High-altitude rocket rarefied flow field reconstruction method and device based on flight test data
By combining the Boltzman-BGK equation with the Shakhov equilibrium distribution function and a deep neural network, a rarefied flow field reconstruction model for high-altitude rockets was constructed, solving the problem of flow field reconstruction in high Mach cross-basin flow problems and achieving efficient flow field data prediction under unknown boundary conditions.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-12-26
- Publication Date
- 2026-04-03
AI Technical Summary
Existing technologies cannot effectively reconstruct the rarefied flow field of high-Mach cross-basin high-altitude rockets, especially when the boundary conditions are unknown, conventional CFD methods are difficult to use known data for accurate solutions.
A method for reconstructing the rarefied flow field of a high-altitude rocket based on flight test data is adopted. This method combines the Boltzman-BGK equation with the Shakhov equilibrium distribution function and a deep neural network. By constructing parallel first and second sub-networks and training the model using various physical constraints and loss functions, the flow field data can be predicted.
By utilizing a small amount of local flow field data obtained from sensors on the surface of high-altitude rockets without the need for precise boundary conditions, more accurate and efficient flow field reconstruction can be achieved, improving the accuracy and efficiency of flow field reconstruction.
Smart Images
Figure CN121787325A_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of physical quantity prediction, specifically to a method and apparatus for reconstructing a rarefied flow field of a high-altitude rocket based on flight test data. Background Technology
[0002] High Mach cross-basin flow field reconstruction is a crucial tool for scientific research, engineering development, and accident analysis of near-space equipment such as high-altitude rockets and spacecraft. However, to date, there is no effective method to directly reconstruct the high Mach cross-basin flow field of near-space equipment. Conventional CFD methods require complete initial flow field and boundary condition information, but in practical engineering problems, some boundary conditions of the computational domain may be unknown. Furthermore, conventional CFD methods struggle to incorporate known data from the computational domain, meaning this known information is almost useless in a single CFD solution. In contrast, the Physical Information Neural Network (PINN) model can use the coordinates and time of the computational domain as input, the dependent variables of the fluid control equations as output, and the output variables to form the fluid control equations. These output variables are then used as the loss function of a deep learning algorithm to train the neural network, thus realizing a deep learning-based method for solving partial differential equations. This method effectively solves the aforementioned problems of conventional CFD methods.
[0003] Extensive research has been conducted by scholars both domestically and internationally on this topic. In 2020, Raissi et al. published a work combining flow experiment visualization data with the Navier-Stokes equations based on PINN, achieving reconstruction of the experimental velocity and pressure fields through concentration field distribution. Wang et al. applied PINN to improve the quality of three-dimensional tomographic PIV experimental measurements of hemispherical bottom wakes; their results, compared with DNS results, demonstrate that PINN has great potential to transform spatially sparse and noisy data into complete three-dimensional flow fields. Lou et al. introduced the Boltzmann-BGK equations into the physical constraints of PINN, constructing a trans-basin PINN model applicable to flow from continuous to rarefied flow. However, this model employs an isothermal BGK model and is therefore not suitable for high Mach trans-basin flow problems. It can be seen that most current research on flow field reconstruction using PINN models is not applicable to high Mach trans-basin flow problems. Summary of the Invention
[0004] The purpose of this application is to propose a method and apparatus for reconstructing a rarefied flow field of a high-altitude rocket based on flight test data, addressing the aforementioned technical problems.
[0005] In a first aspect, the present invention provides a method for reconstructing a rarefied flow field of a high-altitude rocket based on flight test data, comprising the following steps:
[0006] A high-altitude rocket rarefied flow field reconstruction model and a non-equilibrium term prediction model based on the Boltzmann-BGK equation using the Shakhov equilibrium distribution function were constructed and trained to obtain the trained high-altitude rocket rarefied flow field reconstruction model and the trained non-equilibrium term prediction model. The non-equilibrium term prediction model includes a first sub-network and a second sub-network set in parallel. The first sub-network and the second sub-network output the non-equilibrium terms of the density distribution function and the temperature distribution function, respectively, based on the input flow field coordinates. The loss functions used in the training process include a first loss function based on the residual terms of the Boltzmann-BGK equation, a second loss function based on the residual terms of the physical constraints represented by the non-equilibrium terms in the Boltzmann-BGK equation, and a third loss function based on the flight test data and the predicted values of the flow field data output by the high-altitude rocket rarefied flow field reconstruction model based on the input flow field coordinates. The flight test data includes CFD simulation results and sensor data.
