A temperature field prediction method based on multi-model cooperation

Abstract

The application relates to the technical field of numerical simulation of the intersection of computational fluid dynamics and computational heat transfer, and discloses a temperature field prediction method based on multi-model cooperation, which comprises the following steps: training a neural network using an aerodynamic heat data set to obtain an aerodynamic heat rapid prediction model; selecting multiple different flight trajectories within the flight envelope of an aircraft and discretizing the flight trajectories into flight operating points; using the aerodynamic heat rapid prediction model to obtain the cold-wall heat flux of all flight operating points of the flight trajectories under a preset wall surface temperature; using the cold-wall heat flux as a boundary condition to solve a transient heat conduction equation of a thermal protection structure, obtaining the temperature field change of the thermal protection structure along the flight trajectory under aerodynamic heating conditions, and forming a structure temperature field data set; training a neural network using the structure temperature field data set to obtain a structure heat transfer prediction model; and using the structure heat transfer prediction model to predict the temperature field. The application can realize the coupled rapid analysis of aerodynamic heat and wall surface structure heat transfer under any operating condition within the full flight envelope.

Description

Technical Field

[0001] This application relates to the field of numerical simulation technology at the intersection of computational fluid dynamics and computational heat transfer, and in particular to a temperature field prediction method based on multi-model collaboration. Background Technology

[0002] During high-speed flight, the surface temperature of an aircraft can rapidly rise to thousands of degrees Celsius due to intense air compression and friction. This high temperature not only causes overheating and failure of electronic equipment within the cabin but also induces material ablation or thermo-aeroelastic deformation, dynamically altering the aircraft's shape and affecting its aerodynamic performance. To ensure thermal safety, traditional design relies on numerical calculations to conduct coupled analysis of the aero-thermal environment and structural heat transfer; however, this method has significant limitations: 1. High computational cost and long cycle: Coupled analysis based on numerical calculation requires a lot of computational resources and time, which seriously affects the design and evaluation efficiency of the aircraft's thermal protection structure.

[0003] 2. High difficulty in mesh generation: The accuracy of numerical calculations, especially aerodynamic and thermal calculations, is highly dependent on mesh quality. Accurate coupled calculation analysis requires professional technicians to spend a lot of time drawing the relevant computational mesh, which is time-consuming and labor-intensive.

[0004] Existing numerical calculation-based rapid analysis methods for coupling aerodynamic and structural heat transfer are computationally expensive and time-consuming, making it difficult to scale up applications during the design phase and failing to meet the need for real-time assessment of aircraft structural thermal safety. Summary of the Invention

[0005] In view of this, this application provides a temperature field prediction method based on multi-model collaboration. It establishes a rapid prediction model for aerodynamic heat and a rapid prediction model for structural heat transfer based on neural networks. Through the collaborative prediction of multiple models, it realizes rapid coupled analysis of aerodynamic heat and structural heat transfer, which is applicable to aircraft thermal protection or other fluid-thermal coupling analysis scenarios.

