A Calculation Method for Thermochemical Ablation of Composite Materials Based on Physical Information Neural Networks

Abstract

This application provides a method for calculating the thermochemical ablation of composite materials based on a physical information neural network, relating to the field of macroscopic thermochemical ablation prediction technology. The method includes: constructing a macroscopic thermochemical ablation reaction equation for carbon-carbon composite materials; constructing a heat conduction equation for the thermochemical ablation of carbon-carbon composite materials; constructing a physical information neural network model; constructing a macroscopic geometric model of carbon-carbon composite materials; training the physical information neural network model; inputting the carbon-carbon composite material to be detected for macroscopic thermochemical ablation into the trained physical information neural network model to obtain the prediction result of the macroscopic thermochemical ablation behavior of the carbon-carbon composite material. This method, by embedding the physical laws of heat conduction into a neural network, achieves a deep integration of data-driven and physical constraints, enabling rapid and high-precision prediction of the thermochemical ablation of composite materials, and has great application prospects.

Description

Technical Field

[0001] This application relates to the field of macroscopic thermochemical ablation behavior prediction technology for carbon-carbon composite materials, and in particular to a calculation method for thermochemical ablation of composite materials based on physical information neural networks. Background Technology

[0002] In the aerospace field, spacecraft face extremely complex thermal environments. C / C composite materials, with their excellent high-temperature mechanical properties, low density, and thermal shock resistance, have become key materials for thermal protection systems. However, under conditions of high temperature and high-speed airflow, C / C composite materials undergo complex macroscopic thermochemical ablation phenomena, including material mass loss and changes in surface morphology. These directly affect the structural safety of the spacecraft and the success or failure of the mission. Accurately understanding the ablation law is crucial for the design of thermal protection systems. Currently, numerical calculation methods are mainly used to solve the macroscopic thermochemical ablation of C / C composite materials. On the one hand, empirical formula methods rely on fitting experimental data under specific operating conditions, resulting in poor generalization ability and an inability to effectively capture the multi-physics coupling effects such as heat conduction, gas diffusion, and chemical reactions, making it difficult to adapt to the complex and ever-changing actual flight environment. On the other hand, while the method of coupling computational fluid dynamics with ablation models can solve the interaction between the fluid and solid domains, it requires manually defining a large number of reaction mechanisms, diffusion coefficients, and other parameters, leading to extremely high computational costs. At the same time, the dynamic ablation boundary requires frequent mesh reconstruction, resulting in large numerical errors and reducing computational accuracy and efficiency.

[0003] Therefore, there is an urgent need for a new method to predict the macroscopic thermochemical ablation behavior of composite materials in order to solve the problems encountered in the existing technology. Summary of the Invention

[0004] This application provides a calculation method for thermochemical ablation of composite materials based on physical information neural networks, which achieves rapid and high-precision simulation of the ablation process by integrating data-driven and physical constraints.

[0005] According to the first aspect disclosed in this application, this application provides a method for calculating the thermochemical ablation of composite materials based on a physical information neural network, including:

[0006] Construct macroscopic thermochemical ablation reaction equations for carbon-carbon composite materials;

[0007] Based on the macroscopic thermochemical ablation reaction equation, a thermochemical ablation heat conduction equation for carbon-carbon composite materials is constructed.

[0008] Based on the macroscopic thermochemical ablation reaction equation and the thermochemical ablation heat conduction equation, a loss function is constructed using the mean square error method, and a physical information neural network model is built.

[0009] Construct a macroscopic geometric model of carbon-carbon composite materials;

[0010] The macroscopic geometric model of the carbon-carbon composite material is input into the physical information neural network model for iterative training to obtain the trained physical information neural network model.

[0011] The macroscopic thermochemical ablation behavior of the carbon-carbon composite material to be tested is input into the trained physical information neural network model to obtain the prediction results of the macroscopic thermochemical ablation behavior of the carbon-carbon composite material.

[0012] The beneficial effects of adopting the above technical solution are as follows: This application mainly addresses the problems of multi-physics coupling difficulties, low computational efficiency, and poor physical consistency in the numerical calculation of macroscopic thermochemical ablation of C / C composite materials. By embedding the physical laws of heat conduction into a neural network, it achieves a deep integration of data-driven and physical constraints, breaking through the dependence of traditional methods on ultra-fine meshes and a large amount of experimental data. It effectively solves the problems of high computational cost, complex dynamic boundary handling, and insufficient prediction accuracy for small samples. It provides an efficient and reliable computational means for the rapid design and precise optimization of thermal protection systems for hypersonic vehicles and spacecraft, and promotes the engineering application and technological development of C / C composite materials in extreme thermal environments.

[0013] In one feasible implementation, the macroscopic thermochemical ablation reaction equation for constructing the carbon-carbon composite material includes:

[0014] Based on the thermochemical ablation reaction mechanism of carbon-carbon composite materials and the elemental mass conservation equation and thermochemical equilibrium equation, a macroscopic thermochemical ablation reaction equation for carbon-carbon composite materials is constructed; specifically:

[0015]

[0016] in, It is the dimensionless mass ablation rate of the wall surface. Mass concentration of elements Mass concentration of elements Indicates the molecular weight of an element. Indicates the average molecular weight. It is the total pressure. This indicates the number of gas components considered in the surface system. Indicates the number of atoms of an element. This indicates partial pressure.

[0017] In one feasible implementation, the construction of a thermochemical ablation heat conduction equation for carbon-carbon composite materials based on the macroscopic thermochemical ablation reaction equation includes:

[0018] Based on the thermochemical ablation reaction process of carbon-carbon composite materials and the homogeneity and isotropy of carbon-carbon composite materials, a thermochemical ablation heat conduction equation is constructed; specifically:

[0019]

[0020] in, The density of carbon-carbon composite materials, t is the specific heat capacity of carbon-carbon composite material, where T is temperature, t is time, and k is thermal conductivity.