[0007] The coordinates of the flow field to be predicted on the high-altitude rocket are input into the trained high-altitude rocket rarefied flow field reconstruction model to obtain the predicted values of the flow field data.
[0008] Preferably, the flow field data includes density, velocity, pressure, and temperature, and the Boltzman-BGK equation based on the Shakhov equilibrium distribution function is expressed as follows:
[0009] ;
[0010] in, In the flow field coordinates and time At that location, the speed is The particles in The velocity distribution function in the 3D velocity space, Indicates the relaxation time. Indicates unbalanced terms. Represented by Maxwell's distribution function Adding the heat flux correction term to the Shakhov equilibrium distribution function, its expression is:
[0011] ;
[0012] ;
[0013] Where Pr represents the Prandtl number, Velocity in fluid data The characteristic velocity in the vicinity, Indicates heat flux, Represents the gas constant. This represents the temperature in the fluid data. This represents the pressure in fluid data. This represents a vector in a space of dimension L=3-D.
[0014] As a preferred option, the construction process of the first loss function is as follows:
[0015] The non-equilibrium term of the density distribution function predicted by the first and second subnetworks. and the non-equilibrium term of the temperature distribution function Predicted values constituting the unbalanced term The predicted value of the Shakhov equilibrium distribution function was obtained by calculating the predicted value of the flow field data from the high-altitude rocket rarefied flow field reconstruction model. The predicted value of the velocity distribution function is calculated using the following formula. :
[0016] ;
[0017] Will Taking the partial derivatives with respect to time and flow field coordinates respectively, we obtain and , and then combine Substitute these terms into the Boltzmann-BGK equation and calculate the residual terms. As shown in the following formula:
[0018] ;
[0019] in, This represents the i-th flow field coordinate randomly selected for calculating the residual term of the Boltzmann-BGK equations;
[0020] The mean square error of the residual term in the Boltzmann-BGK equation is calculated using the following formula to obtain the first loss function:
[0021] ;
[0022] in, Denotes the first loss function. This represents the total number of flow field coordinates randomly selected for calculating the residual terms of the Boltzmann-BGK equations;
[0023] The process of constructing the second loss function is as follows:
[0024] Construct the following physical constraints:
[0025] ;
[0026] ;
[0027] ;
[0028] in, Represents density in fluid data. express Regarding time Find the partial derivative. Represents velocity in fluid data. This represents the pressure in fluid data. Indicates energy. express Taking the partial derivative with respect to the flow field coordinates, express, Represents viscous stress. Indicates the thermal conductivity term;
[0029] Calculate the second-order central moment of the non-equilibrium term as a prediction of the viscous stress. Calculate the non-equilibrium term of the density distribution function. The first-order moment at the origin is used as the predicted value of the heat conduction term. The residual terms of physical constraints are calculated by combining the predicted values of the flow field data from the high-altitude rocket rarefied flow field reconstruction model. As shown in the following formula:
[0030] ;
[0031] in, This represents the predicted density value in the fluid data. This represents the predicted velocity value in the fluid data. This represents the predicted pressure value in the fluid data. This represents the j-th flow field coordinate randomly selected for calculating the residual term of the physical constraints;
[0032] The mean square error of the residual term of the physical constraints is calculated using the following formula to obtain the second loss function:
[0033] ;
[0034] in, This represents the second loss function. This represents the total number of flow field coordinates randomly selected for calculating the residual terms of physical constraints.
[0035] As a preferred option, the construction process of the third loss function is as follows:
[0036] The mean square error between the flight test data and the predicted values of the flow field data output by the high-altitude rocket rarefied flow field reconstruction model is calculated to obtain the third loss function, as shown in the following equation:
[0037] ;
[0038] in, This represents the third loss function. This represents the total number of samples in the training data. This represents the k-th flow field coordinate in the training data; This represents the predicted value of the flow field data output after the k-th flow field coordinate in the training data is input into the high-altitude rocket rarefied flow field reconstruction model. This represents the flight test data corresponding to the k-th flow field coordinate in the training data;
[0039] The loss function used in the training of the high-altitude rocket rarefied flow field reconstruction model As shown in the following formula:
[0040] ;
[0041] in, , and These represent the weights corresponding to the first loss function, the second loss function, and the third loss function, respectively.
[0042] Preferably, the sensor data is the actual flow field data collected after the predicted values of the flow field data output by the high-altitude rocket rarefied flow field reconstruction model are identified by gradient sensitivity analysis to identify regions of drastic flow field changes and regions of gentle flow field changes, and the sensor distribution is adjusted in the regions of drastic flow field changes and regions of gentle flow field changes respectively.