[0006] This application discloses a temperature field prediction method based on multi-model collaboration, which includes: Step 1: Constructing an aerodynamic thermal dataset: Determine the flight envelope of the aircraft, select multiple combinations of operating conditions including flight altitude, angle of attack, and Mach number within the flight envelope, perform CFD numerical calculations on all operating condition combinations at a preset wall temperature, obtain aerodynamic thermal data of the aircraft under all operating condition combinations, and establish an aerodynamic thermal dataset; the flight envelope includes flight performance and flight range; the aerodynamic thermal dataset includes multiple aerodynamic thermal data sets; Step 2: Train the aerodynamic thermal prediction model: Use the aerodynamic thermal dataset to train a neural network to obtain a fast aerodynamic thermal prediction model with flight altitude, angle of attack, and Mach number as inputs and the cold wall heat flux on the surface of the aircraft at a preset wall temperature as output. Step 3: Construct a structural temperature field dataset: Select multiple different flight trajectories within the flight envelope of the aircraft and discretize them into flight operating points according to a preset time interval. Use the aerodynamic thermal rapid prediction model to obtain the cold wall heat flux of all flight operating points of all selected flight trajectories at a preset wall temperature. Use the cold wall heat flux as a boundary condition to solve the transient heat conduction equation of the thermal protection structure, and obtain the temperature field change of the thermal protection structure along the flight trajectory under aerodynamic heating conditions, thus forming a structural temperature field dataset. Step 4: Train the structural heat transfer prediction model: Based on the neural network and the structural temperature field dataset, construct a structural heat transfer prediction model with the current thermal protection structure temperature field and corresponding heat flow boundary conditions as input and the thermal protection structure temperature field at the next moment as output. Step 5: Multi-model coupling prediction: The temperature field is predicted using a structural heat transfer prediction model. The temperature field and thermal boundary conditions at the current moment are used as inputs to predict the temperature field at the next moment. Then, the temperature field at that moment is used as inputs along with the thermal boundary conditions predicted by the aerodynamic thermal rapid prediction model to iteratively predict the temperature field at subsequent moments until the iterative calculation ends.

[0007] Furthermore, the expression for the rapid aerodynamic thermal prediction model is as follows:

[0008] in, Indicates flight altitude. Represents the Mach number. Indicates angle of attack. This represents the cold wall heat flux on the surface of the aircraft at a preset wall temperature. This represents a rapid prediction model for aerodynamic heat.

[0009] Furthermore, the expression for the flight trajectory is:

[0010]

[0011] in, Indicates the flight path, This indicates that the flight trajectory is discretized according to a preset time interval. Flight conditions at any given moment express Flight altitude at any given moment express Mach number at time, express The angle of attack at any moment.

[0012] Further, the step of using the aerodynamic thermal rapid prediction model to obtain the cold wall heat flux at all flight operating points of all selected flight trajectories at a preset wall temperature includes: The cold wall heat flux at the preset wall temperature at each flight operating point is obtained using the following formula:

[0013] in, This represents the heat flux of the cold wall on the aircraft surface at a preset wall temperature. This indicates that the flight trajectory is discretized according to a preset time interval. Flight conditions at any given moment This represents a rapid prediction model for aerodynamic heat.

[0014] Furthermore, in step 3, in solving the transient heat conduction equation, the cold wall heat flow is corrected based on the total incoming flow temperature and the real-time wall temperature to obtain the hot wall heat flow. Combined with the wall radiation heat dissipation calculation, the net heat flow acting on the thermal protection structure is obtained, thus acquiring the net heat flow acting on the thermal protection structure at all times along the flight trajectory.

[0015] Furthermore, the step of correcting the cold wall heat flux based on the total incoming flow temperature and the real-time wall temperature to obtain the hot wall heat flux includes: The heat flux through the hot wall is obtained using the following formula:

[0016]

[0017] in, express The wall temperature at any given time, express The heat flow from the lower wall express The constant flow of time, express The constant flow of temperature express Mach number at time, Indicates specific heat ratio. This represents the cold wall heat flux of the aircraft surface at a preset wall temperature.

[0018] Furthermore, the calculation of the net heat flow acting on the thermal protection structure in conjunction with wall radiation heat dissipation includes: The net heat flow acting on the thermal protection structure is calculated using the following formula:

[0019] in, This represents the net heat flow acting on the thermal protection structure at the current moment. Indicates the surface emissivity of a material. This represents the Stefan-Boltzmann constant.

[0020] Furthermore, the expression for the transient heat conduction equation is:

[0021] in, For density, For heat capacity, The temperature field changes over time. , , For spatial coordinates, The sign for differentiation is given by K, where K is the thermal conductivity coefficient.

[0022] Furthermore, in step 4, the structural temperature field dataset is divided into a training set and a test set. The structural heat transfer prediction model is trained using the training set and evaluated simultaneously on the test set. During the training process, the error changes of the structural heat transfer prediction model on the training set and the test set are observed to ensure that the model does not overfit.

