solutions in the embodiments of this application will be clearly and thoroughly described below with reference to the accompanying drawings. In the description of the embodiments of this application, unless otherwise stated, the word "and / or" in the text is merely a description of the relationship between related objects, indicating that three relationships can exist. For example, A and / or B can represent: A existing alone, A and B existing simultaneously, and B existing alone.

[0023] Hereinafter, the terms "first" and "second" are used for descriptive purposes only and should not be construed as implying or suggesting relative importance or implicitly indicating the number of indicated technical features. Thus, a feature defined as "first" or "second" may explicitly or implicitly include one or more of that feature. Furthermore, in the description of embodiments in this application, "multiple" refers to two or more.

[0024] References to "one embodiment" or "some embodiments" as described in this specification mean that one or more embodiments of this application include a specific feature, structure, or characteristic described in connection with that embodiment. Therefore, the phrases "in one embodiment," "in some embodiments," "in other embodiments," "in still other embodiments," etc., appearing in different parts of this specification do not necessarily refer to the same embodiment, but rather mean "one or more, but not all, embodiments," unless otherwise specifically emphasized. The terms "comprising," "including," "having," and variations thereof mean "including but not limited to," unless otherwise specifically emphasized.

[0025] To provide a detailed description of the carbon fiber heat treatment optimization method, system, and storage medium based on multiphysics coupling of the present invention, specific embodiments of the present invention will be described in detail below with reference to the accompanying drawings.

[0026] A digital twin is a simulation process that integrates multiple disciplines, physical quantities, scales, and probabilities, utilizing data such as physical models, sensor updates, and operational history to map the entire lifecycle of corresponding physical equipment in a virtual space. A digital twin is a concept that transcends reality; it can be viewed as a digital mapping system of one or more important, interdependent equipment systems. Machine learning models are broadly classified into supervised learning and unsupervised learning based on the types of data they can be used with.

[0027] Supervised learning primarily includes models for classification and models for regression. Examples include linear classifiers (such as Logistic Regression), Support Vector Machines (SVM), Naive Bayes (NB), K-Nearest Neighbors (KNN), Decision Trees (DT), and ensemble models (RF / GDBT, etc.). Unsupervised learning primarily includes data clustering (K-means) and data dimensionality reduction (PCA), etc.

[0028] During the heat treatment of carbon fiber, there is a strong nonlinear coupling effect among the temperature field, tension field, atmospheric environment, and the chemical reaction field occurring inside the fiber. These factors jointly determine the final microstructure (such as crystallite size, orientation, and core-sheath structure) and macroscopic mechanical properties (such as tensile strength and elastic modulus) of the carbon fiber. This invention addresses the continuous and unique characteristics of carbon fiber production by incorporating an embedded automatic matching system to achieve efficient production during the heating or product changeover stages of the production line, thereby enabling effective cost control. Example 1

[0029] Figure 1 A flowchart of a carbon fiber heat treatment optimization method based on multiphysics coupling according to an embodiment of the present invention is shown.

[0030] like Figure 1 As shown, the carbon fiber heat treatment optimization method based on multi-physics field coupling in this embodiment of the invention includes steps S110-S140.

[0031] S110: Obtain process parameters during the carbon fiber heat treatment process; wherein, the process parameters include temperature, tension, fiber linear velocity and furnace atmosphere concentration in each temperature zone.

[0032] In the specific implementation process, production line process parameters are collected in real time and transmitted to the digital twin system of this invention; the process parameter set includes the temperature history of each temperature zone. T zone (t) ,tension F(t) linear velocity vThe goal is to obtain more optimized process parameters based on the current ones, and then control the carbon fiber heat treatment process based on these optimized parameters. For example, a Siemens S7-1500 PLC is used, equipped with a 32-channel analog input module for acquiring signals such as temperature, tension, and atmosphere concentration, and a 16-channel digital / analog output module for outputting control signals. Taking typical polyacrylonitrile (PAN)-based carbon fiber precursor fibers as the research object, simultaneous thermal analysis-mass spectrometry (STA-MS) is used to measure the exothermic curves and gas escape behavior (such as HCN) of the precursor fibers during the pre-oxidation stage at 200-400°C. The modulus changes of fibers at different temperatures were obtained using a dynamic mechanical analyzer (DMA).

[0033] S120: Using a digital twin model coupled with multi-physics fields, simulations are performed based on the process parameters. Key characteristic quantities in the carbon fiber heat treatment process are obtained based on the simulation results and experimental test results.

[0034] It should be noted that, based on the reduced-order multiphysics coupling model, key characteristic quantities of carbon fibers during the heat treatment process are extracted through numerical simulation and symbolic regression analysis, combined with experimental sample test data. The key characteristic quantities involve the microstructure of carbon fibers (such as crystallite size, orientation, core-sheath structure, etc.). Specifically, the key characteristic quantities include, but are not limited to: fiber radial temperature gradient, stress difference between the sheath and the core, cyclization reaction conversion rate, crystallite size evolution rate, and characteristic parameters reflecting the differences in core-sheath structure.

[0035] The key characteristic quantities also include cross-scale energy state characteristics, evolution path discrimination characteristics, by-product / main reaction rate ratio, interfacial coupling stress characteristics, and structural anisotropic evolution characteristics; the cross-scale energy state characteristics include the power spectral density of the radial temperature field.