[0043] As a preferred option, the high-altitude rocket rarefied flow field reconstruction model includes a fully connected network; both the first and second sub-networks use several deconvolution layers connected in sequence.
[0044] Secondly, the present invention provides a high-altitude rocket rarefied flow field reconstruction device based on flight test data, comprising:
[0045] The model building module is configured to construct and train a high-altitude rocket rarefied flow field reconstruction model and a non-equilibrium term prediction model based on the Boltzmann-BGK equation using the Shakhov equilibrium distribution function. The trained high-altitude rocket rarefied flow field reconstruction model and the trained non-equilibrium term prediction model are obtained. The non-equilibrium term prediction model includes a first sub-network and a second sub-network set in parallel. The first and second sub-networks output non-equilibrium terms of the density distribution function and temperature distribution function, respectively, based on the input flow field coordinates. The loss functions used during training include a first loss function constructed based on the residual terms of the Boltzmann-BGK equation, a second loss function constructed based on the residual terms of the physical constraints represented by the non-equilibrium terms in the Boltzmann-BGK equation, and a third loss function constructed based on the flight test data and the predicted values of the flow field data output by the high-altitude rocket rarefied flow field reconstruction model based on the input flow field coordinates. The flight test data includes CFD simulation results and sensor data.
[0046] The prediction module is configured to input the coordinates of the flow field to be predicted on the high-altitude rocket into a trained high-altitude rocket rarefied flow field reconstruction model to obtain the predicted values of the flow field data.
[0047] Thirdly, the present invention provides an electronic device including one or more processors; and a storage device for storing one or more programs, wherein when the one or more programs are executed by the one or more processors, the one or more processors implement the method as described in any implementation of the first aspect.
[0048] Fourthly, the present invention provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the method as described in any of the implementations of the first aspect.
[0049] Fifthly, the present invention provides a computer program product, including a computer program that, when executed by a processor, implements the method as described in any of the implementations in the first aspect.
[0050] Compared with the prior art, the present invention has the following beneficial effects:
[0051] (1) The high-altitude rocket rarefied flow field reconstruction method based on flight test data proposed in this invention adopts the combination of the dual distribution function Boltzmann-BGK equation based on Shakhov equilibrium distribution function and deep neural network, and introduces a variety of physical constraints, such as thermally compressible flow control equation, entropy increase, flow conservation, etc., to ensure the physical rationality of the high-altitude rocket rarefied flow field reconstruction model. It can achieve more accurate and efficient flow field reconstruction by using a small amount of local real flow field data obtained by high-altitude rocket surface sensors without the need for precise boundary conditions.
[0052] (2) The high-altitude rocket rarefaction flow field reconstruction method based on flight test data proposed in this invention uses the high-altitude rocket rarefaction flow field reconstruction model to output the predicted value of the flow field data, and combines the non-equilibrium term output by the non-equilibrium term prediction model to construct the first loss function and the second loss function. For limited high-altitude rocket flight test data, the flow field reconstruction accuracy can be improved under limited data by optimizing the sensor layout. Attached Figure Description
[0053] To more clearly illustrate the technical solutions in the embodiments of the present invention, 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 of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.
[0054] Figure 1 This is a flowchart illustrating the high-altitude rocket rarefied flow field reconstruction method based on flight test data, as an embodiment of this application.
[0055] Figure 2 This is a schematic diagram of a high-altitude rocket rarefied flow field reconstruction device based on flight test data, as an embodiment of this application.
[0056] Figure 3 This is a schematic diagram of the hardware structure of an electronic device provided in an embodiment of the present invention. Detailed Implementation
[0057] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of this invention, and not all of them. Based on the embodiments of this invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this invention.
[0058] Figure 1 An embodiment of this application illustrates a method for reconstructing a rarefied flow field of a high-altitude rocket based on flight test data, comprising the following steps:
[0059] S1. Construct and train a high-altitude rocket rarefied flow field reconstruction model and a non-equilibrium term prediction model based on the Boltzmann-BGK equation using the Shakhov equilibrium distribution function. The trained high-altitude rocket rarefied flow field reconstruction model and the trained non-equilibrium term prediction model are obtained. The non-equilibrium term prediction model includes a first sub-network and a second sub-network set in parallel. The first sub-network and the second sub-network output the non-equilibrium terms of the density distribution function and the temperature distribution function, respectively, based on the input flow field coordinates. The loss functions used in the training process include a first loss function based on the residual terms of the Boltzmann-BGK equation, a second loss function based on the residual terms of the physical constraints represented by the non-equilibrium terms in the Boltzmann-BGK equation, and a third loss function based on the flight test data and the predicted values of the flow field data output by the high-altitude rocket rarefied flow field reconstruction model based on the input flow field coordinates. The flight test data includes CFD simulation results and sensor data.