[0023] Further, step 5 includes: Predict the temperature field at the next moment using the following formula:

[0024] in, This represents a structural heat transfer prediction model. express Temperature field of the thermal protection structure at all times express The heat flow boundary condition at time t. The value range represents 2 to , Represents a positive integer.

[0025] Due to the adoption of the above technical solution, this application has the following advantages: 1. Reduced Dataset Construction Difficulty: This application first establishes a rapid aerodynamic heat prediction model, and then uses it as an alternative to CFD numerical calculation to conduct coupled analysis of aerodynamic heat and structural heat transfer to generate a structural temperature field dataset. The entire process avoids fully numerical coupled analysis, significantly reducing the computational cost of dataset construction.

[0026] 2. Improved efficiency of coupling analysis: This application only requires the establishment of a dataset through numerical simulation in the early stage of predictive model establishment. After the model training is completed, it can completely break away from numerical simulation and achieve coupling analysis along the flight trajectory in seconds by relying solely on the trained neural network model, which greatly improves efficiency. Attached Figure Description

[0027] To more clearly illustrate the technical solutions in the embodiments of this application, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments recorded in the embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings.

[0028] Figure 1 This is a schematic diagram illustrating the specific implementation process of the coupling analysis based on multi-model collaborative prediction in an embodiment of this application. Figure 2 This is a schematic diagram of the multi-model collaborative prediction coupling analysis process in an embodiment of this application; Figure 3 This is a schematic diagram of the aerothermal prediction model network structure according to an embodiment of this application; Figure 4 This is a schematic diagram of a structural heat transfer prediction framework based on Fourier neural operators according to an embodiment of this application; Figure 5 This is a schematic diagram of the temperature distribution on the outer surface of the spherical cone according to an embodiment of this application; Figure 6 This is a schematic diagram of the temperature distribution on the inner surface of the ball-cone according to an embodiment of this application. Detailed Implementation

[0029] The present application will be further described in conjunction with the accompanying drawings and embodiments. The described embodiments are only a part of the embodiments of the present application, and not all of the embodiments. All other embodiments obtained by those skilled in the art should fall within the protection scope of the embodiments of the present application.

[0030] See Figure 1 This application provides an embodiment of a temperature field prediction method based on multi-model collaboration, which includes: Step 1: Constructing an aerodynamic thermal dataset: Determine the flight envelope of the aircraft, select multiple combinations of operating conditions including flight altitude, angle of attack, and Mach number within the flight envelope, and perform CFD (Computational Fluid Dynamics) numerical calculations for all operating condition combinations at a preset wall temperature to obtain aerodynamic thermal data of the aircraft under all operating condition combinations, and establish an aerodynamic thermal dataset; the flight envelope includes flight performance and flight range; the aerodynamic thermal dataset includes multiple aerodynamic thermal data sets; Step 2: Train the aerodynamic thermal prediction model: Use the aerodynamic thermal dataset to train a neural network to obtain a fast aerodynamic thermal prediction model with flight altitude, angle of attack, and Mach number as inputs and the cold wall heat flux on the surface of the aircraft at a preset wall temperature as output. Step 3: Construct a structural temperature field dataset: Select multiple different flight trajectories within the flight envelope of the aircraft and discretize them into flight operating points according to a preset time interval. Use the aerodynamic thermal rapid prediction model to obtain the cold wall heat flux of all flight operating points of all selected flight trajectories at a preset wall temperature. Use the cold wall heat flux as a boundary condition to solve the transient heat conduction equation of the thermal protection structure, and obtain the temperature field change of the thermal protection structure along the flight trajectory under aerodynamic heating conditions, thus forming a structural temperature field dataset. Step 4: Train the structural heat transfer prediction model: Based on the neural network and the structural temperature field dataset, construct a structural heat transfer prediction model with the current thermal protection structure temperature field and corresponding heat flow boundary conditions as input and the thermal protection structure temperature field at the next moment as output. Step 5: Multi-model coupling prediction: The temperature field is predicted using a structural heat transfer prediction model. The temperature field and thermal boundary conditions at the current moment are used as inputs to predict the temperature field at the next moment. Then, the temperature field at that moment is used as inputs along with the thermal boundary conditions predicted by the aerodynamic thermal rapid prediction model to iteratively predict the temperature field at subsequent moments until the iterative calculation ends.

