Error correction method and system for high-speed pneumatic prediction

Abstract

The invention discloses an error correction method and system for high-speed pneumatic prediction, and relates to the related field of data processing, and the method comprises the steps: determining aircraft structure design information, a preset flight path and flight parameters of a target aircraft; analyzing a flowing area experienced by the target aircraft entering the atmosphere based on a preset flight route and the flight parameters, and generating a multi-stage flowing area; gas molecule collision characteristics of the multi-stage flow area are analyzed, aerodynamic prediction is carried out in combination with aircraft structure design information, and multi-stage aerodynamic prediction data are generated; determining a transition area of the multi-stage flow area and transition area characteristics in a preset flight route; and training a transition correction layer based on the transition region features, and performing error correction on the multi-stage pneumatic prediction data. The technical problem that the prediction error of existing high-speed pneumatic prediction is large is solved, and the technical effect of improving the pneumatic prediction accuracy is achieved.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of data processing, and particularly relates to an error correction method and system for high-speed aerodynamic prediction. BACKGROUND

[0002] Aerodynamic prediction of high-speed vehicles is crucial for the design and safe operation of the vehicles, especially when entering the atmosphere. Accurate aerodynamic prediction can help optimize flight trajectories, improve fuel efficiency, and ensure flight safety. Currently, the main methods to solve this problem are through computational fluid dynamics (CFD) simulation and empirical formulas to predict aerodynamic characteristics. These methods can provide estimates of aerodynamic parameters to some extent. However, existing methods have limitations in dealing with complex flow phenomena when high-speed vehicles enter the atmosphere, especially in the transition region between different flow regions, where the variation of aerodynamic parameters is complex, resulting in large prediction errors.

[0003] In the related art, there is a technical problem of large prediction error in high-speed aerodynamic prediction. SUMMARY

[0004] The present application provides an error correction method and system for high-speed aerodynamic prediction, which determines the structural design parameters of the aircraft, the preset flight route and the flight conditions, analyzes the different flow regions experienced by the aircraft when entering the atmosphere based on these input parameters, divides the multi-stage flow region, combines the collision characteristics of gas molecules in different flow regions and the aircraft configuration for aerodynamic prediction, generates aerodynamic data for each stage, further identifies the transition region where the flow state changes on the flight trajectory, extracts its flow characteristics, and uses these characteristic data to train a transition correction model to correct the error of the preliminary aerodynamic prediction result across the flow domain, and obtain the full flight trajectory aerodynamic characteristic prediction result. Technical means such as technical means solve the technical problem of large prediction error in existing high-speed aerodynamic prediction, and achieve the technical effect of improving the accuracy of aerodynamic prediction.

[0005] The present application provides an error correction method for high-speed aerodynamic prediction, comprising: determining the aircraft structural design information of a target aircraft, a preset flight route and flight parameters; analyzing the flow region experienced by the target aircraft when entering the atmosphere based on the preset flight route and the flight parameters, generating a multi-stage flow region; analyzing the collision characteristics of gas molecules in the multi-stage flow region, combining the aircraft structural design information for aerodynamic prediction, generating multi-stage aerodynamic prediction data; determining the transition region of the multi-stage flow region and the transition region characteristics on the preset flight route; training a transition correction layer based on the transition region characteristics, and correcting the error of the multi-stage aerodynamic prediction data.

[0006] In a possible implementation, based on the preset flight path and the flight parameters, the flow regions experienced by the target spacecraft upon entering the atmosphere are analyzed to generate a multi-stage flow region, and the following processing is performed: the Mach number and Reynolds number at each flight position are calculated based on the preset flight path and the flight parameters; the molecular flow partitioning mechanism determined for the free molecular flow region, the rarefied transition flow region, and the continuous flow region is read; based on the molecular flow partitioning mechanism, the continuous region is divided according to the Mach number and Reynolds number at each flight position to generate a flow region map; and the multi-stage flow region is generated using the flow region map.

[0007] In a possible implementation, the gas molecule collision characteristics of the multi-stage flow region are analyzed, and aerodynamic prediction is performed in conjunction with the aircraft structural design information to generate multi-stage aerodynamic prediction data. The following processing is then performed: A pre-built molecular collision database is connected, and the collision frequencies between gas molecules in the free molecular flow region, the rarefied transition flow region, and the continuous flow region are read to generate free molecular flow collision parameters, rarefied transition flow molecular collision parameters, and continuous flow molecular collision parameters; based on the aircraft structural design information, local flow field difference analysis is performed to construct local flow field difference characteristics of the aircraft; combining the local flow field difference characteristics, free molecular flow collision parameters, rarefied transition flow molecular collision parameters, and continuous flow molecular collision parameters, multi-stage local molecular collision parameter analysis of the aircraft is performed to generate molecular collision parameters corresponding to each stage; using the molecular collision parameters corresponding to each stage, aerodynamic prediction is performed by calling the aerodynamic-molecular collision regression relationship trained based on the structural materials in the aircraft structural design information to generate the multi-stage aerodynamic prediction data.

[0008] In possible implementations, the following processing is performed: Local flow field difference analysis includes flow field difference analysis of the wing, tail and fuselage.

[0009] In a possible implementation, based on the aircraft structural design information, a local flow field difference analysis is performed to construct the aircraft's local flow field difference characteristics. The following processing is then performed: a model is created based on the aircraft structural design information to generate a three-dimensional geometric model and flow field correlation parameters; based on the three-dimensional geometric model and the flow field correlation parameters, CFD simulation is used to analyze the interaction between the aircraft surface and the flow field, identify the influence of the wing, tail, and fuselage on the flow field, and generate the aircraft's local flow field difference characteristics.

[0010] In possible implementations, the following processing is performed: the interaction analysis between the aircraft surface and the flow field includes analysis of surface aerodynamic heating, aerodynamic loads, and local differences in the flow field distribution.

[0011] In a possible implementation, the following processing is performed: the transition region includes a first type of transition region and a second type of transition region, wherein the first type of transition region is a transition zone from free molecular flow to rarefied transition flow, and the second type of transition region is a transition zone from rarefied transition flow to continuous flow; the transition region characteristics include a first flow field change characteristic corresponding to the first type of transition region and a second flow field change characteristic corresponding to the second type of transition region.