[0021] The second type of boundary condition is applied to the surface of the carbon-carbon composite material, specifically:

[0022]

[0023] Where q is the heat flux of the carbon-carbon composite material;

[0024] A third type of boundary condition is used at the lower boundary of the carbon-carbon composite material, specifically:

[0025]

[0026] Where k is the thermal conductivity and h is the convective heat transfer coefficient. The ambient temperature;

[0027] Based on the thermochemical ablation mass transfer of carbon-carbon composite materials, the mass ablation rate of the carbon-carbon composite surface is calculated according to the dimensionless mass ablation rate; specifically:

[0028]

[0029] in, To restore enthalpy, and for carbon-carbon composite materials ;

[0030] Based on the aforementioned mass ablation rate, the surface retreat rate of the carbon-carbon composite material was calculated; specifically:

[0031]

[0032] in, The surface receding rate of carbon-carbon composite materials.

[0033] In one feasible implementation, based on the macroscopic thermochemical ablation reaction equation and the thermochemical ablation heat conduction equation, a loss function is constructed using the mean square error method to build a physical information neural network model; including:

[0034] Based on the obtained macroscopic thermochemical ablation reaction equation and thermochemical ablation heat conduction equation, a loss function is constructed using the mean square error method. A physical information neural network model is established with spatial coordinates and time as inputs and the temperature field at various moments of the thermochemical ablation behavior of carbon-carbon composite materials as outputs. The physical information neural network model is then used to predict the temperature field at different moments of the ablation behavior of carbon-carbon composite materials.

[0035] In one feasible implementation, based on the obtained macroscopic thermochemical ablation reaction equation and thermochemical ablation heat conduction equation, a loss function is constructed using the mean square error method. A physical information neural network model is then established, with spatial coordinates and time as inputs and the temperature field at various moments in the thermochemical ablation behavior of the carbon-carbon composite material as the output. This physical information neural network model is used to predict the temperature field at different moments in the ablation behavior of the carbon-carbon composite material. Specifically:

[0036] Taking spatial coordinates and time as input, the data is passed through a fully connected layer. Each neuron in the fully connected layer is connected to all neurons in the previous layer, and each connection corresponds to a weight parameter. Each neuron corresponds to a bias parameter. Substituting the obtained weight parameters and bias parameters into the physical information neural network, the specific calculation formula is as follows:

[0037]

[0038] Where f is the activation function; w and b are the unknown parameters to be optimized in the physical information neural network model, which are the weights and biases of the physical information neural network, respectively, and n is the parameter of the physical information neural network.

[0039] Automatic differentiation techniques are used to calculate the derivative terms in the heat conduction equation and boundary conditions.

[0040] Based on the derivative term, a loss function for the physical information neural network is constructed using mean square error. The loss function includes the residuals of the heat conduction equation, the boundary conditions, and the initial conditions, resulting in a physical information neural network model whose output is the temperature field at various moments of the thermochemical ablation behavior of carbon-carbon composite materials.

[0041] In one feasible implementation, the fully connected layer includes four hidden layers, each with 40 neurons and an activation function of Tanh.

[0042] In one feasible implementation, the loss function consists of the residuals of the heat conduction equation, the residuals of the boundary conditions, and the residuals of the initial conditions, and its calculation formula is as follows:

[0043]

[0044]

[0045]

[0046] in, The density of carbon-carbon composite materials, denoted as , where T is the specific heat capacity of the carbon-carbon composite material, t is the temperature, t is the time, and k is the thermal conductivity. For network parameters, To calculate the number of data points within the domain, The number of training data points for the upper and lower boundary conditions. q represents the number of initial training data points, q represents the heat flux of the carbon-carbon composite material, and h represents the convective heat transfer coefficient. For ambient temperature, The initial temperature.

[0047] In one feasible implementation, the construction of the macroscopic geometric model of the carbon-carbon composite material includes:

[0048] A macroscopic geometric model of carbon-carbon composite materials is established and material properties are assigned to it; the material properties include density, specific heat capacity and thermal conductivity.

[0049] Next, sampling points are taken for the internal boundary and initial conditions of the model. Heat flow is applied to the upper surface of the model, and convective heat transfer boundary conditions are applied to the lower surface.

[0050] In one feasible implementation, the macroscopic geometric model of the carbon-carbon composite material is input into the physical information neural network model for iterative training to obtain the trained physical information neural network model; including:

[0051] Based on the constructed physical information neural network model, a loss function using the augmented Lagrange method is introduced.

[0052] Based on the loss function of the augmented Lagrange method, the constructed physical information neural network is trained using the gradient descent optimization algorithm to obtain the optimal loss function weights and neural network parameters, resulting in the final physical information neural network model whose output is the temperature field at various times of the thermochemical ablation behavior of carbon-carbon composite materials.

[0053] In one feasible implementation, the loss function incorporating the augmented Lagrangian method consists of the residuals of the heat conduction equation, the boundary conditions, and the initial conditions. The augmented Lagrangian method is used to apply the initial and boundary conditions to the residuals, and its calculation formula is as follows:

[0054]

[0055]

[0056]

[0057]

[0058] in, For the density of carbon-carbon composite materials, The specific heat capacity of the carbon-carbon composite material is given by T, where T is the temperature. For time, Let k be the spatial coordinates and k be the thermal conductivity. For network parameters, To calculate the number of data points within the domain, The number of training data points for the upper and lower boundary conditions. q represents the number of initial training data points, q represents the heat flux of the carbon-carbon composite material, and h represents the convective heat transfer coefficient. This represents the predicted temperature field. For ambient temperature, The initial temperature. This represents the residual of the heat conduction equation. and Let these represent the initial conditional loss and boundary loss after introducing the augmented Lagrange multiplier, respectively. For predefined constants, These are Lagrange multipliers (regularization coefficients).

