Satellite load temperature prediction and thermal parameter correction method based on PINN

Abstract

The invention discloses a PINN-based satellite load temperature prediction and thermal parameter correction method, which relates to the technical field of space thermal control, and has the technical key points that the prediction of a satellite load temperature field and the correction of thermal model parameters are realized by effectively combining known temperature information data and thermophysical knowledge by utilizing a thermophysical information neural network; according to the complete satellite load temperature prediction and thermal parameter correction method based on the PINN, the neural network and thermal physical prior information are combined, thermal model parameters which meet the physical law and are high in prediction precision are obtained, and a PINN model and a thermal simulation model which are used for accurately predicting the on-orbit / test temperature are obtained; and the thermal parameter correction efficiency and the temperature prediction accuracy are improved.

Description

Technical Field

[0001] The present invention relates to the technical field of space thermal control, and particularly relates to a method for satellite payload temperature prediction and thermal parameter correction based on PINN. Background Art

[0002] Due to the inaccuracy of the simulation thermal model, the satellite payload thermal model cannot accurately predict the temperature states during experiments and in orbit. Therefore, it is necessary to correct the thermal parameters of the thermal model using experimental temperature data or in-orbit temperature data, and finally obtain accurate thermal parameters and a thermal simulation model. Traditional correction methods are manual trial and error by thermal control engineers, or correction through other optimization methods after sampling by the Monte Carlo method.

[0003] In recent years, to improve the correction efficiency, some researchers have introduced neural network surrogate models into the correction field to replace the simulation model. However, these methods will cause two problems. On the one hand, with the development of future spacecraft missions, the structural layout of satellite payloads and their thermal control subsystems is becoming increasingly complex, which makes there are many uncertain factors involved in thermal model correction, and these thermal parameters may be correlated and coupled with each other, while the temperature measurement points are often insufficient, often resulting in underdetermined solution problems. On the other hand, whether it is the manual trial and error method or the Monte Carlo method, it is necessary to repeatedly substitute groups of thermal parameters back into the simulation software for thousands of groups of simulation calculations. The time cost of this step is quite large, and the obtained dataset is also sparse, which cannot meet the requirements of an increasingly short development cycle and an increasingly high prediction accuracy.

[0004] With the development of computers, some scholars have proposed using machine learning to solve the correction problem. Replacing time-consuming finite element calculations with neural networks is beneficial to greatly improving the correction efficiency. However, as a "black box", the thermal parameters obtained by solving the neural network often lose their physical meaning, making it difficult to deeply explain and elaborate on the thermal mechanism in the problem of thermal model correction, and there is a certain degree of decoupling from the actual thermal physical process.

[0005] Therefore, the present invention aims to provide a method for satellite payload temperature prediction and thermal parameter correction based on PINN, adding thermal boundary conditions and physical constraints to a general surrogate model to overcome the above disadvantages. Summary of the Invention

[0006] The purpose of the present invention is to solve the above problems, and provide a method for satellite payload temperature prediction and thermal parameter correction based on PINN, which is used to solve the problems of the prediction accuracy of the thermal simulation model under sparse temperature measurement data, the low efficiency of thermal parameter correction, and the inconsistency between the corrected thermal parameters and the actual thermal physical phenomenon, and the obtained thermal parameters are not in the true state in the prior art, and improve the efficiency of thermal model correction and the accuracy of the simulation model.

[0007] To achieve the above object, the technical solution of the present invention is as follows:

[0008] The present invention provides a method for satellite payload temperature prediction and thermal parameter correction based on PINN, and the method includes the following steps:

[0009] S1. Establish an original thermal simulation model of the satellite payload, sample the thermal parameters therein, and perform simulation calculations to obtain a training data set;

[0010] S2. Establish a thermophysical mechanism equation of the satellite payload, construct a PINN model, and endow the model with thermophysical constraints based on the thermal equation;

[0011] S3. Use the training data set in S1 as the first data-driven information, use the thermophysical constraints in S2 as the first physical-driven information, train the PINN model, calculate the first loss function, and finally obtain an accurate PINN temperature prediction model;

[0012] S4. Use the heat transfer formula as the second physical-driven information and the on-orbit telemetry test temperature as the second data-driven information to obtain the second loss function of the PINN model;

[0013] S5. Based on the PINN model in S3, adjust the thermal parameters to minimize the second loss function. At this time, the thermal parameters are the corrected thermal parameters, and substitute them into the original thermal simulation model to obtain a corrected accurate thermal simulation model.