[0060] In a specific embodiment, the high-altitude rocket rarefied flow field reconstruction model includes a fully connected network; the first sub-network and the second sub-network both use several deconvolution layers connected in sequence.
[0061] Specifically, the high-altitude rocket rarefied flow field reconstruction model and the non-equilibrium term prediction model in the embodiments of this application employ deep neural networks. In one example, the high-altitude rocket rarefied flow field reconstruction model uses a fully connected network. Its input is flow field coordinates, and its output is the predicted value of the flow field data, which includes physical parameters such as density, velocity, pressure, and temperature. This fully connected network uses 6 hidden layers, each containing 64 neurons. During training, the learning rate is set to lr=0.0001, and the Adam optimizer is used. In the non-equilibrium term prediction model, the first and second sub-networks both use three sequentially connected deconvolutional layers with a kernel size of 3×3. Their inputs are both flow field coordinates, and their corresponding outputs are the non-equilibrium terms of the density distribution function, respectively. and the non-equilibrium term of the temperature distribution function The non-equilibrium term of the density distribution function and the non-equilibrium term of the temperature distribution function This constitutes a non-equilibrium term. Although the first and second subnetworks use the same network structure, they use different outputs during training, resulting in different parameters and causing them to learn different constraints.
[0062] Specifically, in the embodiments of this application, flow field data from high-altitude rocket flight tests is first collected during the training process, and necessary preprocessing, such as normalization and denoising, is performed to ensure that the data is suitable for network training. The preprocessed data, as a dataset, will be used for the training and validation of the high-altitude rocket rarefied flow field reconstruction model and the non-equilibrium term prediction model; wherein, the non-equilibrium term output by the non-equilibrium term prediction model participates in the training process of the high-altitude rocket rarefied flow field reconstruction model, and the training of the non-equilibrium term prediction model is also carried out simultaneously. The samples in the dataset include flow field coordinates and their corresponding true values of flow field data.
[0063] In a specific embodiment, the flow field data includes density, velocity, pressure, and temperature, and is expressed by the Boltzman-BGK equation based on the Shakhov equilibrium distribution function as follows:
[0064] ;
[0065] in, In the flow field coordinates and time At that location, the speed is The particles in The velocity distribution function in the 3D velocity space, Indicates the relaxation time. Indicates unbalanced terms. Represented by Maxwell's distribution function Adding the heat flux correction term to the Shakhov equilibrium distribution function, its expression is:
[0066] ;
[0067] ;
[0068] Where Pr represents the Prandtl number, Velocity in fluid data The characteristic velocity in the vicinity, Indicates heat flux, Represents the gas constant. This represents the temperature in the fluid data. This represents the pressure in fluid data. This represents a vector in a space of dimension L=3-D.
[0069] In a specific embodiment, the construction process of the first loss function is as follows:
[0070] The non-equilibrium term of the density distribution function predicted by the first and second subnetworks. and the non-equilibrium term of the temperature distribution function Constitutes unbalanced terms The predicted value of the Shakhov equilibrium distribution function was obtained by calculating the predicted value of the flow field data from the high-altitude rocket rarefied flow field reconstruction model. The predicted value of the velocity distribution function is calculated using the following formula. :
[0071] ;
[0072] Will Taking the partial derivatives with respect to time and flow field coordinates respectively, we obtain and , and then combine Substitute these terms into the Boltzmann-BGK equation and calculate the residual terms. As shown in the following formula:
[0073] ;
[0074] in, This represents the i-th flow field coordinate randomly selected for calculating the residual term of the Boltzmann-BGK equations;
[0075] The mean square error of the residual term in the Boltzmann-BGK equation is calculated using the following formula to obtain the first loss function:
[0076] ;
[0077] in, Denotes the first loss function. This represents the total number of flow field coordinates randomly selected for calculating the residual terms of the Boltzmann-BGK equations.
[0078] Specifically, embodiments of this application construct a first loss function based on the dual-distribution Boltzmann-BGK equation of the Shakhov equilibrium distribution function, and introduce the loss function used in the training process of the neural network to realize the physical constraints of the cross-basin thermally compressible flow control equations for variable Prandtl number and specific heat ratio. Therefore, the first loss function can be constructed based on the Boltzmann-BGK equation of the Shakhov equilibrium distribution function.