[0059] In one feasible implementation, the step of inputting the macroscopic thermochemical ablation carbon-carbon composite material to be detected into the trained physical information neural network model to obtain the prediction result of the macroscopic thermochemical ablation behavior of the carbon-carbon composite material includes:

[0060] The macroscopic thermochemical ablation carbon-carbon composite material to be tested is input into the trained physical information neural network model to obtain the temperature changes at different times.

[0061] Based on the temperature, the surface retreat rate of carbon-carbon composite materials is calculated according to the formula for calculating the retreat rate of carbon-carbon composite materials at different temperatures, and the prediction results of the macroscopic thermochemical ablation behavior of carbon-carbon composite materials are obtained.

[0062] According to a second aspect disclosed in this application, this application provides a composite material thermochemical ablation calculation device based on a physical information neural network, comprising:

[0063] The macroscopic thermochemical ablation reaction equation construction module is used to construct macroscopic thermochemical ablation reaction equations.

[0064] A module for constructing thermochemical ablation heat conduction equations for carbon-carbon composite materials is used to construct thermochemical ablation heat conduction equations for carbon-carbon composite materials.

[0065] The physical information neural network model construction module is used to construct a physical information neural network model based on the macroscopic thermochemical ablation reaction equation and the thermochemical ablation heat conduction equation, using the mean square error method to construct a loss function.

[0066] The macroscopic geometric model building module for carbon-carbon composite materials is used to build macroscopic geometric models of carbon-carbon composite materials.

[0067] The model training module is used to input the macroscopic geometric model of the carbon-carbon composite material into the physical information neural network model for iterative training to obtain the trained physical information neural network model.

[0068] The macroscopic thermochemical ablation behavior prediction module for carbon-carbon composite materials is used to input the macroscopic thermochemical ablation carbon-carbon composite material to be detected into the trained physical information neural network model to obtain the prediction results of the macroscopic thermochemical ablation behavior of the carbon-carbon composite material.

[0069] According to a third aspect disclosed in this application, an electronic device is provided, including a processor and a memory communicatively connected to the processor;

[0070] The memory stores computer-executed instructions;

[0071] The processor executes computer execution instructions stored in the memory to implement the method described in any one of the first aspects.

[0072] According to a fourth aspect disclosed in this application, a computer-readable storage medium is provided, wherein computer-executable instructions are stored therein, which, when executed by a processor, are used to implement the method described in any one of the first aspects.

[0073] According to the fifth aspect disclosed in this application, a computer program product is provided, comprising a computer program that, when executed by a processor, is used to implement the method described in any one of the first aspects.

[0074] Compared with the prior art, this application has the following beneficial effects:

[0075] This application achieves a deep fusion of data-driven and physical constraints by embedding the physical laws of heat conduction into neural networks. It overcomes the reliance of traditional methods on ultra-fine meshes and large amounts of experimental data, effectively solving problems such as high computational costs, complex dynamic boundary handling, and insufficient prediction accuracy with small samples. Simultaneously, it ensures that the prediction results strictly adhere to physical laws such as mass conservation and energy conservation, providing an efficient and reliable computational tool for the rapid design and precise optimization of thermal protection systems for hypersonic vehicles and spacecraft, and promoting the engineering application and technological development of C / C composite materials in extreme thermal environments. Attached Figure Description

[0076] The accompanying drawings, which are incorporated in and form part of this specification, illustrate embodiments consistent with this application and, together with the description, serve to explain the principles of this application.

[0077] Figure 1 A schematic flowchart illustrating the composite material thermochemical ablation calculation method provided in this application embodiment;

[0078] Figure 2 This is a schematic diagram of the composite material surface element mass conservation provided in the embodiments of this application;

[0079] Figure 3 A schematic diagram of the constructed physical information neural network model provided in the embodiments of this application;

[0080] Figure 4 This is a schematic diagram of the structure of the composite material geometric model provided in the embodiments of this application;

[0081] Figure 5 A schematic diagram of the internal structure and boundary sampling points of the constructed physical information neural network model provided in this application embodiment;

[0082] Figure 6 A schematic diagram illustrating the predicted temperature field and surface degradation changes of the composite material at different times, provided for embodiments of this application;

[0083] Figure 7 This is a schematic diagram of the composite material thermochemical ablation calculation device provided in the embodiments of this application;

[0084] Figure 8 This is a schematic diagram of the structure of an electronic device provided in an embodiment of this application.

[0085] The accompanying drawings illustrate specific embodiments of this application, which will be described in more detail below. These drawings and descriptions are not intended to limit the scope of the concept in any way, but rather to illustrate the concept of this application to those skilled in the art through reference to particular embodiments. Detailed Implementation

[0086] Exemplary embodiments will now be described in detail, examples of which are illustrated in the accompanying drawings. When the following description relates to the drawings, unless otherwise indicated, the same numbers in different drawings denote the same or similar elements. The embodiments described in the following exemplary embodiments do not represent all embodiments consistent with this application. Rather, they are merely examples of apparatuses and methods consistent with some aspects of this application as detailed in the appended claims.

[0087] To improve the efficiency of calculation methods for macroscopic thermochemical ablation behavior of composite materials, this application proposes a calculation method for macroscopic thermochemical ablation behavior of composite materials based on physical information neural networks. This application achieves rapid and high-precision simulation of macroscopic thermochemical ablation process of carbon-carbon composite materials by integrating data-driven and physical constraints.

[0088] The technical solution of the composite material thermochemical ablation calculation method based on physical information neural network provided in this application will be described in detail below through specific embodiments. It should be noted that the following embodiments may exist alone or in combination with each other, and the same or similar content may not be described again in different embodiments.