[0014] In step S1, the establishment of the original thermal simulation model of the satellite payload, sampling the thermal parameters therein, and performing simulation calculations to obtain a training data set includes the following steps:

[0015] S101. Establish an original thermal simulation model in the simulation modeling software according to the physical model of the satellite payload;

[0016] S102. Use finite element analysis software to simplify the geometric body and divide the mesh;

[0017] S103. Perform a solution analysis of the temperature field to obtain the temperature field distribution and temperature level of the key parts of the satellite payload;

[0018] S104. Perform parameter identification on the thermal parameters of the thermal model;

[0019] S105. Determine the value range of the thermal parameters in step S104, perform Latin hypercube uniform sampling, and substitute them into the original thermal simulation model in step S101 for simulation calculations to obtain a data matrix of temperature and coordinates.

[0020] In step S2, establishing the thermophysical mechanism equation of the satellite payload, constructing the PINN model, and imposing thermophysical constraints on the model based on the thermal equation include the following steps:

[0021] S201. Establish the general thermal control equations for each component.

[0022] where ρ is the density; φ is the general solution function, which is represented by the temperature matrix T in this problem; Γ φ is the generalized diffusion coefficient; λ is the thermal conductivity; S is the generalized source term, which is represented by the radiative heat exchange between the component and the space environment in this problem;

[0023] The first term of the thermal control equation is the unsteady term, the second term is the convection term, the third term is the diffusion term, and the fourth term is the source term. Since the satellite payload is in the space environment and the heat transfer modes are thermal radiation and heat conduction, the unsteady term and the convection term can be ignored, and the first and second terms are 0.

[0024] The control equation should be further expanded as follows:

[0025]

[0026] where a is the component thickness, ε is the infrared hemispherical emissivity of the component surface, σ is the Boltzmann constant, T u is the matrix of the surface absolute temperature distribution of the component related to the position, T 0 is the heat sink temperature, x is the abscissa in the Cartesian coordinate system, and y is the ordinate;

[0027] S202. The PINN model should be built on the basis of the general neural network model framework. The input units include the thermal parameters in the satellite payload thermal model. The thermal parameters include the infrared hemispherical emissivity ε, solar absorptivity α, thermal resistance r, heat transfer coefficient k, thermal conductivity λ, heat capacity C, heat dissipation h, etc. of the surface, the space coordinate vector (x, y), and the simulation temperature matrix T required for data driving (x,y) , abbreviated as T.

[0028] The output unit of the PINN model should include the predicted temperature at the position of (x, y) hereinafter referred to as The PINN model should be able to calculate the first-order partial differential and second-order partial differential equations between temperature and coordinates. Therefore, the output unit should also include

[0029] The PINN model should have a data-driven first loss function Loss data , and a physics-driven first loss function Loss PDE ;

[0030] The PINN model should have hidden layers and their weight vectors w ij , bias vector b ij and activation functions;

[0031] S203. When the heat sink is 4K cold air, in the control equation of S201, T 0 4 is negligible relative to T u 4 Therefore, the thermophysical constraint should be

[0032]

[0033] In step S3, using the training dataset of S1 as the first data-driven information and the thermophysical constraint of S2 as the first physics-driven information to train the PINN model and calculate the first loss function, and finally obtaining an accurate PINN temperature prediction model, which includes the following steps:

[0034] S301. Using the data matrix of the simulated calculation temperature and coordinates obtained in S1 as the training dataset, with the coordinate data as the input and the temperature calculation value as the output, to provide data-driven information for the training of the PINN model. The data-driven first loss function should be:

[0035]

[0036] S302. Assign the thermophysical constraint obtained in S2 to the PINN in the form of the first loss function, and calculate the physical constraint first loss function by the output unit, which should be

[0037]

[0038] S303. Perform weighted summation on the data-driven first loss function and the physical constraint first loss function to obtain the total loss function

[0039] Loss f =αLoss data +βLoss PDE ;

[0040] where α is the weight coefficient of the data-driven first loss function and β is the weight coefficient of the physical constraint first loss function;

[0041] S304. Minimize Loss f i.e., min(Loss f)Taking [the specific objective], the PINN model established in S202 is trained based on the temperature data obtained in S105. By means of optimizing the network structure, solver settings, hyperparameter optimization, and even feature extraction, data augmentation, etc. using optimization algorithms, the prediction results of the PINN model can accurately replace the calculation results of the simulation model. At this time, the PINN model is the accurate PINN temperature prediction model.