[0012] In a possible implementation, a transition correction layer is trained based on the transition region features to correct errors in the multi-stage aerodynamic prediction data. The following processing is performed: preset key features are extracted from the first flow field change features and the second flow field change features, including the changing trends of Mach number, Reynolds number, and molecular collision parameters, to generate a first key feature and a second key feature; a regression relationship between the changes in Mach number, Reynolds number, and molecular collision parameters and aerodynamic parameters is constructed; a first transition correction layer is trained based on the first key feature, and a second transition correction layer is trained based on the second key feature; the first transition correction layer and the second transition correction layer are used to correct errors in the multi-stage aerodynamic prediction data.

[0013] In a possible implementation, the following processing is performed: the transition region further includes a third type of transition region, which is an aerodynamic characteristic transition region; the transition characteristics of the third type of transition region are collected, including the aerodynamic heat distribution and aerodynamic load variation characteristics of the aircraft surface, and the third transition correction layer is trained.

[0014] This application also provides an error correction system for high-speed aerodynamic prediction, comprising: a flight information determination module for determining the target aircraft's structural design information, preset flight path, and flight parameters; a flow region analysis module for analyzing the flow regions experienced by the target aircraft upon entering the atmosphere based on the preset flight path and the flight parameters, and generating multi-stage flow regions; an aerodynamic prediction module for analyzing the gas molecule collision characteristics of the multi-stage flow regions, combining the aircraft structural design information to perform aerodynamic prediction, and generating multi-stage aerodynamic prediction data; a transition region feature determination module for determining the transition regions and transition region features of the multi-stage flow regions on the preset flight path; and an error correction module for training a transition correction layer based on the transition region features to correct errors in the multi-stage aerodynamic prediction data.

[0015] The proposed error correction method and system for high-speed aerodynamic prediction, as described in this application, first determines the target aircraft's structural design information, preset flight path, and flight parameters. Then, based on the preset flight path and flight parameters, it analyzes the flow regions experienced by the target aircraft upon entering the atmosphere, generating multi-stage flow regions. Next, it analyzes the gas molecule collision characteristics of these multi-stage flow regions and, combined with the aircraft's structural design information, performs aerodynamic prediction to generate multi-stage aerodynamic prediction data. Then, it determines the transition regions and characteristics of these multi-stage flow regions along the preset flight path. Finally, it trains a transition correction layer based on these transition region characteristics to correct errors in the multi-stage aerodynamic prediction data. This achieves the technical effect of improving the accuracy of aerodynamic prediction. Attached Figure Description

[0016] To more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings of the embodiments of the present invention will be briefly described below. Flowcharts are used in this application to illustrate the operations performed by the system according to the embodiments of the present application. It should be understood that the preceding or following operations are not necessarily performed precisely in sequence. Instead, various steps can be processed in reverse order or simultaneously as needed. Furthermore, other operations can be added to these processes, or one or more steps can be removed from these processes.

[0017] Figure 1 This is a flowchart illustrating the error correction method for high-speed aerodynamic prediction provided in an embodiment of this application.

[0018] Figure 2 This is a schematic diagram of the structure of an error correction system for high-speed aerodynamic prediction provided in an embodiment of this application.

[0019] Explanation of reference numerals in the attached diagram: Flight information determination module 10, flow region analysis module 20, aerodynamic prediction module 30, transition region feature determination module 40, error correction module 50. Detailed Implementation

[0020] The above description is merely an overview of the technical solution of this application. In order to better understand the technical means of this application and to implement it in accordance with the contents of the specification, and to make the above and other objects, features and advantages of this application more obvious and understandable, specific embodiments of this application are given below.

[0021] To make the objectives, technical solutions, and advantages of this application clearer, the following detailed description of this application will be provided in conjunction with the accompanying drawings. The described embodiments should not be considered as limitations on this application. All other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0022] In the following description, references to "some embodiments" describe a subset of all possible embodiments. However, it is understood that "some embodiments" can be the same or different subsets of all possible embodiments and can be combined with each other without conflict. The terms "first" and "second" are used merely to distinguish similar objects and do not represent a specific ordering of objects. The terms "comprising" and "having," and any variations thereof, are intended to cover non-exclusive inclusion. For example, a process, method, system, product, or server that includes a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or modules not explicitly listed or inherent to these processes, methods, products, or devices. Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this application belongs. The terminology used herein is for the purpose of describing embodiments of this application only.

[0023] This application provides an error correction method for high-speed aerodynamic prediction, such as... Figure 1 As shown, the method includes: Step S100: Determine the target aircraft's structural design information, preset flight route, and flight parameters.

[0024] Specifically, aircraft structural design information includes the aircraft's external dimensions, material properties, and mass distribution, used to describe the aircraft's physical structure. The preset flight path refers to the flight route planned by the aircraft according to the predetermined mission, expressed in latitude and longitude coordinates and altitude. Flight parameters refer to parameters that need to be considered during flight, such as speed, angle of attack, and sideslip angle.

[0025] Export the aircraft's 3D model file (e.g., STL format) from aircraft design software (such as CATIA, SolidWorks, etc.) to obtain the aircraft's 3D structural model data, including dimensions, material properties, and mass distribution. Extract the preset flight path (e.g., latitude and longitude coordinate sequence, altitude change curve) and flight parameters (e.g., velocity, angle of attack, sideslip angle, etc.) from the flight mission planning system. Transmit the above data to the aerodynamic prediction system via an API interface.

[0026] Step S200: Based on the preset flight route and the flight parameters, analyze the flow region experienced by the target aircraft as it enters the atmosphere, and generate a multi-stage flow region.

[0027] Specifically, the flow region refers to the area defined based on the aircraft's flight state in the atmosphere, such as hypersonic, supersonic, and subsonic. The target aircraft's structural design information, preset flight path, and flight parameters are imported into Computational Fluid Dynamics (CFD) software (such as ANSYS Fluent or OpenFOAM), and atmospheric environmental parameters (such as temperature and density distribution with altitude) are set. CFD simulations are run to generate flow field data for the target aircraft at different flight stages. A flow region partitioning algorithm is used to divide the flow field into multiple stages based on parameters such as Mach number and Reynolds number.

[0028] For example, when using ANSYS Fluent to simulate the flow field, the flow region is divided into three stages based on the speed and altitude of the target aircraft: hypersonic stage: Mach number > 5, corresponding altitude > 80 km; supersonic stage: 1 < Mach number ≤ 5, corresponding altitude 50 km - 80 km; subsonic stage: Mach number ≤ 1, corresponding altitude < 50 km.