[0089] Figure 1 A flowchart illustrating a composite material thermochemical ablation calculation method based on a physical information neural network, provided in this application embodiment, is shown below. Figure 1 In some embodiments, the process of this composite material thermochemical ablation calculation method based on physical information neural networks includes the following steps:

[0090] S101, construct the macroscopic thermochemical ablation reaction equation of carbon-carbon composite materials;

[0091] S102, Based on the macroscopic thermochemical ablation reaction equation, construct the thermochemical ablation heat conduction equation for carbon-carbon composite materials;

[0092] S103, Based on the macroscopic thermochemical ablation reaction equation and the thermochemical ablation heat conduction equation, a loss function is constructed using the mean square error method to build a physical information neural network model;

[0093] S104, constructing a macroscopic geometric model of carbon-carbon composite materials;

[0094] S105, Construct a macroscopic geometric model of carbon-carbon composite materials;

[0095] S106, input the macroscopic thermochemical ablation carbon-carbon composite material to be detected into the trained physical information neural network model to obtain the prediction results of the macroscopic thermochemical ablation behavior of the carbon-carbon composite material.

[0096] In step S101, the specific steps for constructing the macroscopic thermochemical ablation reaction equation of the carbon-carbon composite material are as follows:

[0097] Based on the thermochemical ablation reaction mechanism of carbon-carbon composite materials and the elemental mass conservation equation and thermochemical equilibrium equation, a macroscopic thermochemical ablation reaction equation for carbon-carbon composite materials is constructed; specifically:

[0098] C / C composites are composed of carbon fibers and a carbon matrix, with carbon as the main element, except for a few impurities. As the temperature increases, the thermochemical ablation reactions of C / C composites mainly include carbon oxidation, carbon-nitrogen reactions, and carbon sublimation. Based on these ablation mechanisms, the massless ablation rate of the material can be obtained by solving the elemental mass conservation equation and the thermochemical equilibrium equation. Figure 2 As shown, the mass loss flux of an element mainly includes diffusion mass flux and convection mass flux, and its expression is:

[0099]

[0100]

[0101] in, The mass loss rate of solid elements on the ablated surface. For the diffusion of elements, The elemental mass concentration in the solid mass loss. The mass flow rate of the gas overflowing from the surface. The mass concentration of the element. The airflow density in the boundary layer; The gas velocity in the boundary layer; This is a dimensionless mass exchange coefficient; The dimensionless mass ablation rate of the wall surface;

[0102] In the above formula It can also be expressed as the mass fraction of gas in the system, i.e.:

[0103]

[0104] in, The molecular weight of the element; The average molecular weight; It is the total pressure; The number of gas components considered in the surface system; The number of atoms of the element; For voltage division;

[0105] Combining the above equations, we can obtain the thermochemical ablation equation for each element in the system:

[0106]

[0107] In step S102, based on the macroscopic thermochemical ablation reaction equation, a thermochemical ablation heat conduction equation for carbon-carbon composite materials is constructed; the specific steps are as follows:

[0108] Based on the thermochemical ablation reaction process of carbon-carbon composite materials and the homogeneous isotropic nature of carbon-carbon composite materials, a thermochemical ablation heat conduction equation is constructed; specifically:

[0109] A large amount of heat is generated at the ablation boundary layer based on the thermochemical ablation equation. When this heat enters the interior of the structure, it is a heat and mass transfer process. At this time, the surface thermochemical ablation of the C / C composite material leads to mass loss and carries away surface heat, which is a multiphysics problem involving heat and mass transfer. When solving the thermochemical ablation heat transfer problem on a macroscopic scale, the C / C composite material is usually regarded as a homogeneous isotropic material, and its heat conduction equation is:

[0110]

[0111] in, The density of the C / C composite material; t is the specific heat capacity of the C / C composite material; T is the temperature; t is the time; k is the thermal conductivity.

[0112] Specifically, a second type of boundary condition is used on the ablated surface, namely:

[0113]

[0114] Where q is the heat flux of the C / C composite material; it is determined by the energy balance of the ablation wall.

[0115] A convective heat transfer boundary condition (third type boundary condition) is applied at the lower boundary, namely:

[0116]

[0117] Where k is the thermal conductivity and h is the convective heat transfer coefficient. The ambient temperature;

[0118] For the thermochemical ablation mass transfer problem of C / C composite materials, the mass ablation rate of the material surface is calculated based on the dimensionless mass ablation rate:

[0119]

[0120] in, To restore enthalpy, and for carbon-carbon composite materials ;

[0121] Based on the above formula for mass ablation rate, the surface retreat rate of the C / C composite material is:

[0122] .

[0123] In step S103, based on the macroscopic thermochemical ablation reaction equation and the thermochemical ablation heat conduction equation, a loss function is constructed using the mean square error method to build a physical information neural network model; the specific steps are as follows:

[0124] Based on the obtained macroscopic thermochemical ablation reaction equation and thermochemical ablation heat conduction equation, a loss function is constructed using the mean square error method. A physical information neural network model is established with spatial coordinates and time as inputs and temperature field at various moments of the thermochemical ablation behavior of carbon-carbon composite materials as output. The physical information neural network model is used to predict the temperature field at different moments of the ablation behavior of carbon-carbon composite materials.

[0125] Specifically, according to one embodiment of this application, the equations are solved based on the aforementioned thermochemical ablation theory, heat and mass transfer equations, and a deep neural network framework to establish a corresponding heat conduction physical information neural network model. First, spatial coordinates x and time t are input, and the data passes through a fully connected layer. Each neuron in the fully connected layer is connected to all neurons in the previous layer, with each connection corresponding to a weight parameter and each neuron corresponding to a bias parameter. Substituting the obtained weight parameters and bias parameters into the physical information neural network, the corresponding mathematical expression is:

[0126]

[0127] Where f is the activation function, which can be ReLU, Sigmoid, Tanh, etc.; w and b are the unknown parameters that the neural network model needs to optimize, which are the weights and biases of the neural network, respectively; n is the parameter of the physical information neural network; where the fully connected layer includes 4 hidden layers, each of which has 40 neurons and uses Tanh as the activation function.