[0079] In a specific embodiment, the construction process of the second loss function is as follows:
[0080] Construct the following physical constraints:
[0081] ;
[0082] ;
[0083] ;
[0084] in, Represents density in fluid data. express Regarding time Find the partial derivative. Represents velocity in fluid data. This represents the pressure in fluid data. Indicates energy. express Taking the partial derivative with respect to the flow field coordinates, express, Represents viscous stress. Indicates the thermal conductivity term;
[0085] Calculate the second-order central moment of the non-equilibrium term as a prediction of the viscous stress. Calculate the non-equilibrium term of the density distribution function. The first-order moment at the origin is used as the predicted value of the heat conduction term. The residual terms of physical constraints are calculated by combining the predicted values of the flow field data from the high-altitude rocket rarefied flow field reconstruction model. As shown in the following formula:
[0086] ;
[0087] in, This represents the predicted density value in the fluid data. This represents the predicted velocity value in the fluid data. This represents the predicted pressure value in the fluid data. This represents the j-th flow field coordinate randomly selected for calculating the residual term of the physical constraints;
[0088] The mean square error of the residual term of the physical constraints is calculated using the following formula to obtain the second loss function:
[0089] ;
[0090] in, This represents the second loss function. This represents the total number of flow field coordinates randomly selected for calculating the residual terms of physical constraints.
[0091] Specifically, embodiments of this application also introduce non-equilibrium terms based on the Boltzmann distribution function. The universal macroscopic equations and other physical constraints are used as the second loss function to ensure that the output of the high-altitude rocket rarefied flow field reconstruction model satisfies the basic laws of fluid mechanics, so that the prediction results of the high-altitude rocket rarefied flow field reconstruction model conform to the basic principles of thermodynamics.
[0092] In a specific embodiment, the construction process of the third loss function is as follows:
[0093] The mean square error between the flight test data and the predicted values of the flow field data output by the high-altitude rocket rarefied flow field reconstruction model is calculated to obtain the third loss function, as shown in the following equation:
[0094] ;
[0095] in, This represents the third loss function. This represents the total number of samples in the training data. This represents the k-th flow field coordinate in the training data; This represents the predicted value of the flow field data output after the k-th flow field coordinate in the training data is input into the high-altitude rocket rarefied flow field reconstruction model. This represents the flight test data corresponding to the k-th flow field coordinate in the training data;
[0096] The loss function used in the training of the high-altitude rocket rarefied flow field reconstruction model As shown in the following formula:
[0097] ;
[0098] in, , and These represent the weights corresponding to the first loss function, the second loss function, and the third loss function, respectively.
[0099] In a specific embodiment, the sensor data is the actual flow field data collected after the gradient sensitivity analysis method is used to identify regions with drastic flow field changes and regions with gentle flow field changes, based on the predicted values of the flow field data output by the high-altitude rocket rarefied flow field reconstruction model. The sensor distribution is adjusted in the regions with drastic flow field changes and the regions with gentle flow field changes respectively.
[0100] Specifically, the actual values of the flow field data collected in the dataset are the flight test data, which includes CFD simulation results and sensor data. The CFD simulation results are rarefied flow field data obtained by simulating typical flow field examples under different Kn numbers using computational fluid dynamics (CFD) methods. The predicted values of the flow field data output by the high-altitude rocket rarefied flow field reconstruction model are compared with the CFD simulation results to evaluate the reconstruction capability of the high-altitude rocket rarefied flow field reconstruction model in cross-basin rarefied flow fields, and to adjust the network structure or parameters of the high-altitude rocket rarefied flow field reconstruction model to improve accuracy. The sensor data are data collected by sensors on the surface of the high-altitude rocket. The sensor data collected by the high-altitude rocket surface sensors are compared with the predicted values of the flow field data output by the high-altitude rocket rarefied flow field reconstruction model to evaluate the degree of mismatch between the two. The comparison results will serve as the basis for training and optimizing the high-altitude rocket rarefied flow field reconstruction model. At the same time, data error is introduced into the loss function, making the model pay more attention to the matching degree of data during training, thereby improving the accuracy of flow field reconstruction. Therefore, a third loss function can be constructed based on the flight test data and the predicted values of the flow field data output by the high-altitude rocket rarefied flow field reconstruction model. By weighting the first, second, and third loss functions using their respective weights, the loss function used during the training of the high-altitude rocket rarefied flow field reconstruction model can be obtained. Based on the convergence of the loss function during the training process of the high-altitude rocket rarefied flow field reconstruction model, the weights of the first, second, and third loss functions are dynamically adjusted according to the magnitude ratio of the loss function to ensure the balance between physical constraints and flight test data in the model.