[0029] In one possible implementation, based on the preset flight path and the flight parameters, the flow regions experienced by the target aircraft upon entering the atmosphere are analyzed to generate multi-stage flow regions. Step S200 further includes step S210, calculating the Mach number and Reynolds number for each flight position based on the preset flight path and the flight parameters. Specifically, the Mach number is the ratio of flight speed to the local speed of sound, used to describe the velocity state of the aircraft. The Reynolds number is a dimensionless number characterizing the fluid flow state, used to distinguish the type of flow region. The altitude and velocity for each flight position are extracted from the preset flight path and flight parameters. Temperature, pressure, and density at different altitudes are obtained using a standard atmospheric model (such as the International Standard Atmosphere ISA) or a custom atmospheric model. The Mach number is calculated based on the flight speed and the local speed of sound. The Reynolds number is calculated based on the flight speed, characteristic length (such as wing chord length), and fluid dynamic viscosity.

[0030] For example, for a specific flight location at an altitude of 50 km and a speed of 3000 m / s, calculations using the international standard atmospheric model yield the following results: local temperature 270 K, local speed of sound 330 m / s, and local density 0.001 kg / m³. 3 Dynamic viscosity 1.68×10 -5 Pa·s, characteristic length 1m (assumed to be a certain dimension of the aircraft, such as wing chord length). Mach number (Ma) is the ratio of flight speed to local speed of sound: Ma = 3000 / 330 ≈ 9.1. Reynolds number (Re) is the ratio of inertial force to viscous force, calculated as: Re = fluid density × fluid velocity × characteristic length / dynamic viscosity of fluid = 0.001 × 3000 × 1 / 1.68 × 10 -5 ≈1.79×10 5 .

[0031] Step S220: Read the molecular flow partitioning mechanism determined for the free molecular flow region, the rarefied transition flow region, and the continuous flow region. Specifically, the free molecular flow region refers to a region where the gas molecule collision frequency is extremely low, and the molecular motion is close to free flight. The rarefied transition flow region refers to a region where the gas molecule collision frequency is moderate, and the flow characteristics are between those of free molecular flow and continuous flow. The continuous flow region refers to a region where the gas molecule collision frequency is high, and the fluid behavior can be described by the continuous medium assumption. The partitioning criteria for the free molecular flow region, the rarefied transition flow region, and the continuous flow region are read from the database and loaded into the calculation module. For example, the partitioning criteria are shown in Table 1.

[0032] Table 1: Examples of molecular flow partitioning mechanisms: .

[0033] Step S230: Based on the molecular flow partitioning mechanism, continuous regions are divided according to the Mach number and Reynolds number of each flight position to generate a flow region map. Specifically, for each flight position, the flow region to which it belongs is determined based on the Mach number and Reynolds number, combined with the partitioning mechanism. Plotting tools (such as MATLAB or Python's Matplotlib) are used to integrate the flow region information of all flight positions to generate a flow region map. The flow region map is used to graphically display the distribution of different flow regions along the flight path.

[0034] Step S240: Generate the multi-stage flow region using the flow region map. Specifically, based on the flow region map, the flight path is divided into multiple stages, each stage corresponding to a different flow region type, and the flow region information for each stage is stored as a data file. For example, the generated multi-stage flow regions are shown in Table 2.

[0035] Table 2: Examples of multi-stage flow regions: .

[0036] This implementation, by calculating Mach and Reynolds numbers and combining them with a molecular flow partitioning mechanism, can precisely divide the flight path into free molecular flow regions, rarefied transition flow regions, and continuous flow regions. This precise partitioning helps to more accurately simulate the aerodynamic characteristics of an aircraft in different flow regions. By precisely partitioning the flow regions, more accurate input conditions can be provided for subsequent aerodynamic predictions, thereby improving the accuracy of aerodynamic predictions.

[0037] Step S300: Analyze the gas molecule collision characteristics of the multi-stage flow region, combine the aerodynamic prediction with the aircraft structural design information, and generate multi-stage aerodynamic prediction data.

[0038] Specifically, direct simulation Monte Carlo (DSMC) methods or continuous medium models (such as the Navier-Stokes equations) are used to analyze gas molecule collision characteristics. Gas molecule collision characteristics are parameters describing the collision behavior of gas molecules in the flow region, such as collision frequency and collision cross-section. Combined with aircraft structural design information, aerodynamic prediction algorithms (such as CFD-based aerodynamic coefficient calculations) are used to generate aerodynamic prediction data, such as lift coefficient and drag coefficient. This aerodynamic prediction data is used to describe the aerodynamic performance of the target aircraft in different flow regions.

[0039] In one possible implementation, the gas molecule collision characteristics of the multi-stage flow region are analyzed, and aerodynamic prediction is performed in conjunction with the aircraft structural design information to generate multi-stage aerodynamic prediction data. Step S300 further includes step S310, connecting to a pre-built molecular collision database, reading the collision frequencies between gas molecules corresponding to the free molecular flow region, the rarefied transition flow region, and the continuous flow region, and generating free molecular flow collision parameters, rarefied transition flow molecular collision parameters, and continuous flow molecular collision parameters. Specifically, the molecular collision database is a pre-built database that stores the collision frequencies and other relevant parameters of gas molecules in different flow regions (free molecular flow, rarefied transition flow, and continuous flow). The system connects to the pre-built molecular collision database. Based on the flow region type, the corresponding gas molecule collision frequencies and other relevant parameters are read to generate free molecular flow collision parameters, rarefied transition flow molecular collision parameters, and continuous flow molecular collision parameters.

[0040] For example, suppose the molecular collision database stores the following collision frequencies: Free molecular flow region: Collision frequency f free =10 3 s －1 rarefied transition flow region: collision frequency f trans =10 5 s －1 Continuous flow region: collision frequency f cont =10 7 s －1 The generated collision parameters are shown in Table 3.

[0041] Table 3: Examples of collision parameters: .

[0042] Step S320: Based on the aircraft structural design information, perform local flow field difference analysis to construct the local flow field difference characteristics of the aircraft. Specifically, import the three-dimensional structural model of the target aircraft into CFD software, run local flow field simulation, calculate the flow field distribution on and around the aircraft surface, and extract the difference characteristics of the local flow field, such as velocity gradient, pressure gradient, temperature gradient, density gradient, etc. For example, the local flow field difference characteristics of the target aircraft are shown in Table 4.

[0043] Table 4: Examples of local flow field differences: .