[0128] Then, by using automatic differentiation techniques to solve the derivative terms in the heat conduction equation and boundary conditions, the temperature T is obtained as the output layer, resulting in a physical information neural network model that outputs the temperature field at various moments in the thermochemical ablation behavior of carbon-carbon composite materials. In this physical information neural network, constructing the loss function is a core step in integrating physical laws into deep learning. Therefore, this application utilizes mean squared error loss to construct the loss function, i.e., calculating the losses in the heat conduction equation, boundary conditions, and initial conditions. Finally, the gradient descent algorithm in deep neural networks is used to optimize the network parameters. The specific process is as follows: Figure 3 As shown; the loss function mainly consists of the residuals of the heat conduction equation, the upper and lower boundary residuals, and the initial condition residuals, and its mathematical expression can be written as:

[0129]

[0130]

[0131]

[0132]

[0133] in, The density of the C / C composite material, t is the specific heat capacity of the C / C composite material, T is the temperature, t is the time, and k is the thermal conductivity. For network parameters, The number of data points within the domain. The number of training data points for the upper and lower boundary conditions. q represents the number of initial training data points, q represents the heat flux of the C / C composite material, and h represents the convective heat transfer coefficient. For ambient temperature, The initial temperature;

[0134] In step S104, a macroscopic geometric model of the carbon-carbon composite material is constructed; the specific steps are as follows:

[0135] According to one embodiment of this application, a macroscopic geometric model of a C / C composite material is established, and material properties are assigned to the model, such as... Figure 4 As shown, for example, density, specific heat capacity, and thermal conductivity; then sampling points are taken for the model's interior, boundaries, and initial conditions, such as... Figure 5 As shown; finally, a heat flow is applied to the upper surface of the model, and a convective heat transfer boundary condition is applied to the lower surface.

[0136] In step S105, the macroscopic geometric model of the carbon-carbon composite material is input into the physical information neural network model for iterative training to obtain the trained physical information neural network model; the specific steps are as follows:

[0137] The macroscopic geometric model of the carbon-carbon composite material is input into the physical information neural network model for iterative training to obtain the trained physical information neural network model.

[0138] Based on the constructed physical information neural network model, a loss function using the augmented Lagrange method is introduced.

[0139] Based on the loss function of the augmented Lagrange method, the constructed physical information neural network is trained by the gradient descent optimization algorithm to obtain the optimal loss function weights and neural network parameters, and the final physical information neural network model is obtained, which outputs the temperature field of the thermochemical ablation behavior of carbon-carbon composite material at each time step.

[0140] According to one embodiment of this application, the specific process is as follows: When solving the heat conduction equation based on the physical information neural network and macroscopic geometric model, the initial conditions and boundary conditions are applied in the form of loss, i.e., "soft constraints". This type of constraint sometimes cannot effectively apply the initial conditions and boundary conditions, especially for transient heat conduction problems. If the initial conditions and boundary conditions cannot be effectively applied, the network training can easily fall into incorrect solutions and trivial solutions. Therefore, an augmented Lagrangian method is introduced into the residuals of the initial conditions and boundary conditions to effectively apply the initial conditions and boundary conditions. Then the loss function can be rewritten as:

[0141]

[0142]

[0143]

[0144]

[0145] in, The density of the C / C composite material, Here, T represents the specific heat capacity of the C / C composite material, and T represents the temperature. For time, Let k be the spatial coordinates and k be the thermal conductivity. For network parameters, To calculate the number of data points within the domain, The number of training data points for the upper and lower boundary conditions. q represents the number of initial training data points, q represents the heat flux of the C / C composite material, and h represents the convective heat transfer coefficient. For ambient temperature, The initial temperature. This represents the residual of the heat conduction equation. and Let these represent the initial conditional loss and boundary loss after introducing the augmented Lagrange multiplier, respectively. For a predefined constant, for A lower bound exists;

[0146] Based on the improved loss function described above, the constructed physical information neural network is trained using the gradient descent optimization algorithm, continuously updating the network parameters. , and Approaching the minimum value makes the loss function Minimize the value to ensure that the physical laws of the heat conduction equation are satisfied at all data points in the domain, thereby enabling the solution of the heat conduction equation.

[0147] In step S106, the macroscopic thermochemical ablation carbon-carbon composite material to be detected is input into the trained physical information neural network model to obtain the prediction results of the macroscopic thermochemical ablation behavior of the carbon-carbon composite material; the specific steps are as follows:

[0148] Based on the trained physical information neural network, the macroscopic thermochemical ablation carbon-carbon composite material to be detected is input into the trained physical information neural network model to obtain the temperature changes at different times; then, based on the temperature, the surface recession rate of the C / C composite material at different temperatures is obtained according to the formula for calculating the surface recession rate of the carbon-carbon composite material, such as... Figure 6 As shown, compared with existing COMSOL numerical simulation results, the physical information neural network improves both computational accuracy and efficiency without the need for mesh generation.

[0149] This application achieves a deep integration of data-driven and physical constraints by embedding the physical laws of heat conduction into neural networks. It breaks through the dependence of traditional methods on ultra-fine meshes and large amounts of experimental data, effectively solving problems such as high computational costs, complex dynamic boundary handling, and insufficient prediction accuracy for small samples. At the same time, by strictly following physical laws such as mass conservation and energy conservation, it provides efficient and reliable computational means for the rapid design and precise optimization of thermal protection systems for hypersonic vehicles and spacecraft, promoting the engineering application and technological development of C / C composite materials in extreme thermal environments.