[0036] The evolution path discrimination features include the microcrystalline ordering rate and the byproduct / main reaction rate ratio; the microcrystalline ordering rate is achieved by the following formula.

[0037] in, The rate of microcrystalline ordering; I G The intensity of the G peak for carbon materials; I D The intensity of the D peak of the carbon material.

[0038] The byproduct / main reaction rate ratio is achieved by the following formula. , in, The instantaneous rate of the main reaction; The release rate of key byproducts.

[0039] The interface coupling stress characteristics are realized by the following formula. , Where K is a constant related to the elastic modulus E, Poisson's ratio ν, and geometric dimensions of the two-layer material; and These represent the average intrinsic strains of the core and the skin, respectively.

[0040] The anisotropic evolution characteristics of the structure, including the orientation consistency index, are achieved through the following formula. , Among them, f H For Hermann orientation factor; The width of the orientation distribution; It is a constant.

[0041] In the specific implementation process, a cross-scale coupled model covering the pre-oxidation and carbonization processes of carbon fiber heat treatment was established based on the COMSOL Multiphysics software platform.

[0042] For the pre-oxidation stage of carbon fiber heat treatment, the modules of "Heat Source," "Solid Mechanics," and "Chemical Reaction" are coupled. The chemical reaction uses a custom PDE, with input cyclization reaction kinetic parameters obtained by fitting STA experimental data. Simultaneously, the exothermic cyclization reaction is introduced into the "Heat Source" module, and the volume change and thermal stress caused by chemical shrinkage are considered in the "Solid Mechanics" module. For the carbonization stage of carbon fiber heat treatment (400-1500°C), the diffusion process of small molecule gases is further added, and its influence on the temperature field is considered. The mechanical behavior of the fiber is described using a modified viscoelastic constitutive model.

[0043] Finally, a multiphysics coupled digital twin model is constructed and subjected to parameterized order reduction. Specifically, intrinsic orthogonal decomposition (POD) is performed to extract the main modes characterizing the spatiotemporal evolution of field variables (temperature, stress, and concentration). Subsequently, radial basis functions (RBF) are used to establish the mapping relationship between process parameters (such as heating rate, tension, and atmosphere concentration) and POD mode coefficients, thus forming a computationally highly efficient parameterized order reduction model (ROM). The completed multiphysics coupled digital twin model can complete a simulation within seconds, providing a large amount of data for subsequent steps.

[0044] S130: Using the process parameters and key feature quantities as input features, input a preset carbon fiber heat treatment model based on multi-physics coupling to predict the performance indicators of carbon fiber; use a multi-objective optimization algorithm to optimize the process parameters in a coordinated manner with the optimization objectives of maximizing fiber tensile strength, maximizing elastic modulus and minimizing energy consumption efficiency, to obtain a Pareto optimal set of process parameters.

[0045] Specifically, a 5-layer residual neural network (ResNet) is constructed. The input layer contains 27 dimensions, including: (a) basic process parameters such as set temperatures for each temperature zone, fiber feed rate, draw ratio, and furnace atmosphere concentration; and (b) key feature quantities obtained from S102 are embedded as physical interaction terms. After the first hidden layer, a physical verification unit based on Fourier's law is embedded to constrain the consistency of the temperature-related intermediate variables predicted by the network. After the third hidden layer, a chemical reaction constraint module based on the Arrhenius equation is added to ensure that the relationship between the predicted reaction rate and temperature conforms to basic chemical laws. Before the output layer, a regularization term based on viscoelastic constitutive model is introduced to constrain the predicted stress-strain relationship. Then, model pre-training is performed, using the ROM of S101 to generate 10,000 sets of simulation data (input being process parameters, output being key feature quantities and performance indicators) to pre-train the network. The loss function is weighted mean square error, with a weight ratio of temperature field:stress field:chemical field = 3:2:1. After training, the model was fine-tuned using 500 sets of rigorously quality-controlled real-world production line experimental data. During this stage, a "physical consistency loss term" was added to the loss function to penalize predictions that violated the underlying physical laws. After training, the model's prediction error for fiber tensile strength on the test set was within ±0.2 GPa, and the tensile modulus error was within ±5 GPa, with R² > 0.90.

[0046] In the specific implementation process, the method for obtaining the multi-physics coupled carbon fiber heat treatment model includes steps S1301-S1304.

[0047] S1301. Construct a digital twin model of multi-physics coupling, wherein the physical fields include temperature field, stress field and chemical reaction field; wherein, a parameterized order reduction modeling method is used to reduce the order of the multi-physics coupling model.

[0048] A multi-physics coupled model for carbon fiber heat treatment, covering the entire process of pre-oxidation, low-temperature carbonization, and high-temperature carbonization, is established. The physical fields include at least temperature field, stress field, and chemical reaction field. A parameterized order reduction modeling method is used to reduce the order of the multi-physics coupled model to improve computational efficiency.

[0049] In other words, the digital twin model of multi-physics coupling is a neural network architecture, which includes a physical constraint subnet; the physical constraint subnet is constructed based on the energy conservation equation, the chemical reaction kinetics equation, and the linear momentum conservation equation.

[0050] The temperature field is constructed based on the energy conservation equation through the following governing equations:

[0051] Where ρ is density, C p For specific heat capacity, k is the thermal conductivity. q reac For the exothermic / endothermic source term of the chemical reaction, represents the conversion rate. and temperature T The function. q vis The heat source term generated by viscous dissipation, and the strain rate Related.

