Airfoil aerodynamic force coefficient prediction method based on attribute reduction, electronic device and storage medium

Abstract

The attribute reduction-based airfoil aerodynamic force coefficient prediction method, electronic equipment and storage medium belong to the technical field of aircrafts.To solve the problem of poor accuracy in predicting the aerodynamic force coefficient of an airfoil under other shape geometric parameters or state parameters, the airfoil is parameterized by CST to obtain airfoil CST parameters representing the shape geometric parameters of the airfoil, then attribute reduction is performed to obtain the attributes of the reduced airfoil, wind tunnel test or CFD numerical simulation is performed to obtain the aerodynamic data of the reduced airfoil, and then a prediction model is established for the obtained aerodynamic data of the reduced airfoil by using an adaptive boosting method and is trained until the target evaluation condition is met.The present application effectively eliminates the influence of data redundancy, reduces the training difficulty of the prediction model, helps to improve the training efficiency, and further improves the prediction accuracy of the aerodynamic force coefficient.

Description

Technical Field

[0001] This invention belongs to the field of aircraft technology, specifically relating to a method for predicting airfoil aerodynamic coefficients based on attribute reduction, electronic equipment, and storage medium. Background Technology

[0002] The two-dimensional interface profiles of aircraft wings, aero-engine propellers, turbine blades, and compressor blades are all airfoils. Aerodynamic optimization design of airfoils plays a crucial role in the aerospace field, requiring extensive wind tunnel testing and CFD calculations. Currently, wind tunnel testing often involves designing a test outline and determining the number of wind tunnel runs based on test requirements. For example, initial tests might be conducted at Mach 0.6, 0.8, 1.0, and 1.2. Subsequent research might require altering some geometric parameters to achieve test data at Mach 0.9, 0.95, and 1.1. Traditional wind tunnel test data analysis methods often rely on linear interpolation for prediction. Similarly, CFD numerical simulations may yield aerodynamic data for a specific airfoil geometry, but optimization of the shape is necessary. This requires re-meshing, numerical calculations, and post-processing for aerodynamic shapes with different geometric parameters. However, traditional interpolation methods and re-performing CFD numerical simulations to obtain aerodynamic characteristics under different geometric or state parameters suffer from poor accuracy, low efficiency, and high cost. Aerodynamic optimization design of airfoils involves a large number of geometric parameters and high data dimensionality, which can lead to the curse of dimensionality, resulting in decreased prediction accuracy and a tendency to overfit. Summary of the Invention

[0003] The problem this invention aims to solve is that traditional interpolation methods and re-perform CFD numerical simulations to obtain the aerodynamic coefficients of airfoils under other shape geometry parameters or state parameters suffer from poor accuracy, low efficiency, high cost, and large number of geometric parameters and high data dimensionality, which can lead to the curse of dimensionality, decreased prediction accuracy, and easy overfitting. The invention proposes an airfoil aerodynamic coefficient prediction method, electronic equipment, and storage medium based on attribute reduction.

[0004] To achieve the above objectives, the present invention provides the following technical solution:

[0005] A method for predicting airfoil aerodynamic coefficients based on attribute reduction includes the following steps:

[0006] S1. Perform CST parameterization on the airfoil to obtain the airfoil CST parameters that characterize the airfoil's shape geometry.

[0007] S2. Perform attribute reduction on the airfoil CST parameters obtained in step S1 to obtain the reduced airfoil attributes;

[0008] S3. Perform wind tunnel tests or CFD numerical simulations on the airfoil reduced properties in step S2 to obtain the aerodynamic data after airfoil reduction.

[0009] S4. A prediction model is established and trained on the aerodynamic data after the airfoil reduction obtained in step S3 using an adaptive boosting method until the target evaluation conditions are met.

[0010] Furthermore, the specific implementation method of step S1 includes the following steps:

[0011] S1.1 The CST parameterization method describes the airfoil using a category function and a shape function, expressed as follows:

[0012]

[0013] in, The dimensionless longitudinal coordinates of the airfoil after CST parameterization; The horizontal coordinate is dimensionless. x is the lateral coordinate of the airfoil curve, and C is the airfoil chord length; For category functions, For shape function, τ is used to control the thickness at the trailing edge of the airfoil. T τ is the dimensionless trailing edge thickness. T =Δz TE / C, Δz TE The thickness of the trailing edge;

[0014] S1.2, Set the expression for the category function as follows:

[0015]

[0016] Wherein, N1 and N2 are coefficients that control the shape of the leading edge and trailing edge of the airfoil, respectively, and the values ​​of N1 and N2 range from 0 to 1;

[0017] S1.3, Set the expression for the shape function as follows:

[0018]

[0019] Where i is the i-th term, n is the n-th term, and A i As a weighting factor, For shape function components, Replace with Bernstein polynomials, the expression for the nth term of the Bernstein polynomial. for:

[0020]

[0021] but:

[0022]

[0023] S1.4, By controlling the weighting factor A i To achieve control over the airfoil geometry, airfoil CST parameters, which characterize the airfoil's shape geometry, are obtained.