[0150] Figure 7 This is a schematic diagram of the composite material thermochemical ablation calculation device based on physical information neural network provided in this application embodiment. (See attached diagram.) Figure 7 The thermochemical ablation calculation device includes various functional modules for implementing the aforementioned thermochemical ablation calculation method, and any functional module can be implemented by software and / or hardware.

[0151] In some embodiments, the composite material thermochemical ablation calculation device 1000 based on physical information neural network includes a macroscopic thermochemical ablation reaction equation construction module 1001, a carbon-carbon composite material thermochemical ablation heat conduction equation construction module 1002, a physical information neural network model construction module 1003, a carbon-carbon composite material macroscopic geometric model construction module 1004, a model training module 1005, and a carbon-carbon composite material macroscopic thermochemical ablation behavior prediction module 1006; wherein:

[0152] The macroscopic thermochemical ablation reaction equation construction module 1001 is used to construct macroscopic thermochemical ablation reaction equations.

[0153] The module 1002 for constructing the thermochemical ablation heat conduction equation for carbon-carbon composite materials is used to construct the thermochemical ablation heat conduction equation for carbon-carbon composite materials.

[0154] The physical information neural network model construction module 1003 is used to construct a physical information neural network model based on the macroscopic thermochemical ablation reaction equation and the thermochemical ablation heat conduction equation, using the mean square error method to construct a loss function.

[0155] The macroscopic geometric model construction module 1004 for carbon-carbon composite materials is used to construct macroscopic geometric models of carbon-carbon composite materials.

[0156] The model training module 1005 is used to input the macroscopic geometric model of the carbon-carbon composite material into the physical information neural network model for iterative training to obtain the trained physical information neural network model.

[0157] The macroscopic thermochemical ablation behavior prediction module 1006 for carbon-carbon composite materials is used to input the macroscopic thermochemical ablation carbon-carbon composite material to be detected into the trained physical information neural network model to obtain the prediction results of the macroscopic thermochemical ablation behavior of carbon-carbon composite materials.

[0158] The composite material thermochemical ablation calculation device 1000 provided in this application embodiment is used to execute the technical solution provided in the aforementioned composite material thermochemical ablation prediction method embodiment. Its implementation principle and technical effect are similar to those in the aforementioned method embodiment, and will not be repeated here.

[0159] It should be noted that the division of the various modules in the above device is merely a logical functional division. In actual implementation, they can be fully or partially integrated into a single physical entity, or they can be physically separated. Furthermore, these modules can be implemented entirely in software through processing element calls, entirely in hardware, or some modules can be implemented in software through processing element calls, while others are implemented in hardware.

[0160] Figure 8 This is a schematic diagram of the structure of an electronic device provided in an embodiment of this application. (See attached diagram.) Figure 8 The electronic device 1100 includes: a processor 1101, and a memory 1102 communicatively connected to the processor 1101;

[0161] Memory 1102 stores computer-executed instructions;

[0162] The processor 1101 executes the computer execution instructions stored in the memory 1102 to implement the technical solution of the aforementioned composite material thermochemical ablation prediction method.

[0163] In the aforementioned electronic device 1100, the memory 1102 and the processor 1101 are electrically connected directly or indirectly to achieve data transmission or interaction. For example, these components can be electrically connected to each other through one or more communication buses or signal lines, such as bus connections. The bus can be an Industry Standard Architecture (ISA) bus, a Peripheral Component Interconnect (PCI) bus, or an Extended Industry Standard Architecture (EISA) bus, etc. Buses can be classified as address buses, data buses, control buses, etc., but this does not mean that there is only one bus or one type of bus. The memory 1102 stores computer execution instructions for implementing the aforementioned method for predicting the macroscopic thermochemical ablation behavior of composite materials, including at least one software functional module that can be stored in the memory 1102 in the form of software or firmware. The processor 1101 executes various functional applications and data processing by running the software programs and modules stored in the memory 1102.

[0164] The memory 1102 includes at least one type of readable storage medium, not limited to Random Access Memory (RAM), Read Only Memory (ROM), Programmable Read-Only Memory (PROM), Erasable Programmable Read-Only Memory (EPROM), Electrically Erasable Programmable Read-Only Memory (EEPROM), etc. The memory 1102 stores programs, and the processor 1101 executes the programs after receiving execution instructions. Furthermore, the software programs and modules within the memory 1102 may also include an operating system, which may include various software components and / or drivers for managing system tasks (e.g., memory management, storage device control, power management, etc.) and can communicate with various hardware or software components to provide an operating environment for other software components.

[0165] Processor 1101 can be an integrated circuit chip with signal processing capabilities. The aforementioned processor 1101 can be a general-purpose processor, including a Central Processing Unit (CPU), a Network Processor (NP), a Digital Signal Processor (DSP), an Application Specific Integrated Circuit (ASIC), etc. It can implement or execute the methods, steps, and logic block diagrams disclosed in the embodiments of this application. The general-purpose processor can be a microprocessor, or processor 1101 can be any conventional processor.

[0166] The electronic device 1100 is used to execute the technical solution provided in the aforementioned embodiment of the composite material thermochemical ablation prediction method. Its implementation principle and technical effect are similar to those in the aforementioned method embodiment, and will not be repeated here.

[0167] This application also provides a computer-readable storage medium storing computer-executable instructions. When a processor executes the computer-executable instructions, it implements the technical solution of the aforementioned composite material thermochemical ablation prediction method.