[0042] In step S4, the heat transfer formula is used as the second physical driving information, and the on-orbit remote test temperature is used as the second data driving information to obtain the second loss function of the PINN model. Different from the first driving information, the first driving information is the field quantity of the temperature field, expressed by a temperature matrix, while the second driving information is the predicted temperature value, expressed by a scalar. It specifically includes the following steps:

[0043] S401. Since there are thermal interfaces between components, the temperature fields of different components are coupled with each other and should satisfy the heat transfer formula:

[0044] Wall steady-state heat conduction formula

[0045]

[0046] where k is the heat transfer coefficient between components, A is the contact area, d is the thickness, T i is the temperature of component i, and T j is the temperature of component j;

[0047] The second loss function of wall steady-state heat conduction is

[0048]

[0049] where is the PINN predicted temperature of component i, is the PINN predicted temperature of component j;

[0050] Thermal balance formula for a single component

[0051]

[0052] where j is the component adjacent to component i, D ji is the conduction network coefficient between components, R ji is the radiation network coefficient between components, Q i is the total heat source of the component;

[0053] The second loss function at steady state of the thermal balance formula

[0054]

[0055] Thermal conduction formula for a single component

[0056]

[0057] Among them, T a is the temperature of temperature measurement point a, and T b is the temperature of temperature measurement point b;

[0058] The physical-driven second loss function should be

[0059] Loss py = mLoss 1 + nLoss 2 + sLoss 3 ;

[0060] Among them, m, n, and s are the proportions of the second loss functions of the three physical constraints respectively;

[0061] S402. The PINN model obtained in S3 no longer adjusts information such as the network structure and hyperparameters, but only adjusts the input parameters, that is, the thermal parameters, and uses the on-orbit telemetry or test-measured temperature as the new data-driven information. The data-driven first loss function should be

[0062]

[0063] Among them, is the predicted temperature value of the PINN model after adjusting the thermal parameters, is the on-orbit telemetry or test-measured temperature value;

[0064] S403. Weighted sum the new data-driven second loss function and the new physical constraint second loss function to obtain the new total second loss function as

[0065] Loss g = aLoss ms + bLoss py ;

[0066] Among them, a is the weight coefficient of the new data-driven second loss function, and b is the weight coefficient of the new physical constraint second loss function.

[0067] In step S5, based on the PINN model obtained in S3, adjust the thermal parameters to minimize the second loss function of S4. At this time, the thermal parameters are the corrected thermal parameters, and substitute them into the simulation model to obtain the corrected accurate thermal simulation model, including the following steps:

[0068] S501. To minimize Loss f That is, min(Loss f)Taking this as the goal, the accurate PINN model obtained in S3 is trained based on the temperature dataset matrix of the on-orbit telemetry temperature values measured in the experiment. Without adjusting the network structure, solver settings, and hyperparameters, only the input parameters are adjusted, that is, thermal parameters such as infrared hemispherical emissivity ε, solar absorptivity α, thermal resistance r, heat transfer coefficient k, thermal conductivity λ, heat capacity C, heat dissipation h, etc. So that the prediction results of the PINN model can accurately match the on-orbit telemetry / experimentally measured temperature results. At this time, the thermal parameters are the accurate corrected thermal parameters.

[0069] S502. Substitute the corrected thermal parameters obtained in S501 into the simulation software in turn to obtain a thermal simulation prediction model that can accurately predict the on-orbit / experimental temperature, and obtain a PINN model that can accurately predict the on-orbit / experimental temperature.

[0070] Compared with the prior art, the beneficial effects of this solution are as follows:

[0071] 1. In the present invention, by using the PINN model to replace the simulation model for temperature field prediction and parameter optimization, the sparse simulation data and measured temperature data are reasonably utilized, greatly improving the efficiency of designing and correcting the thermal model, making the cost lower, being able to be processed and optimized efficiently, significantly reducing the demand for computing resources, and better meeting the requirements of the increasingly short development cycle of the satellite payload thermal control subsystem.