[0101] Furthermore, based on the predicted flow field data output by the high-altitude rocket rarefied flow field reconstruction model, gradient sensitivity analysis is used to identify regions of drastic flow field changes (such as shock waves and boundary layer separation), prioritizing the deployment of more sensors in these regions; while reducing sensor density in regions of gentle flow field changes. This maximizes model reconstruction accuracy with a limited sensor deployment.
[0102] S2 inputs the coordinates of the flow field to be predicted on the high-altitude rocket into the trained high-altitude rocket rarefied flow field reconstruction model to obtain the predicted values of the flow field data.
[0103] Specifically, the trained high-altitude rocket rarefied flow field reconstruction model is applied to the reconstruction of the rarefied flow field of the high-altitude rocket. By inputting the flow field coordinates to be predicted into the trained high-altitude rocket rarefied flow field reconstruction model, the predicted values of physical quantities such as density, velocity, and temperature in the flow field data can be obtained.
[0104] Further reference Figure 2As an implementation of the methods shown in the above figures, this application provides an embodiment of a high-altitude rocket rarefied flow field reconstruction device based on flight test data. This device embodiment is similar to... Figure 1 Corresponding to the method embodiments shown, this device can be specifically applied to various electronic devices.
[0105] This application provides a high-altitude rocket rarefied flow field reconstruction device based on flight test data, including:
[0106] Model building module 1 is configured to construct and train a high-altitude rocket rarefied flow field reconstruction model and a non-equilibrium term prediction model based on the Boltzmann-BGK equation using the Shakhov equilibrium distribution function, resulting in a trained high-altitude rocket rarefied flow field reconstruction model and a trained non-equilibrium term prediction model. The non-equilibrium term prediction model includes a first sub-network and a second sub-network set in parallel. The first and second sub-networks output non-equilibrium terms of the density distribution function and the temperature distribution function, respectively, based on the input flow field coordinates. The loss functions used during training include a first loss function based on the residual terms of the Boltzmann-BGK equation, a second loss function based on the residual terms of the physical constraints represented by the non-equilibrium terms in the Boltzmann-BGK equation, and a third loss function based on the flight test data and the predicted values of the flow field data output by the high-altitude rocket rarefied flow field reconstruction model based on the input flow field coordinates. The flight test data includes CFD simulation results and sensor data.
[0107] Prediction module 2 is configured to input the coordinates of the flow field to be predicted on the high-altitude rocket into the trained high-altitude rocket rarefied flow field reconstruction model to obtain the predicted values of the flow field data.
[0108] Figure 3 This is a schematic diagram of the hardware structure of an electronic device provided in an embodiment of the present invention. For example... Figure 3 As shown, the electronic device in this embodiment includes a processor 301 and a memory 302; wherein the memory 302 is used to store computer execution instructions; and the processor 301 is used to execute the computer execution instructions stored in the memory to implement the various steps performed by the electronic device in the above embodiment. For details, please refer to the relevant descriptions in the foregoing method embodiments.
[0109] Alternatively, the memory 302 can be either standalone or integrated with the processor 301.
[0110] When the memory 302 is set up independently, the electronic device also includes a bus 303 for connecting the memory 302 and the processor 301.
[0111] This invention also provides a computer storage medium storing computer execution instructions, which, when executed by processor 301, implement the above method.
[0112] This invention also provides a computer program product, including a computer program that, when executed by a processor 301, implements the above-described method.
[0113] In the embodiments provided by this invention, it should be understood that the disclosed devices and methods can be implemented in other ways. For example, the device embodiments described above are merely illustrative; for instance, the division of modules is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple modules may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be indirect coupling or communication connection through some interfaces, devices, or modules, and may be electrical, mechanical, or other forms.
[0114] The modules described as separate components may or may not be physically separate. The components shown as modules may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the modules can be selected to implement the solution of this embodiment according to actual needs.
[0115] Furthermore, the functional modules in the various embodiments of this invention can be integrated into one processing unit, or each module can exist physically separately, or two or more modules can be integrated into one unit. The unit formed by the above modules can be implemented in hardware or in the form of hardware plus software functional units.
[0116] The integrated modules implemented as software functional modules described above can be stored in a computer-readable storage medium. These software functional modules, stored in a storage medium, include several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) or processor 301 to execute some steps of the methods of the various embodiments of this application.
[0117] It should be understood that the processor 301 described above can be a Central Processing Unit (CPU), or other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), etc. The general-purpose processor can be a microprocessor, or the processor 301 can be any conventional processor 301. The steps of the method disclosed in this invention can be directly manifested as the hardware processor 301 executing the steps, or as a combination of hardware and software modules within the processor 301 executing the steps.