[0168] The aforementioned computer-readable storage medium can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as static random access memory (SRAM), electrically erasable programmable read-only memory (EEPROM), erasable programmable read-only memory (EPROM), programmable read-only memory (PROM), read-only memory (ROM), magnetic storage, flash memory, magnetic disk, or optical disk. The computer-readable storage medium can be any available medium accessible to a general-purpose or special-purpose computer.

[0169] An exemplary readable storage medium is coupled to a processor, enabling the processor to read information from and write information to the readable storage medium. Alternatively, the readable storage medium can be an integral part of the processor. The processor and the readable storage medium can reside in an Application Specific Integrated Circuit (ASIC). Alternatively, the processor and the readable storage medium can exist as discrete components in the control unit of a computational device for macroscopic thermochemical ablation behavior of composite materials.

[0170] This application also provides a computer program product, including a computer program that, when executed by a processor, is used to implement the aforementioned technical solution for predicting thermochemical ablation of composite materials.

[0171] In the above embodiments, those skilled in the art will understand that the above method embodiments can be implemented entirely or partially by software, hardware, firmware, or any combination thereof. When implemented in software, it can be implemented entirely or partially in the form of a computer program product. A computer program product includes one or more computer instructions. When the computer program instructions are loaded and executed on a computer, all or part of the flow or function according to the embodiments of the present invention is generated. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The computer instructions can be stored in a computer-readable storage medium or transmitted from one computer-readable storage medium to another. For example, computer instructions can be transmitted from one website, computer, server, or data center to another website, computer, server, or data center via wired (e.g., coaxial cable, fiber optic, digital subscriber line (DSL)) or wireless (e.g., infrared, wireless network, microwave, etc.) means. The computer-readable storage medium can be any available medium that a computer can access or a data storage device such as a server or data center that integrates one or more available media. The available media can be magnetic media (e.g., floppy disks, hard disks, magnetic tapes), optical media (e.g., DVDs), or semiconductor media (e.g., solid-state drives (SSDs)).

[0172] It should be understood that this application is not limited to the precise structure described above and shown in the accompanying drawings, and various modifications and changes can be made without departing from its scope. The scope of this application is limited only by the appended claims.

Claims

1. A method for calculating thermal-chemical ablation of composite materials based on physical information neural networks, characterized by, The method comprises the following steps: constructing a macroscopic thermo-chemical ablation reaction equation of carbon-carbon composite material; based on the macroscopic thermo-chemical ablation reaction equation, constructing a thermo-chemical ablation heat conduction equation of carbon-carbon composite material; based on the macroscopic thermo-chemical ablation reaction equation and the thermo-chemical ablation heat conduction equation, constructing a loss function by using a mean square error method to construct a physical information neural network model; constructing a macroscopic geometric model of carbon-carbon composite material; inputting the macroscopic geometric model of carbon-carbon composite material into the physical information neural network model for iterative training to obtain a trained physical information neural network model; inputting a macroscopic thermo-chemical ablation carbon-carbon composite material to be detected into the trained physical information neural network model to obtain a prediction result of the thermo-chemical ablation behavior of carbon-carbon composite material; based on the macroscopic thermo-chemical ablation reaction equation and the thermo-chemical ablation heat conduction equation, constructing a loss function by using a mean square error method to construct a physical information neural network model; comprising: based on the obtained macroscopic thermo-chemical ablation reaction equation and the thermo-chemical ablation heat conduction equation, constructing a loss function by using a mean square error method to construct a physical information neural network model, which takes spatial coordinates and time as input and outputs temperature fields of the thermo-chemical ablation behavior of carbon-carbon composite material at different times, and uses the physical information neural network model to predict the temperature fields of the ablation behavior of carbon-carbon composite material at different times; based on the obtained macroscopic thermo-chemical ablation reaction equation and the thermo-chemical ablation heat conduction equation, constructing a loss function by using a mean square error method to construct a physical information neural network model, which takes spatial coordinates and time as input and outputs temperature fields of the thermo-chemical ablation behavior of carbon-carbon composite material at different times, and uses the physical information neural network model to predict the temperature fields of the ablation behavior of carbon-carbon composite material at different times; specifically: taking spatial coordinates and time as input, passing through a full connection layer, each neuron in the full connection layer is connected with all neurons of the previous layer, each connection corresponds to a weight parameter, and each neuron corresponds to a bias parameter, and the obtained weight parameter and bias parameter are substituted into the physical information neural network, and the specific calculation formula is: wherein f is an activation function; w and b are unknown parameters to be optimized in the physical information neural network model, which are weight and bias values of the physical information neural network, respectively; and n is a parameter of the physical information neural network; using automatic differentiation technology to calculate the derivative terms in the heat conduction equation and the boundary condition; according to the derivative terms, a loss function of the physical information neural network is constructed by using a mean square error; the loss function comprises a heat conduction equation residual error, a boundary condition residual error and an initial condition residual error, and a physical information neural network model outputting temperature fields of the thermo-chemical ablation behavior of carbon-carbon composite material at different times is obtained.

2. The method of claim 1, wherein, The method comprises the following steps: based on the macroscopic thermo-chemical ablation reaction equation, constructing a thermo-chemical ablation heat conduction equation of carbon-carbon composite material; comprising: wherein is the dimensionless mass ablation rate of the wall surface, is the mass concentration of the element, denotes the mass concentration of the element in the boundary layer, denotes the molecular weight of the element, denotes the average molecular weight, is the total pressure, denotes the number of gas components considered in the surface system, denotes the number of atoms of the element, denotes the partial pressure.