[0044] Step S330 involves performing multi-stage local molecular collision parameter analysis of the aircraft by combining the local flow field difference characteristics, free molecular flow collision parameters, rarefied transition flow molecular collision parameters, and continuous flow molecular collision parameters, generating molecular collision parameters for each stage. Specifically, the difference characteristics of the local flow field are extracted from the flow field simulation of the aircraft. The velocity gradient represents the rate of change of velocity in the flow field; a larger velocity gradient indicates a more drastic velocity change. The pressure gradient represents the rate of change of pressure in the flow field; a larger pressure gradient indicates a more drastic pressure change. The temperature gradient represents the rate of change of temperature in the flow field; a larger temperature gradient indicates a more drastic temperature change. The density gradient represents the rate of change of density in the flow field; a larger density gradient indicates a more drastic density change. The influence of these local flow field difference characteristics on the gas molecule collision parameters is analyzed. Specifically, the velocity gradient affects the relative velocity of gas molecules. In regions with larger velocity gradients, the relative velocity of gas molecules increases, leading to an increase in collision frequency. For example, in regions with drastic velocity changes, such as the nose or tail of the aircraft, the collision frequency will be higher than in other regions. Pressure gradients affect the density distribution of gas molecules. In regions with larger pressure gradients, the density of gas molecules is higher, leading to an increase in collision frequency and collision cross-section. Temperature gradients affect the thermal motion of gas molecules. In regions with larger temperature gradients, the thermal motion of gas molecules is more intense, also increasing collision frequency and collision cross-section. For example, in high-temperature regions of an aircraft (such as near engine nozzles), the large temperature gradient causes significant changes in collision parameters. Density gradients affect the distribution of gas molecules. In regions with large density gradients, the distribution of gas molecules is uneven, and the collision frequency changes accordingly. For example, in the edge regions of an aircraft, the density gradient is large, and the collision parameters will differ.

[0045] For each flow region stage, the molecular collision parameters are adjusted based on local flow field characteristics. The specific steps are as follows: In the free molecular flow region, the gas molecule collision frequency is low, but local flow field characteristics (such as velocity gradient and temperature gradient) still affect the collision parameters. For example, in regions with large velocity gradients, the collision frequency increases slightly. In the rarefied transition flow region, the gas molecule collision frequency is moderate, and the influence of local flow field characteristics on the collision parameters is more significant. For example, in regions with large pressure and temperature gradients, the collision frequency and collision cross-section increase significantly. In the continuous flow region, the gas molecule collision frequency is high, and the influence of local flow field characteristics on the collision parameters is also significant. For example, in regions with large density and pressure gradients, the collision frequency and collision cross-section increase further. By comprehensively considering the effects of velocity gradient, pressure gradient, temperature gradient, and density gradient, the original collision parameters for each flow region stage are adjusted to generate the molecular collision parameters for each stage.

[0046] Step S340: Using the molecular collision parameters corresponding to each stage, the aerodynamic-molecular collision regression relationship trained based on the structural materials in the aircraft structural design information is invoked to perform aerodynamic prediction, generating the multi-stage aerodynamic prediction data. Specifically, the regression relationship is a prediction model established using the aircraft structural materials and based on machine learning or statistical methods, used to predict aerodynamic characteristics based on input parameters. First, the regression model trained based on the aircraft structural materials is loaded. For each stage, the corresponding molecular collision parameters are input, the regression model is invoked, and the aerodynamic characteristics of the aircraft in different flow regions, such as lift coefficient and drag coefficient, are predicted through calculation or modeling, generating aerodynamic prediction data for each stage.

[0047] The aerodynamic-molecular collision regression model is based on a three-layer fully connected neural network model. This model achieves accurate prediction through the strong correlation between input multi-dimensional features and output aerodynamic parameters. Its core logic is as follows: The input layer receives 6-dimensional fused data, including molecular collision frequency, collision cross section, velocity gradient, pressure gradient, temperature gradient, and density gradient. These data directly correspond to the physical characteristics of different flow regions and the flow field influence of the local structure of the aircraft. The hidden layer extracts nonlinear correlation features through the ReLU activation function of 32 neurons, and then transforms the original features into higher-order features that can reflect aerodynamic laws through a deeper mapping of 16 neurons. At the same time, the Dropout layer avoids overfitting. The output layer directly outputs the lift coefficient and drag coefficient. During model training, historical flight data and high-precision CFD simulation data corresponding to the target aircraft's structural materials are first collected. The training, validation, and test sets are then divided in a 7:2:1 ratio. The Adam optimizer is used to minimize the mean squared error loss. A batch size of 32 and 100 training epochs are set. Learning rate decay is initiated when the validation set loss shows no decrease for 10 consecutive epochs, with a decay coefficient of 0.5. Finally, the optimal model with a test set accuracy ≥95% is saved. In practical applications, for different regions of free molecular flow, rarefied transition flow, and continuous flow, the model automatically matches the molecular collision parameters and local flow field differences of the corresponding region as input, and outputs aerodynamic parameters adapted to the characteristics of that region.

[0048] This approach, by considering the differences in local flow fields, can more accurately adjust molecular collision parameters, thereby improving the accuracy of aerodynamic prediction. This provides a more accurate basis for the structural design and material selection of aircraft, helping to optimize the aerodynamic performance of aircraft.

[0049] In one possible implementation, step S320 further includes: local flow field difference analysis including flow field difference analysis of the wing, tail and fuselage parts.

[0050] Specifically, the wing is the primary component generating lift in an aircraft, and its flow field characteristics are crucial to the aircraft's performance. The tail fin's main function is to provide directional stability and control torque. The fuselage is the main structural component of the aircraft, and its flow field characteristics affect the aircraft's drag and stability. This approach, through refined analysis of local flow field differences, particularly detailed analysis of the wing, tail, and fuselage, allows for more precise adjustment of molecular collision parameters.

[0051] In one possible implementation, based on the aircraft structural design information, a local flow field difference analysis is performed to construct the local flow field difference characteristics of the aircraft. Step S320 further includes step S321, which involves modeling based on the aircraft structural design information to generate a three-dimensional geometric model and flow field correlation parameters. Specifically, the aircraft structural design information includes the aircraft's external shape design, including the dimensions and shapes of the wings, tail, and fuselage, as well as material properties such as density and thermal conductivity, and flight parameters such as flight speed, altitude, and angle of attack. A three-dimensional geometric model is generated using computer-aided design (CAD) software (such as SolidWorks or CATIA) based on the aircraft's structural design information. The aircraft surface and surrounding space are divided into small mesh cells for CFD simulation. Boundary conditions for the aircraft surface and surrounding environment, such as velocity, pressure, and temperature, are defined, and the initial state of the flow field, such as the initial velocity field and pressure field, is defined, generating flow field correlation parameters.