[0052] The chemical reaction kinetics equations are used to construct the chemical reaction field through the following governing equations:

[0053] in, A Pre-exponential factor, E a For activation energy, R is the ideal gas constant. This is a function that describes the reaction mechanism.

[0054] The stress field is constructed using the following governing equations based on the linear momentum conservation equation:

[0055] in, C Let be the elastic stiffness tensor. and These are thermal strain and chemical contraction strain, respectively. α and β These are the coefficients of thermal expansion and chemical shrinkage, respectively.

[0056] It should be noted that the core of the hybrid order reduction strategy guided by physical mechanisms lies in designing a hybrid loss function.

[0057] During the training of the multiphysics-coupled digital twin model using the carbon fiber heat treatment dataset, the loss function is implemented using the following formula. , in, , , It is a constant.

[0058] L data For data loss; , L pde This represents a loss due to physical mechanisms; this is key to embedding physical laws as "hard constraints." Random sampling is performed across the entire computational domain. Np residual points {x j , t j The network's predictions must approximately satisfy the governing equations.

[0059] , L bc Loss is for boundary / initial conditions; , in, N d This represents the total number of training data points; these data come from high-fidelity simulations or limited experimental measurements. i For the index of the data points, traverse from 1 to... N d . x i ,t i ,p i represents the input coordinates of the i-th data point; represents the spatial coordinates, time coordinates, and process parameter vector, respectively. T(x i ,t i ,p i ) , These are the predicted values ​​from the neural network. Given input coordinates and process parameters, the neural network outputs predicted values ​​for temperature and reaction conversion rate.

[0060] For the first i The true values ​​of each data point come from high-fidelity simulation or experimental measurements. N p This represents the total number of residuals. j This is the index of the residual point. ρ The density of the material. C p This represents the specific heat capacity of the material. k The thermal conductivity of the material. This is the heat source term for the chemical reaction. It represents the reaction conversion rate. and temperature T The function is defined according to the chemical kinetic model. N b This represents the total number of boundary points and initial condition points. k The index of the boundary / initial point. k , t k , p k For the first k The coordinates of the boundary / initial point. For the boundary point x... k Located on the geometric boundary of the computational domain (such as a fiber surface). t k For any time period, the initial point is... t k =0 (start of process), x k Let be any point within the domain. T (x k , t k , p k ) represents the neural network's prediction at the boundary / initial point. T bc,k For the first k The known values ​​specified by the boundary / initial point.

[0061] The parametric order reduction modeling method employs a combination of intrinsic orthogonal decomposition (POD) and radial basis functions (RBF). A POD is performed on the multiphysics coupled digital twin model to extract the main modes characterizing the field variables (temperature, stress, concentration). Subsequently, RBF is used to establish a mapping relationship between process parameters (such as heating rate, tension, and atmosphere concentration) and the POD mode coefficients, thus forming a computationally efficient parametric order reduction model (ROM). This ROM can complete a simulation within seconds, providing a large amount of data for subsequent steps. In summary, the parametric order reduction modeling technique simplifies complex multiphysics models, significantly reducing computational costs while preserving key physical information. This makes it possible to use simulation data to support machine learning model training, solving the efficiency bottleneck of directly using high-fidelity simulations for optimization.

[0062] S1302. Based on the digital twin model of the multi-physics coupling, the characteristic vector of carbon fiber heat treatment is obtained through numerical simulation and symbolic regression analysis.

[0063] The eigenvectors include stress field characteristics, chemical reaction characteristics, and structural evolution characteristics. Stress field characteristics include the maximum tensile stress along the fiber axis and the stress difference between the skin and core layers. Chemical reaction characteristics include the cyclization conversion rate at specific temperature points (e.g., 260°C, 300°C). Structural evolution characteristics are the crystallite size (L) estimated from temperature and stress history using empirical formulas. a , L c The initial values ​​and their evolution rates. These characteristic quantities together constitute the connection process, the bridge between the final microstructure and the final microstructure.

[0064] The key features also include cross-scale energy state features, fiber radial thermal distribution features, structural evolution features, core-sheath structure features, and structural anisotropic evolution features; the cross-scale energy state features include the power spectral density of the radial temperature field. Among these, the fiber radial thermal distribution features not only monitor the radial temperature gradient of the fiber, but more importantly, extract its time-varying power spectral density to quantify the impact of thermal fluctuations at different spatial frequencies on the uniformity of chemical reactions, thereby correlating with the formation of surface defects in the final fiber.

[0065] radial temperature field T (r,t) First, the temperature distribution in the radial direction (r direction) of the fiber is obtained through a reduced-order model.

[0066] radial temperature gradient G(r,t) : Computation of the first derivative in space .

[0067] Power spectral density PSD(f,t) : Radial temperature gradient signal at a certain time t G(r,t) Perform a spatial Fourier transform along spatial coordinate r and calculate its power spectrum. This transforms it from the spatial domain to the frequency domain.

[0068]

[0069] .

[0070] in, PSD(f,t) This describes the intensity of temperature fluctuation energy at spatial frequency f. High-frequency components correspond to sharp temperature abrupt changes, possibly related to surface cracks; low-frequency components correspond to gentle temperature gradients, affecting the overall formation of the core-skin structure. The cyclization reaction conversion rate was determined by infrared spectroscopy to measure the ratio of C,N double to triple bonds, and the crystallite size evolution rate was obtained by differentiating the Miller equation.