[0031] Optionally, the expression for the rapid aerodynamic thermal prediction model is:

[0032] in, Indicates flight altitude. Represents the Mach number. Indicates angle of attack. This represents the cold wall heat flux on the aircraft surface at a preset wall temperature (e.g., 300K wall temperature). This represents a rapid prediction model for aerodynamic heat.

[0033] Optionally, the expression for the flight trajectory is:

[0034]

[0035] in, Indicates the flight path, This indicates that the flight trajectory is discretized according to a preset time interval. Flight conditions at any given moment express Flight altitude at any given moment express Mach number at time, express The angle of attack at any moment.

[0036] Optionally, the step of using the aerodynamic thermal rapid prediction model to obtain the cold wall heat flux at a preset wall temperature for all flight operating points of all selected flight trajectories includes: The cold wall heat flux at the preset wall temperature at each flight operating point is obtained using the following formula:

[0037] in, This represents the heat flux of the cold wall on the aircraft surface at a preset wall temperature. This indicates that the flight trajectory is discretized according to a preset time interval. Flight conditions at any given moment This represents a rapid prediction model for aerodynamic heat.

[0038] Optionally, in step 3, in solving the transient heat conduction equation, the cold wall heat flow is corrected based on the total incoming flow temperature and the real-time wall temperature to obtain the hot wall heat flow. Combined with the wall radiation heat dissipation calculation, the net heat flow acting on the thermal protection structure is obtained to obtain the net heat flow acting on the thermal protection structure at all times along the flight trajectory.

[0039] Optionally, the step of correcting the cold wall heat flux based on the total incoming flow temperature and the real-time wall temperature to obtain the hot wall heat flux includes: The heat flux through the hot wall is obtained using the following formula:

[0040]

[0041] in, express The wall temperature at any given time, express The heat flow from the lower wall express The constant flow of time, express The constant flow of temperature express Mach number at time, This indicates the specific heat ratio (for example, a value of 1.4). This represents the cold wall heat flux of the aircraft surface at a preset wall temperature.

[0042] Optionally, the calculation of the net heat flow acting on the thermal protection structure in conjunction with wall radiation heat dissipation includes: The net heat flow acting on the thermal protection structure is calculated using the following formula:

[0043] in, This represents the net heat flow acting on the thermal protection structure at the current moment. Indicates the surface emissivity of a material. This represents the Stefan-Boltzmann constant. Optionally, W stands for watt, m for meter, and K for Kelvin.

[0044] Optionally, the expression for the transient heat conduction equation is:

[0045] in, For density, For heat capacity, The temperature field changes over time. , , For spatial coordinates, The sign for differentiation is given by K, where K is the thermal conductivity coefficient.

[0046] Optionally, in step 4, the structural temperature field dataset is divided into a training set and a test set. The structural heat transfer prediction model is trained using the training set and evaluated simultaneously on the test set. During the training process, the error changes of the structural heat transfer prediction model on the training set and the test set are observed to ensure that the model does not overfit.

[0047] Optionally, step 5 includes: Predict the temperature field at the next moment using the following formula:

[0048] in, This represents a structural heat transfer prediction model. express Temperature field of the thermal protection structure at all times express The heat flow boundary condition at time t. The value range represents 2 to , Represents a positive integer.

[0049] Alternatively, in selecting a neural network, different network frameworks such as fully connected neural networks, convolutional neural networks, or graph neural networks can be chosen according to the complexity of the actual problem.