[0052] Step S322: Based on the three-dimensional geometric model and the flow field correlation parameters, CFD simulation is used to analyze the interaction between the aircraft surface and the flow field, identify the influence of the wing, tail, and fuselage on the flow field, and generate the local flow field difference characteristics of the aircraft. Specifically, computational fluid dynamics (CFD) software (such as ANSYS Fluent or OpenFOAM) is used for flow field simulation. The three-dimensional geometric model and flow field correlation parameters are input, the simulation is run, and the interaction between the aircraft surface and the surrounding flow field is calculated. The velocity distribution on the aircraft surface is analyzed, especially the velocity gradients of the wing, tail, and fuselage; the pressure distribution on the aircraft surface is analyzed, especially the pressure gradients of the wing, tail, and fuselage; the temperature distribution on the aircraft surface is analyzed, especially the temperature gradients of the wing, tail, and fuselage; the density distribution on the aircraft surface is analyzed, especially the density gradients of the wing, tail, and fuselage. The flow field difference characteristics of the wing, tail, and fuselage are extracted, including velocity gradient, pressure gradient, temperature gradient, and density gradient. These characteristics are stored as data files for subsequent use. The flow field difference features were extracted using a multi-scale feature fusion algorithm. This algorithm designed differentiated extraction logic for the dominant flow field characteristics of different parts of the wing, tail, and fuselage, ensuring a high degree of consistency between the features and the aerodynamic influence laws. First, the raw velocity, pressure, temperature, and density fields output from the CFD simulation were smoothed using Gaussian filtering to remove simulation noise. Based on the aircraft's three-dimensional geometric model, the wing, tail, and fuselage mesh regions were automatically segmented. A 1mm fine-scale mesh was used for the lift / moment dominant characteristics of the wing and tail, while a 5mm medium-scale mesh was used for the drag dominant characteristics of the fuselage. This differentiated mesh resolution improved the targeting of feature extraction. The central difference method was used to calculate the velocity, pressure, temperature, and density gradients at each local region's mesh points. Finally, the gradient values ​​were normalized according to region type: the gradient value of the wing region was divided by the maximum gradient value of the wing, and the gradient value of the fuselage region was divided by the maximum gradient value of the fuselage. The output was a standardized 4D feature vector, which directly reflects the differentiated influence of different local structures on the flow field. This approach, through CFD simulation, can accurately extract the differences in flow field characteristics between the spacecraft surface and its surroundings, providing more accurate data support for subsequent molecular collision parameter analysis.

[0053] In one possible implementation, step S322 further includes: the interaction analysis between the aircraft surface and the flow field includes the analysis of local differences in surface aerodynamic heating, aerodynamic loads, and flow field distribution.

[0054] Specifically, surface aerodynamic heating analysis refers to analyzing the aerodynamic heating effect on the surface of an aircraft. Aerodynamic load analysis refers to analyzing the distribution of aerodynamic loads on the surface of an aircraft. Local difference analysis of flow field distribution refers to analyzing the local differences in the flow field between the aircraft surface and its surroundings. Surface aerodynamic heating analysis primarily focuses on heat flux density and temperature distribution. Heat flux density refers to the heat flow per unit area, and temperature distribution reflects the concentrated areas of heat load. These data directly affect the thermal motion of gas molecules, thus influencing molecular collision parameters. Aerodynamic loads refer to the aerodynamic forces and moments acting on the surface of the aircraft. Aerodynamic load analysis primarily focuses on pressure and load distribution. Pressure distribution reflects the distribution of lift and drag, while load distribution helps assess the aerodynamic performance and structural strength requirements of the aircraft. These data directly affect the density distribution of gas molecules, thus influencing molecular collision parameters. Local difference analysis of flow field distribution primarily focuses on velocity distribution, pressure distribution, temperature distribution, and density distribution. Velocity distribution reflects the flow characteristics of the flow field, pressure distribution reflects the pressure characteristics, temperature distribution reflects the thermal characteristics, and density distribution reflects the density characteristics. These data directly affect the relative velocity and density distribution of gas molecules, thus influencing molecular collision parameters.

[0055] This approach allows for more precise adjustment of molecular collision parameters through detailed analysis of local differences in surface aerodynamic heating, aerodynamic loads, and flow field distribution. This method not only improves the accuracy of aerodynamic predictions but also provides crucial data support for aircraft design and optimization.

[0056] Step S400: Determine the transition region and transition region characteristics of the multi-stage flow region in the preset flight path.

[0057] Specifically, the transition region is the part between multiple flow stages, and its characteristic parameters change in a complex manner. Within a pre-defined flight path, numerical analysis methods (such as interpolation and fitting) are used to determine the transition region between flow stages, and characteristic parameters of the transition region, such as the rate of change of aerodynamic coefficients and temperature gradients, are extracted. The transition region and its characteristic parameters are then stored as a data file.

[0058] In one possible implementation, step S400 further includes: the transition region includes a first type of transition region and a second type of transition region, wherein the first type of transition region is a transition zone from free molecular flow to rarefied transition flow, and the second type of transition region is a transition zone from rarefied transition flow to continuous flow; the transition region features include a first flow field change feature corresponding to the first type of transition region and a second flow field change feature corresponding to the second type of transition region.

[0059] Specifically, the first type of transition region is the transition zone from free molecular flow to rarefied flow. In this region, the collision frequency of gas molecules gradually increases, but remains at a low level. The flow field changes are mainly characterized by: a gradual increase in collision frequency from low to moderate; significant changes in velocity, pressure, temperature, and density gradients; and complex changes in aerodynamic coefficients (such as lift and drag coefficients), requiring special attention. Due to the large variations in collision frequency and flow field gradients in the first type of transition region, aerodynamic prediction errors may be high. Detailed analysis of these characteristics is used to adjust the aerodynamic prediction model and reduce errors.

[0060] The second type of transition region is the transition zone from rarefied to continuous flow. In this region, the collision frequency of gas molecules further increases, and the flow field gradually tends towards continuous flow characteristics. The main characteristics of flow field changes are: an increase in collision frequency from moderate to high; gradual stabilization of velocity, pressure, temperature, and density gradients; and gradual stabilization of aerodynamic coefficients, although the influence of local flow field differences still needs to be considered. Because the collision frequency and flow field gradients gradually stabilize in the second type of transition region, the aerodynamic prediction error is relatively low, but the influence of local flow field differences still needs to be considered.

[0061] This approach, by refining the classification and feature analysis of transition regions, can more accurately identify the location and characteristics of these transition regions, providing a basis for subsequent error correction, thereby optimizing the aerodynamic prediction model and improving the accuracy and reliability of predictions.