[0071] Regarding structural evolution characteristics, the microcrystalline ordering rate based on the dynamic changes in Raman spectral peak shift and full width at half maximum (FWHM), and the ratio of byproduct release rate to main reaction rate during cyclization reaction are introduced as key characteristic quantities. These two ratio parameters can sensitively distinguish whether the heat treatment process is on an ideal steady-state optimization path or deviating from an abnormal path, thereby achieving early warning.

[0072] The evolution path discrimination features include the microcrystalline ordering rate and the byproduct / main reaction rate ratio; the microcrystalline ordering rate is achieved by the following formula.

[0073] in, The rate of microcrystalline ordering; I G The intensity of the G peak for carbon materials; I D The intensity of the D peak for carbon materials is given. Based on online Raman spectroscopy data, the D peak of carbon materials (~1350) is typically used. (representing disordered structure) and G peak (~1580) (representing the graphite lattice). This rate quantifies the dynamic changes in the graphitization process, and its time integral is the final degree of order.

[0074] The byproduct / main reaction rate ratio is achieved by the following formula.

[0075] in, It is the instantaneous rate of the main reaction; it can be measured by the reaction conversion rate. The derivative with respect to time is obtained as follows: .

[0076] The release rate of key byproducts; their concentration can be measured using online detection equipment such as mass spectrometers. C byproduct And differentiate: byproduct (t)=dC byproduct / dt .

[0077] For the core-sheath structure, the evolution of the "sheath-core interface shear stress" induced by the combined effects of chemical shrinkage and thermal expansion coefficient mismatch is extracted, rather than the static stress difference. This dynamic shear stress history is a direct indicator for predicting the initiation and development of internal fiber cracks.

[0078] First, the fiber is simplified into a two-layer composite cylindrical model, and the intrinsic strain difference is set: chemical shrinkage and thermal strain act together on the intrinsic strain.

[0079] Interfacial shear stress is based on a shear hysteresis model, and refers to the interfacial shear stress between the cortex and the core. It is directly related to the intrinsic strain difference between the two layers; it is achieved through the following formula:

[0080] Wherein, K is a constant related to the elastic modulus E, Poisson's ratio ν, and geometric dimensions of the two-layer material; and These are the average intrinsic strains of the core layer and the skin layer, respectively.

[0081] Regarding the anisotropic evolution characteristics of the structure, this index quantifies the change in the orientation distribution function of graphite microcrystals along the fiber axis using online wide-angle X-ray diffraction data, and couples it with the aforementioned interfacial shear stress field in real time to predict the final compressibility of high-modulus fibers.

[0082] The anisotropic evolutionary characteristics of the structure include an orientation consistency index, which is achieved through the following formula.

[0083] in, f H The Hermann orientation factor is first extracted from the online wide-angle X-ray diffraction pattern. It is a standard parameter that measures the degree of orientation of the microcrystal relative to the fiber axis. The orientation distribution width is usually represented by the half-width at half maximum (WHM) of the orientation distribution function; It is a constant, a small constant added to prevent the denominator from being zero.

[0084] It should be noted that the higher the OCI value, the more ideal the microstructure of the carbon fiber. This is related to interfacial shear stress. Coupled calculations can establish a quantitative prediction model of process stress history and final compression performance.

[0085] S1303. Construct a machine learning model with embedded physical constraints, and train the machine learning model using a carbon fiber heat treatment dataset to obtain a trained multi-physics coupled carbon fiber heat treatment model; wherein, the process parameter set and the carbon fiber heat treatment feature vector are used as input features of the multi-physics coupled carbon fiber heat treatment model, and the carbon fiber performance index is used as output features of the multi-physics coupled carbon fiber heat treatment model.

[0086] A machine learning model with embedded physical constraints is constructed. The process parameter set and key structural features are used as input features, and the macroscopic performance indicators of carbon fiber are used as output. The model is trained using multiphysics simulation data and experimental data, enabling it to learn the precise mapping relationship between process parameters, structural features, and final performance. In other words, the model input consists of two parts: the process parameter set P and the key structural features S. The process parameter set includes the temperature history T for each temperature region. zone The parameters include tensile strength (F(t)), tensile modulus (F(t)), and linear velocity (v). Key structural features S include higher-order features extracted from the reduced-order model that represent the evolution of the material's internal state. In specific implementation, the macroscopic performance indicators of carbon fiber include tensile strength, tensile modulus, and density. The carbon fiber heat treatment dataset includes simulation data and experimental data generated using a multiphysics coupled digital twin model.

[0087] .

[0088] S1304. The trained multi-physics coupled carbon fiber heat treatment model is optimized using the improved dung beetle algorithm.

[0089] In other words, the trained multiphysics coupled carbon fiber heat treatment model utilizes an improved dung beetle algorithm to optimize process parameters. Using a trained machine learning model combined with a multi-objective optimization algorithm, with fiber tensile strength, elastic modulus, and energy efficiency as optimization objectives, the heat treatment process parameters are collaboratively optimized to obtain a Pareto-optimal set of process parameters. The multi-objective optimization algorithm employs an improved dung beetle algorithm and integrates a random forest algorithm for feature importance evaluation and selection.

[0090] In a specific example, multi-objective process optimization is performed based on a machine learning model.