[0050] This application generates a dataset based on traditional numerical calculations and trains a neural network to obtain a fast prediction model. Once the model is trained, it can realize rapid analysis of the coupling of aerodynamic heat and structural heat transfer under any working condition without the need for further numerical calculations.

[0051] The flight condition sampling method, aerodynamic heat prediction neural network model, and structural heat transfer prediction neural network model of this application can be selected according to the complexity of the actual problem. The implementation process of this application will be explained in detail below with a specific implementation case.

[0052] Case setting: For a cone with a length of 2m and a cone angle of... A three-dimensional spherical cone with a head radius of 0.1m was analyzed. The cone has a two-layer heat-resistant structure: an outer layer of 5mm thick fiberglass and an inner layer of 5mm thick aluminum alloy.

[0053] Implementation steps: Step 1: Construction of Aerothermal Dataset Without considering the effects of surface ablation, pyrolysis, and thermochemical non-equilibrium, the analysis is conducted only under non-ablative passive heat protection conditions; therefore, the flight envelope is set at a relatively low Mach number. Specifically, the flight altitude range is 25km~50km, the Mach number range is 5~10, and the angle of attack range is... Within the flight envelope, 252 sample points were selected using a full factorial design method, and the value levels of each variable are shown in Table 1. Due to the symmetry of the geometric shape, the angle of attack can be obtained simultaneously. Calculation results under flight conditions. CFD numerical calculations were performed under a constant wall temperature of 300K to establish an aerothermal dataset.

[0054] Table 1. Values ​​of various variables under flight conditions

[0055] Step 2: Training the aerodynamic thermal prediction model Regarding the aerodynamic thermal prediction model, considering that the aircraft's shape remains unchanged and only flight conditions affect surface heat flux, a fully connected neural network is used to establish the aerodynamic thermal prediction model. The inputs are flight altitude, Mach number, and angle of attack, and the output is the cold wall heat flux. The model structure is as follows: Figure 3 As shown, Figure 3 The circles in the diagram represent neurons. The aerothermal dataset established in step one is divided into training and testing sets in a 9:1 ratio for model training. During training, the errors in both the training and testing sets are monitored and adjusted to the appropriate training parameters (learning rate, etc.) to ensure training convergence while preventing overfitting.

[0056] Step 3: Construction of the structural temperature field dataset One hundred 100-second flight trajectories were discretized into different operating points at 1-second time intervals. The cold wall heat flux at all operating points was predicted using a trained aerodynamic thermal prediction model. Combining hot wall correction and structural heat transfer calculations, the structural temperature field of all flight trajectories was calculated, and a temperature field dataset was established.

[0057] Step 4: Training the structural heat transfer prediction model In structural heat transfer prediction, compared with aerodynamic heat transfer prediction, in addition to the increase in spatial dimension, structural heat transfer is also affected by the temporal dimension, which places higher demands on the model's learning ability. Since structural heat transfer is controlled by explicit PDEs (partial differential equations) and is relatively simpler than CFD, we consider using FNO (Fourier Neural Operator), which has strong learning capabilities for PDEs, for modeling and learning. The entire prediction period is discretized into multiple time points, each with different heat flux boundary inputs. Therefore, a recurrent neural network is used to learn the mapping between the temperature field at time t and the heat flux input to the temperature field at time t+dt. By predicting along time, the temperature field under unsteady thermal boundary conditions over a period of time is predicted. The specific framework is as follows: Figure 4 As shown, the model is trained using the dataset established in step three to obtain a fast predictive model for structural heat transfer. Figure 4 middle, Indicates temperature difference. This indicates a summation.