[0062] Step S500: Train a transition correction layer based on the transition region features to correct errors in the multi-stage aerodynamic prediction data.

[0063] Specifically, the transition correction layer is used to correct errors in aerodynamic prediction data, making it closer to the true values. Linear regression, multinomial regression, or other regression methods are used to establish the relationship between transition region characteristic parameters and aerodynamic coefficient correction values. Historical flight data or high-precision CFD simulation data are used as training datasets. By training the transition correction layer, the regression model is used to calculate aerodynamic coefficient correction values, thus correcting errors in multi-stage aerodynamic prediction data.

[0064] In one possible implementation, a transition correction layer is trained based on the transition region features to correct errors in the multi-stage aerodynamic prediction data. Step S500 further includes step S510, extracting preset key features from the first and second flow field change features, including the changing trends of Mach number, Reynolds number, and molecular collision parameters, to generate a first key feature and a second key feature. Specifically, the changing trends of Mach number, Reynolds number, and molecular collision parameters (such as collision frequency and collision cross-section) in the first type of transition region are analyzed, and these trends are integrated into the first key feature. The changing trends of Mach number, Reynolds number, and molecular collision parameters in the second type of transition region are analyzed, and these trends are integrated into the second key feature.

[0065] Step S520: Construct a regression relationship between changes in Mach number, Reynolds number, and molecular collision parameters and aerodynamic parameters. Then, train a first transition correction layer based on the first key feature and a second transition correction layer based on the second key feature. Specifically, use changes in Mach number, Reynolds number, and molecular collision parameters as independent variables, and corrected values ​​of aerodynamic parameters (such as lift coefficient and drag coefficient) as dependent variables. Establish regression relationships between these variables using historical flight data or high-precision CFD simulation data. Train the first transition correction layer using the first key feature (key features of the first type of transition region) and train the second transition correction layer using the second key feature (key features of the second type of transition region).

[0066] Step S530: Error correction is performed on the multi-stage aerodynamic prediction data using the first transition correction layer and the second transition correction layer. Specifically, for aerodynamic prediction data in the first type of transition region, the first transition correction layer is used for error correction, adjusting the predicted values ​​of the aerodynamic coefficients according to the first key feature to reduce errors. For aerodynamic prediction data in the second type of transition region, the second transition correction layer is used for error correction, adjusting the predicted values ​​of the aerodynamic coefficients according to the second key feature to reduce errors.

[0067] This approach, by extracting key features, constructing regression relationships, and training transition correction layers, can more accurately correct multi-stage aerodynamic prediction data. This not only improves the accuracy of aerodynamic prediction but also provides important data support for aircraft design and optimization.

[0068] In one possible implementation, step S500 further includes step S540, wherein the transition region further includes a third type of transition region, which is an aerodynamic characteristic transition region; the transition characteristics of the third type of transition region are collected, including the aerodynamic heat distribution and aerodynamic load variation characteristics of the aircraft surface, and the third transition correction layer is trained.

[0069] Specifically, the third transition region is the aerodynamic characteristic transition region, focusing on changes in aerodynamic characteristics (aerodynamic heat distribution and aerodynamic loads). This region covers the entire process from free molecular flow to continuous flow. The aerodynamic heat distribution on the aircraft surface is analyzed, including heat flux density and temperature distribution. Changes in aerodynamic loads on the aircraft surface are analyzed, including lift coefficient, drag coefficient, and moment coefficient. The aerodynamic heat distribution and aerodynamic load variation characteristics for each flow region are collected. Historical flight data or high-precision CFD simulation data are used as training datasets. Regression relationships are constructed, using the aerodynamic heat distribution and aerodynamic load variation characteristics as inputs and aerodynamic coefficient correction values ​​as outputs. The model is trained using the training dataset, adjusting model parameters to minimize the error between predicted and true values. The trained model is the third transition correction layer, used to adjust the predicted values ​​of aerodynamic coefficients. For the aerodynamic prediction data of each flow region, the third transition correction layer is used for error correction. Based on the aerodynamic heat distribution and aerodynamic load variation characteristics, the predicted values ​​of aerodynamic coefficients are adjusted to reduce errors. This implementation, by introducing a third transition correction layer, can more accurately correct aerodynamic prediction data, reduce errors, and improve prediction accuracy.

[0070] The first and second transition correction layers employ a gradient boosting regression tree (GBRT) model, while the third transition correction layer uses a convolutional neural network (CNN) model. Both models achieve error correction through precise matching of features with aerodynamic parameter correction requirements. The core logic is as follows: For the first and second transition correction layers, the model uses a decision tree of depth 5 as the base learner. A feature mapping module standardizes key features such as the rate of change of Mach number and Reynolds number. A gradient calculation module captures the residual gradient between the predicted and actual aerodynamic parameter values ​​in real time. A loss update module iteratively reduces the residual. During training, a learning rate of 0.01, a subsample ratio of 0.8, and a feature sampling ratio of 0.7 are set. Training stops when the validation set residual is less than 0.001. In application, the model inputs the key feature sequence corresponding to the transition region and outputs the correction amounts for the lift and drag coefficients. The multi-stage aerodynamic prediction data is directly corrected using the formula: predicted value + correction amount. For the third transition correction layer, the CNN model is designed based on the two-dimensional distribution features of the aerodynamic characteristic transition region. The input layer receives three-dimensional data H×W×C of the aerodynamic and thermal distribution and aerodynamic load distribution of the aircraft surface grid, where H and W are the number of rows and columns of the grid, and C=2. Spatial distribution features are extracted through two layers of convolution and pooling operations, and then the feature mapping is deepened through a fully connected layer, finally outputting the accurate correction amount. During training, the SGD optimizer and Huber loss function are used, combined with data augmentation techniques to improve generalization ability. After training, models with a correction error ≤3% on the validation set are saved. When applied, the two-dimensional distribution features of the transition region are directly input, and the correction amount is output to complete the final optimization of the aerodynamic parameters of the entire flight trajectory.

[0071] This application employs techniques such as determining the structural design parameters of the aircraft, the preset flight path, and flight conditions, analyzing the different flow regions experienced by the aircraft when entering the atmosphere based on these input parameters, dividing the flow regions into multiple stages, combining the collision characteristics of gas molecules in different flow regions and the aircraft configuration to perform aerodynamic prediction, generating aerodynamic data for each stage, further identifying transition regions where the flow state changes on the flight trajectory, extracting their flow characteristics, using these characteristic data to train a transition correction model, and performing cross-flow region error correction on the preliminary aerodynamic prediction results to obtain the full flight trajectory aerodynamic characteristic prediction results. These techniques solve the technical problem of large prediction errors in existing high-speed aerodynamic predictions and achieve the technical effect of improving the accuracy of aerodynamic predictions.