[0091] The trained machine learning model is used as a surrogate model and integrated into the optimization framework. The goal is to maximize the fiber tensile strength (…). ), maximize tensile modulus ( Minimize unit energy consumption ( Set it as the optimization target.

[0092] An improved Dung Beetle Optimizer (DBO) is employed for multi-objective optimization. For example, the algorithm parameters can be a population size of 50 and a maximum of 500 iterations. To enhance global search capability, a dynamic adjustment mechanism for inertia weights is introduced, and Cauchy mutation is performed on individuals with a 10% probability. Through optimization, a series of non-dominated solutions, i.e., a Pareto optimal set, is obtained. Production operators can select the most suitable combination of process parameters from this frontier according to actual needs.

[0093] S140: Control the carbon fiber heat treatment process according to the Pareto optimal process parameter set.

[0094] The carbon fiber heat treatment optimization method based on multi-physics coupling of the present invention further includes, S150: collecting heat treatment operation data and carbon fiber product performance data adjusted according to the Pareto optimal process parameter set; updating the digital twin model of multi-physics coupling and the carbon fiber heat treatment model based on the heat treatment operation data and carbon fiber product performance data.

[0095] In the specific implementation process, the aforementioned multi-physics coupled digital twin model and the carbon fiber heat treatment model based on multi-physics coupling are integrated into a single digital twin system. For example, a three-layer distributed software architecture is constructed. First, a digital twin layer is built, deploying a parameterized reduced-order model developed using S101 on the ANSYS Twin Builder platform. This layer communicates with the physical production line in real time via the OPC UA over TSN protocol, receiving sensor data from the production line (such as actual temperature and tension) and dynamically updating the virtual model state to achieve real-time, high-fidelity mapping of the production line. Second, an intelligent optimization layer is built. This layer runs machine learning models and optimization algorithms. It periodically (e.g., every 10 minutes) obtains the current state from the digital twin layer and calculates the optimal process parameter settings for the next time period based on the latest model and optimization objectives. Finally, a control execution layer is included. Specifically, COMSOL is used for model building, and ANSYS is used for digital twin layer deployment. A Siemens S7-1500 series PLC is used as the lower-level machine. By receiving instructions from the intelligent optimization layer and employing a model predictive control (MPC) algorithm, the system drives actuators such as heaters, drafting rollers, and atmosphere flow valves to precisely adjust process parameters to target values, completing closed-loop control. Furthermore, the computing server can be a Dell PowerEdge R750 server equipped with an NVIDIA A100 GPU for running multiphysics ROMs and training and inferring large neural networks. The industrial control network can utilize an industrial-grade Profine network, divided into three IEEE 802.1Q VLANs, with DSCP 46 priority marking and bandwidth reservation ensuring high-quality real-time communication between the digital twin layer, optimization layer, and control execution layer.

[0096] Figure 2 This is a schematic diagram illustrating the principle of a carbon fiber heat treatment optimization method based on multiphysics coupling. Figure 2As shown, the first part involves constructing a parameterized multi-physics coupling digital twin model and a mechanistic multi-physics coupling carbon fiber heat treatment model. Then, the parameterized reduced-order digital twin model and the mechanistic machine learning model are integrated, and the interaction effects in the multi-field coupling are analyzed. Simulation is used to understand and quantify how physical fields interact, and key feature quantities are extracted from the simulation. To achieve a weight allocation model for the bidirectional feedback between the non-uniform thermal stress field and the fiber microstructure, as well as the difference between the core and skin structure, key feature quantities are used as model inputs. The second part involves establishing a multi-layer neural network including a physical constraint subnet module, introducing physical field weights for data training: data preparation and model training verification. During model training, considerations of physical laws are incorporated to enhance the model's interpretability. Ultimately, the predicted key performance index R>0.90 is achieved. The third part, to implement the concepts of the first and second parts, constructs an operational framework capable of real-time optimization and closed-loop control. This involves building the system's operational skeleton, consisting of a digital twin layer, an intelligent optimization layer, and a control execution layer.

[0097] Specifically, the model framework is a proposed physical information feedforward network architecture, and its computational flow is as follows: Feedforward network basic mapping: This includes several hidden layers: ； ； Where σ is the activation function, and W... l ,b l These are the weights and biases of the l-th layer. The above describes common construction methods and processes.

[0098] To obtain more reasonable and accurate results, this patent introduces a physical constraint layer. This is achieved through a differentiable correction function: .

[0099] This system is used to map the production line status in real time and dynamically adjust process parameters based on optimization results, achieving intelligent closed-loop control of the carbon fiber heat treatment process. Considering the continuous and unique nature of carbon fiber production, an embedded automatic matching system is implemented to ensure efficient production during heating or product changeover stages, thereby achieving effective cost control. The control and execution layer of the digital twin system drives the production line actuators, dynamically adjusting temperature, tension, and atmosphere concentration to achieve real-time closed-loop control of the carbon fiber heat treatment process.

[0100] like Figure 3As shown, this invention provides a carbon fiber heat treatment optimization system based on multiphysics coupling, which is evaluated using the carbon fiber heat treatment optimization method based on multiphysics coupling as described above. Depending on the functions implemented, the carbon fiber heat treatment optimization system 300 based on multiphysics coupling may include a detection data acquisition unit 310, a simulation optimization unit 320, and an execution unit 330. The unit in this invention can also be called a module, which refers to a series of computer program segments that can be executed by an electronic device processor and can perform a fixed function, stored in the memory of the electronic device.