[0072] 2. In the present invention, by applying the thermal physical information constraint conditions, the defect of insufficient interpretability of the neural network as a "black box" is improved by using physical information. Combining data information, physical mechanism information, and the required expert prior knowledge, the obtained thermal parameters have certain physical meanings, conform to physical laws, and establish a thermal connection with the actual physical process of thermotics. Description of the Drawings

[0073] Figure 1 is the flowchart of the method for satellite payload temperature prediction and thermal parameter correction based on PINN in the embodiment of the present invention;

[0074] Figure 2 is the schematic diagram of the low-temperature optical link in the embodiment of the present invention;

[0075] Figure 3 is the plan view of the heat dissipation surface of the low-temperature optical link in the embodiment of the present invention;

[0076] Figure 4 is the PINN network structure diagram in the embodiment of the present invention.

[0077] In the figure: 1. Heat dissipation surface; 2. Cold end of the heat pipe; 3. Heat pipe; 4. Hot end of the heat pipe; 5. Mirror; 6. Cold box; 7. Base plate; 8. Mirror frame. Detailed Embodiment

[0078] To enable those skilled in the art to better understand the solution of the present invention, the technical solution of the present invention will be further described in detail below in conjunction with the embodiments and drawings of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative work shall fall within the protection scope of the present invention.

[0079] It should be noted that, without conflict, the embodiments in the present invention and the features in the embodiments can be combined with each other. The present invention will be described in detail below in conjunction with the embodiments.

[0080] Embodiment:

[0081] (1) Establish an original thermal simulation model of the satellite payload, sample the thermal parameters therein and perform simulation calculations to obtain a training dataset;

[0082] In the simulation modeling software, perform thermal simulation modeling according to the physical model of the cryogenic optical link, simplify the geometric body and divide the grid using the finite element analysis software, perform the solution analysis of the temperature field, and obtain the temperature field distribution of the heat dissipation surface 1 and the temperature level of the cold end 2 of the heat pipe.

[0083] Perform parameter identification on the thermal parameters of all thermal models, including the infrared hemispherical emissivity, solar absorptivity of the heat dissipation surface 1, the heat transfer coefficient between the cold end 2 of the heat pipe and the heat dissipation surface 1, the heat conduction thermal resistance of the heat pipe, etc. Determine the value range of the thermal parameters, perform Latin hypercube uniform sampling, and substitute them into the thermal simulation model for simulation calculations to obtain the data matrix of the temperature and coordinates of the heat dissipation surface 1.

[0084] (2) Establish the thermophysical mechanism equation of the satellite payload, construct the PINN model and endow the model with thermophysical constraints based on the thermal equation; establish the thermal control equation of the heat dissipation surface 1,

[0085]

[0086] where a is the thickness of the heat dissipation surface 1, ε is the infrared hemispherical emissivity of the surface of the heat dissipation surface 1, σ is the Boltzmann constant, T u is the matrix related to the surface absolute temperature distribution and position of the heat dissipation surface 1, x is the abscissa in the Cartesian coordinate system, and y is the ordinate.

[0087] As Figure 4 shown, build the PINN model on the basis of the general neural network model framework. The input units are thermal parameters such as the infrared hemispherical emissivity ε, solar absorptivity α, heat transfer coefficient k between the cold end 2 of the heat pipe and the heat dissipation surface 1, heat conduction thermal resistance r of the heat pipe, and the spatial coordinate vector (x, y) as Figure 3, the simulated temperature matrix T required for data driving (x,y) , abbreviated as T. The output unit includes the predicted temperature of the heat dissipation surface 1 at the (x,y) position hereinafter referred to as Calculate the first-order partial differential and second-order partial differential equations between temperature and coordinates. Therefore, the output unit also includes

[0088] Set the data-driven first loss function Loss data , the physics-driven first loss function Loss PDE , the hidden layer and its weight vector w ij , the bias vector b ij and the activation function.

[0089] (3) Use the training data set in (1) as the data-driven information and the thermophysical constraints in (2) as the physics-driven information to train the PINN model, calculate the first loss function, and finally obtain an accurate PINN temperature prediction model;

[0090] Use the sparse simulation calculation temperature results and the coordinate (x,y) matrix as the training data set, with the coordinate data as the input and the temperature calculation value as the output, to provide data-driven information for the training of the PINN model. The data-driven first loss function should be:

[0091]

[0092] Endow the PINN with the thermophysical constraints obtained in S2 in the form of the first loss function. From the Calculate the physics constraint first loss function as

[0093]

[0094] Perform weighted summation on the data-driven first loss function and the physics constraint first loss function to obtain the total breeding loss function

[0095] Loss f =αLoss data +βLoss PDE ;

[0096] Among them, α is the weight coefficient of the data-driven first loss function, and β is the weight coefficient of the physics constraint first loss function.