[0118] The memory 302 may include high-speed RAM memory, and may also include non-volatile memory (NVM), such as at least one disk storage device, and may also be a USB flash drive, portable hard drive, read-only memory, disk or optical disc, etc.
[0119] Bus 303 can be an Industry Standard Architecture (ISA), a Peripheral Component Interconnect (PCI) bus, or an Extended Industry Standard Architecture (EISA) bus, etc. Bus 303 can be divided into address bus, data bus, control bus, etc. For ease of illustration, the bus 303 in the accompanying drawings of this application is not limited to only one bus 303 or one type of bus 303.
[0120] The aforementioned storage medium can be implemented from 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 storage medium can be any available medium accessible to general-purpose or special-purpose computers.
[0121] An exemplary storage medium is coupled to a processor 301, enabling the processor 301 to read information from and write information to the storage medium. Alternatively, the storage medium can be an integral part of the processor 301. The processor 301 and the storage medium can reside in an application-specific integrated circuit (ASIC). Alternatively, the processor 301 and the storage medium can exist as discrete components in an electronic device or a host device.
[0122] Those skilled in the art will understand that all or part of the steps of the above-described method embodiments can be implemented by hardware related to program instructions. The aforementioned program can be stored in a computer-readable storage medium. When executed, the program performs the steps of the above-described method embodiments; and the aforementioned storage medium includes various media capable of storing program code, such as ROM, RAM, magnetic disks, or optical disks.
[0123] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and not to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some or all of the technical features; and these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of the present invention.
Claims
1. A method for reconstructing a rarefied flow field in a high-altitude rocket based on flight test data, characterized in that, Includes the following steps: A high-altitude rocket rarefied flow field reconstruction model and a non-equilibrium term prediction model based on the Boltzmann-BGK equation using the Shakhov equilibrium distribution function were constructed and trained to obtain the trained high-altitude rocket rarefied flow field reconstruction model and the trained non-equilibrium term prediction model. The non-equilibrium term prediction model includes a first sub-network and a second sub-network set in parallel. The first sub-network and the second sub-network output non-equilibrium terms of the density distribution function and the temperature distribution function, respectively, based on the input flow field coordinates. The loss functions used in the training process include a first loss function based on the residual terms of the Boltzmann-BGK equation, a second loss function based on the residual terms of the physical constraints represented by the non-equilibrium terms in the Boltzmann-BGK equation, and a third loss function based on the flight test data and the predicted values of the flow field data output by the high-altitude rocket rarefied flow field reconstruction model based on the input flow field coordinates. The flight test data includes CFD simulation results and sensor data. The coordinates of the flow field to be predicted on the high-altitude rocket are input into the trained high-altitude rocket rarefied flow field reconstruction model to obtain the predicted values of the flow field data.
2. The method for reconstructing a rarefied flow field of a high-altitude rocket based on flight test data according to claim 1, characterized in that, The flow field data includes density, velocity, pressure, and temperature, and the Boltzman-BGK equation based on the Shakhov equilibrium distribution function is expressed as: ； in, In the flow field coordinates and time At that location, the speed is The particles in The velocity distribution function in the 3D velocity space, Indicates the relaxation time. Indicates unbalanced terms. Represented by Maxwell's distribution function Adding the heat flux correction term to the Shakhov equilibrium distribution function, its expression is: ； ； Where Pr represents the Prandtl number, Velocity in fluid data The characteristic velocity in the vicinity, Indicates heat flux, Represents the gas constant. This represents the temperature in the fluid data. This represents the pressure in fluid data. It represents a vector in a space of dimension L=3-D.