[0101] In this embodiment, the functions of each module / unit are as follows: The detection data acquisition unit 310 is used to acquire process parameters during the heat treatment of carbon fiber; wherein, the process parameters include temperature, tension, fiber linear velocity and furnace atmosphere concentration in each temperature zone; The simulation optimization unit 320 uses a multi-physics coupled digital twin model to perform simulations based on the process parameters, and obtains key characteristic quantities in the carbon fiber heat treatment process based on the simulation results. It uses the process parameters and key characteristic quantities as input features to input a pre-defined multi-physics coupled carbon fiber heat treatment model to predict the performance indicators of the carbon fiber. It then uses a multi-objective optimization algorithm to collaboratively optimize the process parameters with the optimization objectives of maximizing fiber tensile strength, maximizing elastic modulus, and minimizing energy consumption efficiency, thereby obtaining a Pareto optimal set of process parameters. The execution unit 330 is used to control the carbon fiber heat treatment process according to the Pareto optimal process parameter set.

[0102] More specific implementations of the carbon fiber heat treatment optimization system based on multi-physics coupling can be found in the foregoing description of the embodiments of the carbon fiber heat treatment optimization method based on multi-physics coupling, and will not be detailed here.

[0103] like Figure 4 As shown, the present invention also provides an electronic device 1 based on a carbon fiber heat treatment optimization method using multi-physics coupling.

[0104] The electronic device 1 may include a processor 10, a memory 11, and a bus. It may also include a computer program stored in the memory 11 and executable on the processor 10, such as a carbon fiber heat treatment optimization program 12 based on multiphysics coupling. The memory 11 may include both internal storage units for the carbon fiber heat treatment optimization system based on multiphysics coupling and external storage devices. The memory 11 can be used not only to store application software and various types of data, such as the code for the carbon fiber heat treatment optimization program based on multiphysics coupling, but also to temporarily store data that has been output or will be output.

[0105] The memory 11 includes at least one type of readable storage medium, including flash memory, portable hard drive, multimedia card, card-type memory (e.g., SD or DX memory), magnetic memory, magnetic disk, optical disk, etc. In some embodiments, the memory 11 can be an internal storage unit of the electronic device 1, such as a portable hard drive. In other embodiments, the memory 11 can be an external storage device of the electronic device 1, such as a plug-in portable hard drive, smart media card (SMC), secure digital card (SD), flash card, etc., equipped on the electronic device 1. Furthermore, the memory 11 can include both internal and external storage units of the electronic device 1. The memory 11 can be used not only to store application software and various types of data installed on the electronic device 1, such as code for a single mechanical component life prediction program for aircraft, but also to temporarily store data that has been output or will be output.

[0106] In some embodiments, the processor 10 may be composed of integrated circuits, such as a single packaged integrated circuit or multiple integrated circuits with the same or different functions, including combinations of one or more central processing units (CPUs), microprocessors, digital processing chips, graphics processors, and various control chips. The processor 10 is the control unit of the electronic device, connecting various components of the entire electronic device through various interfaces and lines. It executes programs or modules stored in the memory 11 (e.g., a carbon fiber heat treatment optimization program based on multiphysics coupling) and calls data stored in the memory 11 to perform various functions of the electronic device 1 and process data.

[0107] The bus can be a peripheral component interconnect (PCI) bus or an extended industry standard architecture (EISA) bus, etc. This bus can be divided into an address bus, a data bus, a control bus, etc. The bus is configured to enable communication between the memory 11 and at least one processor 10, etc.

[0108] Figure 4 Only electronic devices with components are shown; those skilled in the art will understand that... Figure 4The structure shown does not constitute a limitation on the electronic device 1, and may include fewer or more components than shown, or combine certain components, or have different component arrangements.

[0109] For example, although not shown, the electronic device 1 may also include a power supply (such as a battery) to power the various components. Preferably, the power supply can be logically connected to the at least one processor 10 through a power management system, thereby enabling functions such as charging management, discharging management, and power consumption management through the power management system. The power supply may also include one or more DC or AC power supplies, recharging systems, power fault detection circuits, power converters or inverters, power status indicators, and other arbitrary components. The electronic device 1 may also include various sensors, Bluetooth modules, Wi-Fi modules, etc., which will not be described in detail here.

[0110] Furthermore, the electronic device 1 may also include a network interface. Optionally, the network interface may include a wired interface and / or a wireless interface (such as a Wi-Fi interface, a Bluetooth interface, etc.), which is typically used to establish communication connections between the electronic device 1 and other electronic devices.

[0111] Optionally, the electronic device 1 may further include a user interface, which may be a display, an input unit (such as a keyboard), and optionally, a standard wired interface or a wireless interface. Optionally, in some embodiments, the display may be an LED display, a liquid crystal display, a touch-sensitive liquid crystal display, or an OLED (Organic Light-Emitting Diode) touchscreen, etc. The display may also be appropriately referred to as a screen or display unit, used to display information processed in the electronic device 1 and to display a visual user interface.

[0112] It should be understood that the embodiments described are for illustrative purposes only and are not limited to this structure in the scope of the patent application.