[0058] Step 5: Coupled prediction of aerodynamic heat transfer and structural heat transfer After the model training is completed, the flight trajectories that need to be analyzed are then... Figure 2 The process shown is analyzed. Figure 5 and Figure 6 The temperature distributions of the outer and inner surfaces of the spherical cone at different times are presented, showing that the predicted temperature distributions are basically consistent with the numerical calculations. The outer surface temperature initially shows an upward trend, then gradually decreases. This is because the aerodynamic heating effect is relatively low in the later stages, and the heat has already been transferred to the supporting layer (aluminum alloy layer), where a large amount of heat is absorbed, leading to a gradual decrease in surface temperature. The inner surface temperature consistently shows an upward trend, but is generally significantly lower than the outer surface temperature, indicating that the heat-insulating layer provides a certain degree of insulation. The predicted temperature distributions are basically consistent with the numerical calculations, demonstrating that the model established for the surface has high accuracy. In terms of efficiency, this application can achieve predictions on the order of seconds after the model is established. For example, the prediction time for one flight trajectory in this example is approximately 0.7 seconds, which is a significant improvement compared to numerical calculations that use a large amount of computing resources, giving it an irreplaceable advantage in real-time prediction and other aspects.

[0059] Figure 1This paper demonstrates the complete modeling steps of this application. First, sampling is performed within the flight envelope to obtain a large number of flight condition combinations. Then, CFD numerical simulation is used to calculate all the flight condition combinations to form an aerodynamic heat dataset. The aerodynamic heat dataset is used to train a neural network model to obtain a rapid aerodynamic heat prediction model. The rapid aerodynamic heat prediction model is combined with structural heat transfer numerical calculation to perform structural heat transfer calculations on a large number of flight trajectories to form a structural temperature field dataset. The structural temperature field dataset is used to train a neural network model to obtain a rapid structural heat transfer prediction model. Finally, the rapid aerodynamic heat prediction model and the rapid structural heat transfer prediction model are used together for coupled calculation to achieve rapid prediction.

[0060] Figure 2 This paper demonstrates how to conduct coupled analysis using a trained aerodynamic heating rapid prediction model and a structural heat transfer rapid prediction model. First, the given flight trajectory is discretized into multiple flight conditions at a certain time step. Then, the aerodynamic heating prediction model is used to predict the cold wall heat flux under all conditions. In heat transfer prediction, starting from the initial moment, the cold wall heat flux at that moment is corrected based on the wall temperature to obtain the actual hot wall heat flux entering the structure. Then, using the structural temperature and hot wall heat flux at that moment as input, the structural temperature at the next moment is predicted by the heat transfer prediction model. This heat transfer prediction process is repeated until all discrete flight conditions are calculated, obtaining the structural temperature field change history under aerodynamic heating conditions.

[0061] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of this application and not to limit them. Although this application has been described in detail with reference to the above embodiments, those skilled in the art should understand that modifications or equivalent substitutions can still be made to the specific implementation of this application. Any modifications or equivalent substitutions that do not depart from the spirit and scope of this application should be covered within the protection scope of the claims of this application.