[0097] To minimize Loss f That is, min(Loss f) Taking [the target], based on the temperature dataset matrix obtained in S105, train the PINN model established in S202. By means of optimizing the network structure, solver settings, hyperparameter optimization, and even feature extraction, data augmentation, etc. using the optimization algorithm, make the prediction results of the PINN model accurately replace the calculation results of the simulation model. At this time, the PINN model is the accurate PINN temperature prediction model.

[0098] (4) Taking the heat transfer formula as the new physical driving information and the on-orbit telemetry / test temperature as the new data driving information, obtain the second loss function of the PINN model;

[0099] The second loss function of steady-state heat conduction is

[0100]

[0101] where k is the heat transfer coefficient between the heat dissipation surface 1 and the cold end 2 of the heat pipe, A is the contact area, d is the thickness, is the PINN predicted temperature of the heat dissipation surface 1, is the PINN predicted temperature of the cold end 2 of the heat pipe.

[0102] It should also satisfy the heat balance formula of the heat dissipation surface 1

[0103] The second loss function of the heat balance formula at steady state is

[0104]

[0105] The heat conduction formula of the heat pipe 3 is

[0106]

[0107] where T a is the temperature of the cold end 2 of the heat pipe, T b is the temperature of the hot end 4 of the heat pipe, and r is the thermal resistance of the heat pipe itself. It is considered that the temperature of the hot end of the heat pipe can represent the temperature of the cold box 6 and its optical components 5, 7.

[0108] The second loss function of the physical drive should be

[0109] Loss py = mLoss 1 + nLoss 2 + sLoss 3 ;

[0110] where m, n, and s are the weights of the second loss functions of the three physical constraints respectively.

[0111] Taking the on-orbit telemetry or test measurement temperature as the new data driving information, the second loss function of the data drive is

[0112]

[0113] Among them, is the predicted temperature value of the PINN model after adjusting the thermal parameters, T′ (x,y) is the temperature value measured by on-orbit telemetry or experiment.

[0114] The data-driven second loss function and the physical constraint second loss function are weighted and summed to obtain a new total second loss function as

[0115] Loss g = aLoss ms + bLoss py

[0116] Among them, a is the weight coefficient of the new data-driven second loss function, and b is the weight coefficient of the new physical constraint second loss function.

[0117] (5) Based on the PINN model obtained in (3), adjust the thermal parameters so that the second loss function in (4) is minimized. The thermal parameters at this time are the corrected thermal parameters, and substitute them into the simulation model to obtain the corrected accurate thermal simulation model.

[0118] To minimize Loss f That is, min(Loss f ), based on the temperature dataset matrix of the measured temperature values by experiment / on-orbit telemetry, train the accurate PINN model obtained in (3), and adjust the input parameters, that is, the thermal parameters, including the infrared hemispherical emissivity ε, solar absorptivity α of the heat dissipation surface 1, the heat transfer coefficient k between the cold end 2 of the heat pipe and the heat dissipation surface 1, the thermal resistance r of the heat pipe heat conduction, etc., so that the prediction result of the PINN model can accurately match the temperature result measured by on-orbit telemetry / experiment. The thermal parameters at this time are the accurate corrected thermal parameters.

[0119] Substitute the corrected thermal parameters obtained in S501 into the simulation software in turn to obtain a thermal simulation prediction model that can accurately predict the on-orbit / experimental temperature, and obtain a PINN model that can accurately predict the on-orbit / experimental temperature. The above specific embodiments are only explanations of the present invention, and they are not limitations of the present invention. Those skilled in the art can make modifications without creative contributions to this embodiment according to needs after reading this specification, but as long as they are within the scope of the claims of the present invention, they are protected by the patent law.