[0072] In the above text, refer to Figure 1 An error correction method for high-speed aerodynamic prediction according to an embodiment of the present invention is described in detail. Next, reference will be made to... Figure 2 An error correction system for high-speed aerodynamic prediction according to an embodiment of the present invention is described.

[0073] The error correction system for high-speed aerodynamic prediction according to embodiments of the present invention is used to solve the technical problem of large prediction errors in existing high-speed aerodynamic prediction systems, thereby improving the accuracy of aerodynamic prediction. The error correction system for high-speed aerodynamic prediction includes: a flight information determination module 10, a flow region analysis module 20, an aerodynamic prediction module 30, a transition region feature determination module 40, and an error correction module 50.

[0074] The flight information determination module 10 is used to determine the target aircraft's structural design information, preset flight route, and flight parameters; the flow region analysis module 20 is used to analyze the flow regions experienced by the target aircraft upon entering the atmosphere based on the preset flight route and the flight parameters, and generate multi-stage flow regions; the aerodynamic prediction module 30 is used to analyze the gas molecule collision characteristics of the multi-stage flow regions, and perform aerodynamic prediction in conjunction with the aircraft structural design information, generating multi-stage aerodynamic prediction data; the transition region feature determination module 40 is used to determine the transition regions and transition region features of the multi-stage flow regions on the preset flight route; and the error correction module 50 is used to train a transition correction layer based on the transition region features and perform error correction on the multi-stage aerodynamic prediction data.

[0075] The specific configuration of the flow region analysis module 20 will be described in detail below. As mentioned above, based on the preset flight path and the flight parameters, the flow region experienced by the target spacecraft upon entering the atmosphere is analyzed to generate a multi-stage flow region. The flow region analysis module 20 may further include: a Mach-Reynolds number calculation unit for calculating the Mach number and Reynolds number at each flight position based on the preset flight path and the flight parameters; a molecular flow partitioning mechanism reading unit for reading the molecular flow partitioning mechanism determined for the free molecular flow region, the rarefied transition flow region, and the continuous flow region; a continuous region segmentation unit for segmenting the continuous region based on the molecular flow partitioning mechanism and the Mach number and Reynolds number at each flight position to generate a flow region map; and a multi-stage flow region generation unit for generating the multi-stage flow region from the flow region map.

[0076] The specific configuration of the aerodynamic prediction module 30 will be described in detail below. As described above, the gas molecule collision characteristics of the multi-stage flow region are analyzed, and aerodynamic prediction is performed in conjunction with the aircraft structural design information to generate multi-stage aerodynamic prediction data. The aerodynamic prediction module 30 may further include: a collision parameter generation unit for connecting to a pre-built molecular collision database, reading the collision frequencies between gas molecules corresponding to the free molecular flow region, the rarefied transition flow region, and the continuous flow region, and generating free molecular flow collision parameters, rarefied transition flow molecular collision parameters, and continuous flow molecular collision parameters; a local flow field difference analysis unit for performing local flow field difference analysis based on the aircraft structural design information and constructing local flow field difference characteristics of the aircraft; a local molecular collision parameter analysis unit for performing multi-stage local molecular collision parameter analysis of the aircraft by combining the local flow field difference characteristics, free molecular flow collision parameters, rarefied transition flow molecular collision parameters, and continuous flow molecular collision parameters of the aircraft, and generating molecular collision parameters for each stage; and an aerodynamic prediction unit for using the molecular collision parameters for each stage to call the aerodynamic-molecular collision regression relationship trained based on the structural materials in the aircraft structural design information to perform aerodynamic prediction and generate the multi-stage aerodynamic prediction data.

[0077] The local flow field difference analysis unit may further include: local flow field difference analysis includes flow field difference analysis of the wing, tail and fuselage.

[0078] Specifically, the local flow field difference analysis is performed based on the aircraft structural design information to construct the local flow field difference characteristics of the aircraft. The local flow field difference analysis unit may further include: a modeling subunit for modeling according to the aircraft structural design information, generating a three-dimensional geometric model and flow field correlation parameters; and a CFD simulation subunit for analyzing the interaction between the aircraft surface and the flow field through CFD simulation based on the three-dimensional geometric model and the flow field correlation parameters, identifying the influence of the wing, tail and fuselage on the flow field, and generating the local flow field difference characteristics of the aircraft.

[0079] The CFD simulation sub-unit can further include: the interaction analysis between the aircraft surface and the flow field, including the analysis of local differences in surface aerodynamic heating, aerodynamic loads, and flow field distribution.

[0080] The specific configuration of the transition region feature determination module 40 will be described in detail below. As mentioned above, the transition region feature determination module 40 may further include: the transition region includes a first type of transition region and a second type of transition region, wherein the first type of transition region is a transition zone from free molecular flow to rarefied transition flow, and the second type of transition region is a transition zone from rarefied transition flow to continuous flow; the transition region features include a first flow field change feature corresponding to the first type of transition region and a second flow field change feature corresponding to the second type of transition region.

[0081] The specific configuration of the error correction module 50 will be described in detail below. As mentioned above, the transition correction layer is trained based on the transition region features to correct the error in the multi-stage aerodynamic prediction data. The error correction module 50 may further include: a preset key feature extraction unit for extracting preset key features from the first flow field change features and the second flow field change features, including the changing trends of Mach number, Reynolds number, and molecular collision parameters, and generating a first key feature and a second key feature; a transition correction layer training unit for constructing the regression relationship between the changes in Mach number, Reynolds number, and molecular collision parameters and aerodynamic parameters, and then training a first transition correction layer based on the first key feature and a second transition correction layer based on the second key feature; and an error correction unit for correcting the error in the multi-stage aerodynamic prediction data using the first transition correction layer and the second transition correction layer.

[0082] The error correction module 50 may further include: a third transition correction layer training unit for the transition region including a third type of transition region, the third type of transition region being an aerodynamic characteristic transition region, collecting the transition characteristics of the third type of transition region, including the aerodynamic heat distribution and aerodynamic load change characteristics of the aircraft surface, and training the third transition correction layer.

[0083] The error correction system for high-speed aerodynamic prediction provided in this embodiment of the invention can execute the error correction method for high-speed aerodynamic prediction provided in any embodiment of the invention, and has the corresponding functional modules and beneficial effects of the method.