3. The method of claim 1, wherein, ​ Based on the thermal chemical ablation reaction process of carbon-carbon composite material and the homogeneous isotropy of carbon-carbon composite material, a thermal chemical ablation heat conduction equation is constructed; specifically: wherein, is the density of the carbon-carbon composite material, is the specific heat capacity of the carbon-carbon composite material, T is the temperature, t is the time, and k is the thermal conductivity; A second type of boundary condition is adopted on the surface of the carbon-carbon composite material, specifically: Wherein, q is the heat flux of the carbon-carbon composite material; A third type of boundary condition is adopted on the lower boundary of the carbon-carbon composite material, specifically: where k is the thermal conductivity, h is the convective heat transfer coefficient, Tambient is the ambient temperature; Based on the mass ablation rate of the thermal chemical ablation of the carbon-carbon composite material, the mass ablation rate of the surface of the carbon-carbon composite material is calculated according to the dimensionless mass ablation rate; specifically: wherein to restore enthalpy, and for carbon-carbon composites ; Based on the mass ablation rate, the surface recession rate of the carbon-carbon composite material is calculated; specifically: wherein is the surface recession rate of the carbon-carbon composite.

4. The method of claim 1, wherein, The full connection layer includes 4 hidden layers, the number of neurons in each hidden layer is 40, and the activation function is Tanh.

5. The method of claim 1, wherein, The loss function is composed of heat conduction equation residual, boundary condition residual and initial condition residual, and its calculation formula is: wherein, is the density of the carbon-carbon composite material, is the specific heat capacity of the carbon-carbon composite material, T is the temperature, t is the time, k is the thermal conductivity, is the network parameter, is the number of data points in the calculation domain, is the number of training data points for the upper and lower boundary conditions, is the number of training data points for the initial conditions, q is the heat flux of the carbon-carbon composite material, h is the convective heat transfer coefficient, is the ambient temperature, is the temperature at the initial time.

6. The method of claim 5, wherein, The carbon-carbon composite material macroscopic geometric model is constructed; including: The carbon-carbon composite material macroscopic geometric model is established, and material properties are given; the material properties include density, specific heat capacity and thermal conductivity; The internal boundary and initial condition of the model are sampled, and the heat flow is applied on the upper surface of the model, and the convective heat transfer boundary condition is applied on the lower surface.

7. The method of claim 6, wherein, The carbon-carbon composite material macroscopic geometric model is input into the physical information neural network model for iterative training to obtain the trained physical information neural network model; including: Based on the constructed physical information neural network model, a loss function introducing the augmented Lagrangian method is constructed; Based on the loss function introducing the augmented Lagrangian method, the constructed physical information neural network is trained by using the gradient descent optimization algorithm to obtain the optimal loss function weight and neural network parameter, and the final physical information neural network model with the output of the temperature field of the carbon-carbon composite material at each time of the thermal chemical ablation behavior is obtained.

8. The method of claim 7, wherein, The loss function introducing the augmented Lagrangian method is composed of heat conduction equation residual, boundary condition residual and initial condition residual, wherein the initial condition and boundary condition residual introduce the initial condition and boundary condition by using the augmented Lagrangian method, and the calculation formula is: where, is the density of carbon-carbon composites, is the specific heat capacity of carbon-carbon composites, T is the temperature, is the time, is the spatial coordinate, k is the thermal conductivity, is the network parameter, is the number of data points in the computational domain, is the number of training data points for the upper and lower boundary conditions, is the number of training data points for the initial conditions, q is the heat flux of carbon-carbon composites, h is the convective heat transfer coefficient, denotes the predicted temperature field, is the ambient temperature, is the temperature at the initial time, denotes the residual of the heat conduction equation, and denote the initial condition loss and the boundary loss after introducing the augmented Lagrangian, respectively, is a predefined constant, is the Lagrange multiplier.

9. The method of claim 1, wherein, The to-be-detected macroscopic thermal chemical ablation carbon-carbon composite material is input into the trained physical information neural network model to obtain the prediction result of the macroscopic thermal chemical ablation behavior of the carbon-carbon composite material; including: The to-be-detected macroscopic thermal chemical ablation carbon-carbon composite material is input into the trained physical information neural network model to obtain the temperature change at different times; Based on the temperature, the recession rate of the carbon-carbon composite material at different temperatures is obtained according to the recession rate calculation formula of the surface of the carbon-carbon composite material, and the prediction result of the macroscopic thermal chemical ablation behavior of the carbon-carbon composite material is obtained.

10. A device for calculating thermal-chemical ablation of composite materials based on physical information neural networks, which performs the method for calculating thermal-chemical ablation of composite materials based on physical information neural networks according to any one of claims 1 to 9, characterized in that, Including: The macroscopic thermal chemical ablation reaction equation construction module is used to construct the macroscopic thermal chemical ablation reaction equation; The carbon-carbon composite material thermal chemical ablation heat conduction equation construction module is used to construct the carbon-carbon composite material thermal chemical ablation heat conduction equation; The physical information neural network model construction module is configured to construct a loss function by using a mean square error method based on the macroscopic thermo-chemical ablation reaction equation and the thermo-chemical ablation heat conduction equation, and to construct a physical information neural network model. The carbon-carbon composite macroscopic geometric model construction module is configured to construct a carbon-carbon composite macroscopic geometric model. The model training module is configured to input the carbon-carbon composite macroscopic geometric model into the physical information neural network model for iterative training, and to obtain a trained physical information neural network model. The carbon-carbon composite macroscopic thermo-chemical ablation behavior prediction module is configured to input a macroscopic thermo-chemical ablation carbon-carbon composite material to be detected into the trained physical information neural network model, and to obtain a carbon-carbon composite macroscopic thermo-chemical ablation behavior prediction result.

11. An electronic device, comprising: It comprises: a processor and a memory connected in communication with the processor; the memory stores computer execution instructions; the processor executes the computer execution instructions stored in the memory to implement the method of any one of claims 1 to 9.

12. A computer-readable storage medium, characterized in that, The computer readable storage medium stores computer execution instructions, and the computer execution instructions are executed by the processor to implement the method of any one of claims 1 to 9.

Images