Claims

1. A method for satellite payload temperature prediction and thermal parameter correction based on PINN, characterized in that: The method comprises the following steps: S1. Establish an original thermal simulation model of the satellite payload, sample the thermal parameters therein and perform simulation calculations to obtain a training data set; S2. Establish the thermophysical mechanism equation of satellite payload, construct the PINN model and give the model thermophysical constraints based on the thermal equation; S3, using the training data set of S1 as the first data driving information and the thermal physical constraint of S2 as the first physical driving information, training the PINN model, calculating the first loss function, and finally obtaining an accurate PINN temperature prediction model; S4, using the heat transfer formula as the second physical driving information and the on-orbit telemetry test temperature as the second data driving information to obtain the second loss function of the PINN model; S5. Based on the PINN model of S3, adjust the thermal parameters to minimize the second loss function. The thermal parameters at this time are the corrected thermal parameters, which are substituted into the original thermal simulation model to obtain the corrected accurate thermal simulation model.

2. The method for satellite payload temperature prediction and thermal parameter correction based on PINN as claimed in claim 1, characterized in that: In step S1, the original thermal simulation model of the satellite payload is established, the thermal parameters thereof are sampled and simulated to obtain a training data set, including the following steps: S101, establishing an original thermal simulation model in simulation modeling software according to a physical model of a satellite payload; S102. Use finite element analysis software to simplify the geometry and divide the mesh; S103, performing temperature field solution analysis to obtain temperature field distribution and temperature level of key parts of satellite payload; S104, identifying the thermal parameters of the thermal model, including but not limited to the infrared hemispherical emissivity, solar absorptivity, thermal resistance, heat transfer coefficient, thermal conductivity, heat capacity, heat consumption and other thermal parameters of the surface; S105, determine the value interval of the thermal parameter in step S104, perform Latin hypercube uniform sampling, and substitute it into the original thermal simulation model in step S101 for simulation calculation to obtain a data matrix of temperature and coordinates.

3. The method for satellite payload temperature prediction and thermal parameter correction based on PINN as claimed in claim 1, characterized in that: In step S2, the establishment of the thermophysical mechanism equation of the satellite payload, the construction of the PINN model and the assignment of thermophysical constraints to the model based on the thermal equation include the following steps: S201. Establish the general thermal control equations for each component Where ρ is the density; φ is the general solution function, which is represented by the temperature matrix T in this problem; Γ φ is the generalized diffusion coefficient; λ is the thermal conductivity; S is the generalized source term, in which the source term is represented by the radiation heat exchange between the component and the space environment; The first term of the heat control equation is a non-steady-state term, the second term is a convection term, the third term is a diffusion term, and the fourth term is a source term. Since the satellite payload is located in a space environment, the heat transfer mode is thermal radiation and heat conduction, the non-steady-state term and the convection term can be ignored, and the first term and the second term are 0. The control equation should be further expanded as follows: Where a is the thickness of the component, ε is the infrared hemispherical emissivity of the component surface, σ is the Boltzmann constant, and T u is the matrix of the absolute surface temperature distribution of the component and its position, T0 is the heat sink temperature, x is the abscissa in the Cartesian coordinate system, and y is the ordinate; S202. The PINN model should be built on the basis of a general neural network model framework. The input unit includes thermal parameters in the satellite payload thermal model, including the surface infrared hemispherical emissivity ε, solar absorptivity α, thermal resistance r, heat transfer coefficient k, thermal conductivity λ, heat capacity C, heat consumption h and other thermal parameters, spatial coordinate vector (x, y), and simulation temperature matrix T required for data driving. (x,y) , referred to as T; The PINN model output cell should include the predicted temperature at the (x,y) location Hereinafter referred to as The PINN model should be able to calculate the first-order partial differential and second-order partial differential equations between temperature and coordinates. Therefore, the output unit should also include The PINN model should have a data-driven first loss function Loss data , Physically driven first loss function Loss PDE ; The PINN model should have a hidden layer and its weight vector w ij , bias vector b ij and activation functions; S203, when the heat sink is 4K cold air, T0 in the control equation of S201 4 Relative to T u 4 can be neglected, so the thermophysical constraint should be 4. The method for satellite payload temperature prediction and thermal parameter correction based on PINN as claimed in claim 1, characterized in that: In step S3, the training data set of S1 is used as the first data driving information, the thermal physical constraint of S2 is used as the first physical driving information, the PINN model is trained, the first loss function is calculated, and finally an accurate PINN temperature prediction model is obtained, including the following steps: S301, the data matrix of simulated temperature and coordinates obtained in S1 is used as a training data set, the coordinate data is input, and the temperature calculation value is output, to provide data-driven information for the training of the PINN model. The data-driven first loss function should be: S302, the thermal physical constraints obtained in S2 are assigned to PINN in the form of the first loss function, and the output unit The first loss function for calculating physical constraints should be S303: Perform a weighted summation of the data-driven first loss function and the physical constraint first loss function to obtain a total first loss function. Loss f =αLoss data +βLoss PDE ; Among them, α is the weight coefficient of the data-driven first loss function, and β is the weight coefficient of the physical constraint first loss function; S304, to minimize Loss f That is min(Loss f ) is taken as the goal, and the PINN model built in S202 is trained based on the temperature data obtained in S105. The optimization algorithm is used to optimize the network structure, solver settings, hyperparameter optimization, and even feature extraction and data enhancement, so that the prediction results of the PINN model can accurately replace the calculation results of the simulation model. At this time, the PINN model is an accurate PINN temperature prediction model.