[0113] The carbon fiber heat treatment optimization program 12 based on multiphysics coupling, stored in the memory 11 of the electronic device 1, is a combination of multiple instructions. When run in the processor 10, it can: acquire process parameters in the carbon fiber heat treatment process; wherein, the process parameters include temperature, tension, fiber linear velocity, and furnace atmosphere concentration in each temperature zone; perform simulation based on the process parameters using a digital twin model of multiphysics coupling, and acquire key feature quantities in the carbon fiber heat treatment process based on the simulation results; use the process parameters and the key feature quantities as input features, input to a preset carbon fiber heat treatment model based on multiphysics coupling, and predict the performance indicators of the carbon fiber; use a multi-objective optimization algorithm to perform collaborative optimization of the process parameters with the optimization objectives of maximizing fiber tensile strength, maximizing elastic modulus, and minimizing energy consumption efficiency, to obtain a Pareto optimal process parameter set; and control the carbon fiber heat treatment process according to the Pareto optimal process parameter set.

[0114] Specifically, the processor 10's implementation method for the above instructions can be found in [reference needed]. Figure 1 The descriptions of the relevant steps in the corresponding embodiments are not repeated here. Furthermore, if the modules / units integrated in the electronic device 1 are implemented as software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. The computer-readable medium may include: any entity or system capable of carrying the computer program code, a recording medium, a USB flash drive, a portable hard drive, a magnetic disk, an optical disk, a computer memory, or a read-only memory (ROM).

[0115] This invention also provides a computer-readable storage medium, which can be non-volatile or volatile. The storage medium stores a computer program, which, when executed by a processor, performs the following: acquiring process parameters during the carbon fiber heat treatment process; wherein the process parameters include temperature, tension, fiber linear velocity, and furnace atmosphere concentration in each temperature zone; performing simulation based on the process parameters using a multi-physics coupled digital twin model, and acquiring key characteristic quantities during the carbon fiber heat treatment process based on the simulation results; using the process parameters and the key characteristic quantities as input features, inputting them into a preset multi-physics coupled carbon fiber heat treatment model to predict the performance indicators of the carbon fiber; using a multi-objective optimization algorithm to collaboratively optimize the process parameters with the optimization objectives of maximizing fiber tensile strength, maximizing elastic modulus, and minimizing energy consumption efficiency, obtaining a Pareto optimal process parameter set; and controlling the carbon fiber heat treatment process according to the Pareto optimal process parameter set.

[0116] Specifically, the specific implementation method of the computer program when executed by the processor can be referred to the description of the relevant steps in the embodiment of the carbon fiber heat treatment optimization method based on multi-physics field coupling, and will not be repeated here.

[0117] In the several embodiments provided by this invention, it should be understood that the disclosed devices, systems, and methods can be implemented in other ways. For example, the system embodiments described above are merely illustrative; for instance, the division of modules is only a logical functional division, and other division methods may be used in actual implementation.

[0118] The modules described as separate components may or may not be physically separate. The components shown as modules may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs.

[0119] Furthermore, the functional modules in the various embodiments of the present invention can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or in the form of hardware plus software functional modules.

[0120] It will be apparent to those skilled in the art that the present invention is not limited to the details of the exemplary embodiments described above, and that the present invention can be implemented in other specific forms without departing from the spirit or essential characteristics of the present invention.

[0121] Therefore, the embodiments should be considered exemplary and non-limiting in all respects, and the scope of the invention is defined by the appended claims rather than the foregoing description. Thus, all variations falling within the meaning and scope of equivalents of the claims are intended to be embraced within the invention. No appended diagram markings in the claims should be construed as limiting the scope of the claims.

[0122] Furthermore, it is clear that the word "comprising" does not exclude other units or steps, and the singular does not exclude the plural. Multiple units or systems stated in a system claim may also be implemented by a single unit or system through software or hardware.

[0123] However, those skilled in the art should understand that various modifications can be made to the carbon fiber heat treatment optimization method and system based on multiphysics field coupling proposed in this invention without departing from the scope of this invention. Therefore, the scope of protection of this invention should be determined by the appended claims.

Claims

1. A carbon fiber heat treatment optimization method based on multiphysics coupling, applied to electronic devices, characterized in that, include: Obtain process parameters for the heat treatment of carbon fiber; wherein, the process parameters include temperature, tension, fiber linear velocity and furnace atmosphere concentration in each temperature zone; A digital twin model with multi-physics coupling is used to simulate the process parameters, and key characteristic quantities in the carbon fiber heat treatment process are obtained based on the simulation results and experimental test data. The process parameters and key features are used as input features, and a preset carbon fiber heat treatment model based on multi-physics coupling is input to predict the performance indicators of carbon fiber. A multi-objective optimization algorithm is used to optimize the process parameters in a coordinated manner with the optimization objectives of maximizing fiber tensile strength, maximizing elastic modulus and minimizing energy consumption efficiency, so as to obtain the Pareto optimal process parameter set. The carbon fiber heat treatment process is controlled based on the Pareto optimal process parameter set.

2. The carbon fiber heat treatment optimization method based on multiphysics coupling according to claim 1, characterized in that, It also includes, Collect heat treatment operation data and carbon fiber product performance data after adjustment based on the Pareto optimal process parameter set; The digital twin model of multiphysics coupling and the carbon fiber heat treatment model based on the heat treatment operation data and carbon fiber product performance data are updated.