Claims

1. A temperature field prediction method based on multi-model collaboration, characterized in that, include: Step 1: Constructing an aerodynamic thermal dataset: Determine the flight envelope of the aircraft, select multiple combinations of operating conditions including flight altitude, angle of attack, and Mach number within the flight envelope, perform CFD numerical calculations on all operating condition combinations at a preset wall temperature, obtain aerodynamic thermal data of the aircraft under all operating condition combinations, and establish an aerodynamic thermal dataset; the flight envelope includes flight performance and flight range; the aerodynamic thermal dataset includes multiple aerodynamic thermal data sets; Step 2: Train the aerodynamic thermal prediction model: Use the aerodynamic thermal dataset to train a neural network to obtain a fast aerodynamic thermal prediction model with flight altitude, angle of attack, and Mach number as inputs and the cold wall heat flux on the surface of the aircraft at a preset wall temperature as output. Step 3: Construct a structural temperature field dataset: Select multiple different flight trajectories within the flight envelope of the aircraft and discretize them into flight operating points according to a preset time interval. Use the aerodynamic thermal rapid prediction model to obtain the cold wall heat flux of all flight operating points of all selected flight trajectories at a preset wall temperature. Use the cold wall heat flux as a boundary condition to solve the transient heat conduction equation of the thermal protection structure, and obtain the temperature field change of the thermal protection structure along the flight trajectory under aerodynamic heating conditions, thus forming a structural temperature field dataset. Step 4: Train the structural heat transfer prediction model: Based on the neural network and the structural temperature field dataset, construct a structural heat transfer prediction model with the current thermal protection structure temperature field and corresponding heat flow boundary conditions as input and the thermal protection structure temperature field at the next moment as output. Step 5: Multi-model coupling prediction: The temperature field is predicted using a structural heat transfer prediction model. The temperature field and thermal boundary conditions at the current moment are used as input to predict the temperature field at the next moment. Then, the temperature field at that moment is used as input along with the thermal boundary conditions predicted by the aerodynamic heat rapid prediction model to iteratively predict the temperature field at subsequent moments until the iterative calculation ends. In step 3, in solving the transient heat conduction equation, the cold wall heat flow is corrected based on the total incoming flow temperature and the real-time wall temperature to obtain the hot wall heat flow. Combined with the wall radiation heat dissipation calculation, the net heat flow acting on the thermal protection structure is obtained, and the net heat flow acting on the thermal protection structure at all times on the flight trajectory is obtained. The process of correcting the cold wall heat flux based on the total incoming flow temperature and the real-time wall temperature to obtain the hot wall heat flux includes: The heat flux through the hot wall is obtained using the following formula: in, express The wall temperature at any given time, express The heat flow from the lower wall express The constant flow of time, express The constant flow of temperature express Mach number at time, Indicates specific heat ratio. This represents the cold wall heat flux of the aircraft surface at a preset wall temperature; The calculation of the net heat flow acting on the thermal protection structure, combined with wall radiation heat dissipation, includes: The net heat flow acting on the thermal protection structure is calculated using the following formula: in, This represents the net heat flow acting on the thermal protection structure at the current moment. Indicates the surface emissivity of a material. This represents the Stefan-Boltzmann constant; In step 4, the structural temperature field dataset is divided into a training set and a test set. The structural heat transfer prediction model is trained using the training set and evaluated simultaneously on the test set. During the training process, the error changes of the structural heat transfer prediction model on the training set and the test set are observed to ensure that the model does not overfit.

2. The method according to claim 1, characterized in that, The expression for the rapid aerodynamic thermal prediction model is: in, Indicates flight altitude. Represents the Mach number. Indicates angle of attack. This represents the cold wall heat flux on the surface of the aircraft at a preset wall temperature. This represents a rapid prediction model for aerodynamic heat.

3. The method according to claim 1, characterized in that, The expression for the flight trajectory is: in, Indicates the flight path, This indicates that the flight trajectory is discretized according to a preset time interval. Flight conditions at any given moment express Flight altitude at any given moment express Mach number at time, express The angle of attack at any moment.

4. The method according to claim 1, characterized in that, The step of using the aerodynamic thermal rapid prediction model to obtain the cold wall heat flux at all flight operating points of all selected flight trajectories at a preset wall temperature includes: The cold wall heat flux at the preset wall temperature at each flight operating point is obtained using the following formula: in, This represents the heat flux of the cold wall on the aircraft surface at a preset wall temperature. This indicates that the flight trajectory is discretized according to a preset time interval. Flight conditions at any given moment This represents a rapid prediction model for aerodynamic heat.

5. The method according to claim 1, characterized in that, The expression for the transient heat conduction equation is: in, For density, For heat capacity, The temperature field changes over time. , , For spatial coordinates, The sign for differentiation is given by K, which is the thermal conductivity coefficient.

6. The method according to claim 1, characterized in that, Step 5 includes: Predict the temperature field at the next moment using the following formula: in, This represents a structural heat transfer prediction model. express Temperature field of the thermal protection structure at all times express The heat flow boundary condition at time t. The value range represents 2 to , Represents a positive integer.

Images