5. The method for satellite payload temperature prediction and thermal parameter correction based on PINN as claimed in claim 1, characterized in that: In step S4, the heat transfer formula is used as the second physical driving information, and the on-orbit telemetry test temperature is used as the second data driving information to obtain a new second loss function of the PINN model, including the following steps: S401. Since there are thermal interfaces between components, the temperature fields of different components are coupled with each other and should satisfy the heat transfer formula: Steady-state heat conduction formula for the wall Where k is the heat transfer coefficient between components, A is the contact area, d is the thickness, and T i is the temperature of component i, T j is the temperature of component j; The second loss function of steady-state heat conduction on the wall is in, Predicted temperature for PINN of component i, Predict the temperature of PINN for component j; Heat balance formula for a single component Where j is the component adjacent to component i, D ji is the inter-component conduction network coefficient, R ji is the radiation network coefficient between components, Q i is the total heat source of the component; The second loss function of the thermal balance formula in steady state Heat transfer formula for a single component Among them, T a is the temperature of the temperature measuring point a, T b is the temperature of temperature measuring point b; The second physics-driven loss function should be Loss py =mLoss1+nLoss2+sLoss3; Among them, m, n, and s are the proportions of the second loss functions of the three physical constraints respectively; The PINN model obtained in S402 and S3 no longer adjusts the network structure, hyperparameters and other information, but only adjusts the input parameters, namely the thermal parameters, and uses the on-orbit telemetry or experimental measurement temperature as the new data-driven information. The data-driven first loss function should be: in, The temperature value predicted by the PINN model after adjusting the thermal parameters, T′ (x,y) Measuring temperature values ​​for on-orbit telemetry or testing; S403: Perform a weighted summation of the data-driven second loss function and the physical constraint second loss function to obtain a total second loss function: Loss g =aLoss ms +bLoss py ; Among them, a is the weight coefficient of the new data-driven second loss function, and b is the weight coefficient of the new physical constraint second loss function.

6. The method for satellite payload temperature prediction and thermal parameter correction based on PINN as claimed in claim 1, characterized in that: In step S5, the PINN model obtained in step S3 is used as the basis to adjust the thermal parameters so that the second loss function of step S4 is minimized. The thermal parameters at this time are the corrected thermal parameters, which are substituted into the simulation model to obtain the corrected accurate thermal simulation model, including the following steps: S501, to minimize Loss f That is min(Loss f ) is the goal, and the accurate PINN model obtained in S3 is trained based on the temperature data set matrix of the on-orbit telemetry temperature value measured experimentally. The network structure, solver settings and hyperparameters are no longer adjusted. Only the input parameters, namely thermal parameters such as infrared hemispherical emissivity ε, solar absorptivity α, thermal resistance r, heat transfer coefficient k, thermal conductivity λ, heat capacity C, heat consumption h, etc., are adjusted so that the prediction results of the PINN model can accurately match the temperature results of the on-orbit telemetry / experimental measurement. At this time, the thermal parameters are the accurate corrected thermal parameters; S502, the corrected thermal parameters obtained in S501 are sequentially re-substituted into the simulation software to obtain a thermal simulation prediction model that can accurately predict the on-orbit / test temperature, and to obtain a PINN model that can accurately predict the on-orbit / test temperature.

Images