[0084] Although this application makes various references to certain modules in the system according to the embodiments of this application, any number of different modules can be used and run on user terminals and / or servers. The various units and modules included are only divided according to functional logic, but are not limited to the above division, as long as the corresponding functions can be achieved; in addition, the specific names of each functional unit are only for easy distinction between each other and are not used to limit the scope of protection of this invention.

[0085] The specific embodiments described above do not constitute a limitation on the scope of protection of this application. Those skilled in the art should understand that various modifications, combinations, and substitutions can be made according to design requirements and other factors. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of this application should be included within the scope of protection of this application. In some cases, the actions or steps described in this application can be performed in a different order than that shown in the embodiments and still achieve the desired results. Furthermore, the processes depicted in the accompanying drawings do not necessarily require a specific or sequential order to achieve the desired results. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

Claims

1. An error correction method for high-speed aerodynamic prediction, characterized in that, include: Determine the target aircraft's structural design information, preset flight path, and flight parameters; Based on the preset flight path and the flight parameters, the flow region experienced by the target aircraft upon entering the atmosphere is analyzed, and a multi-stage flow region is generated. The collision characteristics of gas molecules in the multi-stage flow region are analyzed, and aerodynamic prediction is performed in combination with the aircraft structural design information to generate multi-stage aerodynamic prediction data. The transition zone and characteristics of the multi-stage flow region are determined in the preset flight path. A transition correction layer is trained based on the transition region features to correct errors in the multi-stage aerodynamic prediction data.

2. The error correction method for high-speed aerodynamic prediction as described in claim 1, characterized in that, Based on the preset flight path and the flight parameters, the flow region experienced by the target aircraft upon entering the atmosphere is analyzed, and a multi-stage flow region is generated, including: Calculate the Mach number and Reynolds number for each flight position based on the preset flight route and the flight parameters; Read the molecular flow partitioning mechanisms determined for the free molecular flow region, the rarefied transition flow region, and the continuous flow region; Based on the molecular flow partitioning mechanism, continuous regions are divided according to the Mach number and Reynolds number of each flight position to generate a flow region map; The multi-stage flow region is generated using the flow region map.

3. The error correction method for high-speed aerodynamic prediction as described in claim 1, characterized in that, Analyzing the gas molecule collision characteristics of the multi-stage flow region and combining them with the aircraft structural design information, aerodynamic prediction is performed to generate multi-stage aerodynamic prediction data, including: Connect to a pre-built molecular collision database, read the collision frequencies between gas molecules in the free molecular flow region, the rarefied transition flow region, and the continuous flow region, and generate free molecular flow collision parameters, rarefied transition flow molecular collision parameters, and continuous flow molecular collision parameters. Based on the aircraft structural design information, a local flow field difference analysis is performed to construct the local flow field difference characteristics of the aircraft. By combining the local flow field difference characteristics of the aircraft, the collision parameters of free molecular flow, the collision parameters of rarefied transition flow and the collision parameters of continuous flow, a multi-stage local molecular collision parameter analysis of the aircraft is performed to generate the molecular collision parameters corresponding to each stage. Using the molecular collision parameters corresponding to each stage, aerodynamic prediction is performed by calling the aerodynamic-molecular collision regression relationship trained based on the structural materials in the spacecraft structural design information, thereby generating the multi-stage aerodynamic prediction data.

4. The error correction method for high-speed aerodynamic prediction as described in claim 3, characterized in that, Local flow field difference analysis includes flow field difference analysis of the wing, tail and fuselage.

5. The error correction method for high-speed aerodynamic prediction as described in claim 3, characterized in that, Based on the aforementioned aircraft structural design information, a local flow field difference analysis is performed to construct the aircraft's local flow field difference characteristics, including: Modeling is performed based on the aircraft structural design information to generate a three-dimensional geometric model and flow field correlation parameters; Based on the three-dimensional geometric model and the flow field correlation parameters, CFD simulation is used to analyze the interaction between the aircraft surface and the flow field, identify the influence of the wing, tail and fuselage on the flow field, and generate the local flow field difference characteristics of the aircraft.

6. The error correction method for high-speed aerodynamic prediction as described in claim 5, characterized in that, in, The interaction analysis between the aircraft surface and the flow field includes analysis of local differences in surface aerodynamic heating, aerodynamic loads, and flow field distribution.

7. The error correction method for high-speed aerodynamic prediction as described in claim 1, characterized in that, The transition region includes a first type of transition region and a second type of transition region, wherein the first type of transition region is a transition zone from free molecular flow to rarefied transition flow, and the second type of transition region is a transition zone from rarefied transition flow to continuous flow. The transition region features include the first flow field change features corresponding to the first type of transition region and the second flow field change features corresponding to the second type of transition region.

8. The error correction method for high-speed aerodynamic prediction as described in claim 7, characterized in that, A transition correction layer is trained based on the transition region features to correct errors in the multi-stage aerodynamic prediction data, including: Preset key features are extracted from the first flow field change features and the second flow field change features, including the changing trends of Mach number, Reynolds number and molecular collision parameters, to generate the first key feature and the second key feature. The regression relationship between changes in Mach number, Reynolds number, and molecular collision parameters and aerodynamic parameters is constructed. Then, a first transition correction layer is trained based on the first key feature, and a second transition correction layer is trained based on the second key feature. The first transition correction layer and the second transition correction layer are used to correct errors in the multi-stage aerodynamic prediction data.

9. The error correction method for high-speed aerodynamic prediction as described in claim 8, characterized in that, The transition region also includes a third type of transition region, which is an aerodynamic characteristic transition region; The transition characteristics of the third type of transition region are collected, including the aerodynamic heat distribution and aerodynamic load variation characteristics of the aircraft surface, and the third transition correction layer is trained.

10. An error correction system for high-speed aerodynamic prediction, characterized in that, The system is used to implement the error correction method for high-speed aerodynamic prediction according to any one of claims 1-9, and the system comprises: The flight information determination module is used to determine the target aircraft's structural design information, preset flight route, and flight parameters. The flow region analysis module is used to analyze the flow region experienced by the target aircraft as it enters the atmosphere based on the preset flight route and the flight parameters, and to generate a multi-stage flow region. The aerodynamic prediction module is used to analyze the collision characteristics of gas molecules in the multi-stage flow region, combine the aerodynamic prediction with the aircraft structural design information, and generate multi-stage aerodynamic prediction data. The transition region feature determination module is used to determine the transition region and transition region features of the multi-stage flow region on the preset flight path. An error correction module is used to train a transition correction layer based on the transition region features and to correct errors in the multi-stage aerodynamic prediction data.

Images