3. The method for reconstructing a rarefied flow field of a high-altitude rocket based on flight test data according to claim 2, characterized in that, The construction process of the first loss function is as follows: The non-equilibrium term of the density distribution function predicted by the first and second subnetworks and the non-equilibrium term of the temperature distribution function Predicted values constituting the unbalanced term The predicted value of the Shakhov equilibrium distribution function is calculated from the predicted value of the flow field data obtained by the high-altitude rocket rarefied flow field reconstruction model. The predicted value of the velocity distribution function is calculated using the following formula. : ； Will Taking the partial derivatives with respect to time and flow field coordinates respectively, we obtain and , and then combine Substitute these terms into the Boltzmann-BGK equation and calculate the residual terms of the Boltzmann-BGK equation. As shown in the following formula: ； in, This represents the i-th flow field coordinate randomly selected for calculating the residual term of the Boltzmann-BGK equations; The mean square error of the residual term in the Boltzmann-BGK equation is calculated using the following formula to obtain the first loss function: ； in, Denotes the first loss function. This represents the total number of flow field coordinates randomly selected for calculating the residual terms of the Boltzmann-BGK equations; The construction process of the second loss function is as follows: Construct the following physical constraints: ； ； ； in, Represents density in fluid data. express Regarding time Find the partial derivative. Represents velocity in fluid data. This represents the pressure in fluid data. Indicates energy. express Taking the partial derivative with respect to the flow field coordinates, express, Represents viscous stress. Indicates the thermal conductivity term; The second central moment of the non-equilibrium term is calculated as a predicted value of the viscous stress. Calculate the non-equilibrium term of the density distribution function. The first-order moment at the origin is used as the predicted value of the heat conduction term. The residual terms of physical constraints are calculated by combining the predicted values of the flow field data from the high-altitude rocket rarefied flow field reconstruction model. As shown in the following formula: ； in, This represents the predicted density value in the fluid data. This represents the predicted velocity value in the fluid data. This represents the predicted pressure value in the fluid data. This represents the j-th flow field coordinate randomly selected for calculating the residual term of the physical constraints; The mean square error of the residual term of the physical constraints is calculated using the following formula to obtain the second loss function: ； in, This represents the second loss function. This represents the total number of flow field coordinates randomly selected for calculating the residual terms of physical constraints.
4. The method for reconstructing a rarefied flow field of a high-altitude rocket based on flight test data according to claim 3, characterized in that, The construction process of the third loss function is as follows: The mean square error between the flight test data and the predicted values of the flow field data output by the high-altitude rocket rarefied flow field reconstruction model is calculated to obtain the third loss function, as shown in the following formula: ； in, This represents the third loss function. This represents the total number of samples in the training data. This represents the k-th flow field coordinate in the training data; This represents the predicted value of the flow field data output after the k-th flow field coordinate in the training data is input into the high-altitude rocket rarefied flow field reconstruction model. This represents the flight test data corresponding to the k-th flow field coordinate in the training data; The loss function used in the training of the high-altitude rocket rarefied flow field reconstruction model As shown in the following formula: ； in, , and These represent the weights corresponding to the first loss function, the second loss function, and the third loss function, respectively.
5. The method for reconstructing a rarefied flow field of a high-altitude rocket based on flight test data according to claim 1, characterized in that, The sensor data is the actual flow field data collected after the gradient sensitivity analysis method is used to identify regions of drastic flow field change and regions of gentle flow field change, based on the predicted values of the flow field data output by the high-altitude rocket rarefied flow field reconstruction model. The sensor distribution is adjusted in the regions of drastic flow field change and regions of gentle flow field change, respectively.
6. The method for reconstructing a rarefied flow field of a high-altitude rocket based on flight test data according to claim 1, characterized in that, The high-altitude rocket rarefied flow field reconstruction model includes a fully connected network; both the first and second sub-networks use several deconvolution layers connected in sequence.
7. A high-altitude rocket rarefied flow field reconstruction device based on flight test data, characterized in that, include: The model building module is configured to construct and train a high-altitude rocket rarefied flow field reconstruction model and a non-equilibrium term prediction model based on the Boltzmann-BGK equation using the Shakhov equilibrium distribution function, resulting in a trained high-altitude rocket rarefied flow field reconstruction model and a trained non-equilibrium term prediction model. The non-equilibrium term prediction model includes a first sub-network and a second sub-network set in parallel. The first sub-network and the second sub-network output non-equilibrium terms of the density distribution function and the temperature distribution function, respectively, based on the input flow field coordinates. The loss functions used during training include a first loss function based on the residual terms of the Boltzmann-BGK equation, a second loss function based on the residual terms of the physical constraints represented by the non-equilibrium terms in the Boltzmann-BGK equation, and a third loss function based on the flight test data and the predicted values of the flow field data output by the high-altitude rocket rarefied flow field reconstruction model based on the input flow field coordinates. The flight test data includes CFD simulation results and sensor data. The prediction module is configured to input the flow field coordinates to be predicted on the high-altitude rocket into the trained high-altitude rocket rarefied flow field reconstruction model to obtain the predicted values of the flow field data.
8. An electronic device, comprising: One or more processors; Storage device for storing one or more programs. When the one or more programs are executed by the one or more processors, the one or more processors implement the method as described in any one of claims 1-6.
9. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the program is executed by the processor, it implements the method as described in any one of claims 1-6.
10. A computer program product, comprising a computer program, characterized in that, When the computer program is executed by a processor, it implements the method as described in any one of claims 1-6.