3. The carbon fiber heat treatment optimization method based on multiphysics coupling according to claim 1, characterized in that, The method for obtaining the multiphysics coupled carbon fiber heat treatment model includes, A digital twin model of multi-physics coupling is constructed, wherein the physical fields include temperature field, stress field and chemical reaction field; wherein, a parameterized order reduction modeling method is used to reduce the order of the multi-physics coupling model; Based on the digital twin model of multi-physics coupling, the characteristic vector of carbon fiber heat treatment is obtained through numerical simulation and symbolic regression analysis. The characteristic vector of carbon fiber heat treatment includes fiber radial temperature gradient, skin / core stress difference, cyclization reaction conversion rate, crystallite size evolution rate, and characteristic parameters reflecting the differences in skin and core structure. A machine learning model with embedded physical constraints is constructed and trained using a carbon fiber heat treatment dataset to obtain a trained multi-physics coupled carbon fiber heat treatment model. The process parameter set and the carbon fiber heat treatment feature vector are used as input features of the multi-physics coupled carbon fiber heat treatment model, and the carbon fiber performance index is used as output features of the multi-physics coupled carbon fiber heat treatment model. The trained multiphysics coupled carbon fiber heat treatment model was optimized using an improved dung beetle algorithm.

4. The carbon fiber heat treatment optimization method based on multiphysics coupling according to claim 3, characterized in that, The digital twin model of multiphysics coupling is a neural network architecture, which includes a physical constraint subnet. The physical constraint subnet is constructed based on the energy conservation equation, the chemical reaction kinetics equation, and the linear momentum conservation equation.

5. The carbon fiber heat treatment optimization method based on multiphysics coupling according to claim 1, characterized in that, The key features include cross-scale energy state features, evolution path discrimination features, by-product / main reaction rate ratio, interfacial coupling stress features, and structural anisotropic evolution features. The cross-scale energy state characteristics include the power spectral density of the radial temperature field; The evolution path discrimination features include the microcrystalline ordering rate and the byproduct / main reaction rate ratio; the microcrystalline ordering rate is achieved by the following formula. ， in, The rate of microcrystalline ordering; I G The intensity of the G peak for carbon materials; I D The intensity of the D peak of the carbon material; The byproduct / main reaction rate ratio is achieved by the following formula. ， in, The instantaneous rate of the main reaction; The release rate of key byproducts; The interface coupling stress characteristics are realized by the following formula. ， Where K is a constant related to the elastic modulus E, Poisson's ratio ν, and geometric dimensions of the two-layer material; and These are the average intrinsic strains of the core layer and the skin layer, respectively; The anisotropic evolution characteristics of the structure, including the orientation consistency index, are achieved through the following formula. ， Among them, f H For Hermann orientation factor; The width of the orientation distribution; It is a constant.

6. The carbon fiber heat treatment optimization method based on multiphysics coupling according to claim 3, characterized in that, During the training of the machine learning model using the carbon fiber heat treatment dataset, the loss function is implemented using the following formula. ， in, L data For data loss; L pde Loss due to physical mechanisms; L bc Loss is for boundary / initial conditions; , , It is a constant; ， ， ， in, N d This represents the total number of training data points. x i , t i , p i These are the spatial coordinates, time coordinates, and process parameter vector of the i-th data point, respectively. T ( x i , t i , p i ), These are the predicted values ​​of temperature and reaction conversion rate output by the neural network, respectively. T i , They were not the first i The true value of each data point; N p This represents the total number of residual points. ρ The density of the material; C p The specific heat capacity of the material; k The thermal conductivity of the material; This is the heat source term for the chemical reaction; N b This represents the total number of boundary points and initial condition points. k Index of the boundary / initial point; T (x k , t k , p k () represents the neural network's prediction at the boundary / initial point; T bc,k For the first k The known values ​​specified by the boundary / initial point.

7. The carbon fiber heat treatment optimization method based on multiphysics coupling according to claim 3, characterized in that, The carbon fiber heat treatment dataset includes simulation data and experimental data generated using a digital twin model coupled with multiphysics fields.

8. A carbon fiber heat treatment optimization system based on multiphysics coupling, characterized in that, The system includes: The detection data acquisition unit is used to acquire process parameters during the heat treatment of carbon fiber; wherein, the process parameters include temperature, tension, fiber linear velocity and furnace atmosphere concentration in each temperature zone; A digital twin model coupled with multiphysics is used to simulate the process parameters, and key characteristic quantities in the carbon fiber heat treatment process are obtained based on the simulation results. The simulation optimization unit is used to take the process parameters and key feature quantities as input features, input a pre-prepared carbon fiber heat treatment model based on multi-physics field coupling, and predict the performance indicators of carbon fiber; and use a multi-objective optimization algorithm to optimize the process parameters in a coordinated manner with the optimization objectives of maximizing fiber tensile strength, maximizing elastic modulus and minimizing energy consumption efficiency, so as to obtain the Pareto optimal process parameter set. An execution unit is used to control the carbon fiber heat treatment process according to the Pareto optimal process parameter set.

9. An electronic device, characterized in that, The system includes a memory, a processor, and a carbon fiber heat treatment optimization program based on multiphysics coupling stored in the memory and executable on the processor. When the carbon fiber heat treatment optimization prediction program based on multiphysics coupling is executed by the processor, it implements the carbon fiber heat treatment optimization method based on multiphysics coupling as described in any one of claims 1 to 7.

10. A computer-readable storage medium storing a computer program, characterized in that, When the computer program is executed by the processor, it implements the carbon fiber heat treatment optimization method based on multiphysics coupling as described in any one of claims 1 to 7.

Images