[0024] Furthermore, step S2 performs attribute reduction on the airfoil CST parameters obtained in step S1. The attribute reduction method is one of principal component analysis, multidimensional scaling, or t-distribution random neighborhood embedding. The CST parameter data obtained in step S1 is set to be weighted by the upper surface weighting factor A. ui and lower surface weighting factor A li The CST parameter dataset S1 obtained in step S1 is: S1 = {A} u1 A u2 ,…,A u(n+1) A l1 A l2 ,…,A l(n+1)};

[0025] The reduced dataset S2 is:

[0026] S2={B1,B2,…B m}, m≤2(n+1)

[0027] Among them, B m This represents the reduced parameter data.

[0028] Furthermore, the specific implementation process of principal component analysis in step S2 is as follows: For the CST parameter dataset S1 = {A} obtained in step S1... u1 A u2 ,…,A u(n+1) A l1 A l2 ,…,A l(n+1)}, calculate the pulsation quantity:

[0029]

[0030] Among them, A j For the j-th item in dataset S1, j = 1, ..., 2(n+1), we obtain matrix A. j Then calculate the autocorrelation matrix A'A'. T A' is all A j The matrix is ​​constructed, and the autocorrelation matrix is ​​decomposed into eigenvalues ​​to obtain the eigenvectors W and their eigenvalues, thus obtaining the low-dimensional output data B = WA.

[0031] Furthermore, the specific implementation process of the multidimensional scaling method in step S2 is as follows: Data points are distinguished by calculating the Euclidean distance. The original 2(n+1)-dimensional CST parameter dataset S1 is input, and the distance B between any two samples after dimensionality reduction is calculated. m The distance is equal to the distance in 2(n+1)-dimensional space, resulting in the reduced dataset S2.

[0032] Furthermore, the specific implementation method of the t-distribution random neighborhood embedding method in step S2 is as follows: The gradient formula is simplified using the cost function of the symmetric SNE; the Gaussian distribution is replaced by the t-distribution; the original 2(n+1)-dimensional CST parameter dataset S1 is input; the number of iterations and the learning rate parameters are set; and the probability distribution p is calculated. ij The distance between two points is expressed by replacing the Gaussian distribution with the t-distribution. By comparing the probability distribution after dimensionality reduction with the probability distribution of the original space, the reduced dataset S2 is obtained.

[0033] Furthermore, the aerodynamic data obtained in step S3 includes: target independent variable data and target dependent variable data. The target independent variable data includes: reduced CST parameters of the airfoil, as well as Mach number Ma, Reynolds number Re, and angle of attack state parameter data α. The target dependent variable data includes six-dimensional aerodynamic coefficients, used to characterize the aerodynamic performance of the model. The target dependent variable data is obtained by adjusting the target independent variable data through wind tunnel tests on a vehicle-by-vehicle basis or through CFD numerical simulations on a parameter-by-parameter basis.

[0034] The six-element aerodynamic coefficient is specifically the wind shaft lift coefficient C. L Wind axis drag coefficient C D Wind axis lateral force coefficient C C Body axis pitching moment coefficient C m Body axis yaw moment coefficient C n and body axis rolling moment coefficient C l This constitutes the dataset Y;

[0035] The airfoil aerodynamic data can be obtained through existing wind tunnel tests, computer CFD numerical simulations, or Xfoil's dedicated airfoil aerodynamic calculation tool.

[0036] Furthermore, step S4 uses the Adaboost algorithm to process the reduced target independent variable data X = {B1, B2, ..., B...} m {Ma,Re,α} and the target dependent variable data Y={C L C D C C C m C n C lThe established prediction model is trained until the target evaluation conditions are met; the target evaluation conditions compare the mean square error and the mean absolute error as prediction errors.

[0037] An electronic device includes a memory and a processor, the memory storing a computer program, and the processor executing the computer program to implement the steps of the airfoil aerodynamic coefficient prediction method based on attribute reduction.

[0038] A computer-readable storage medium having a computer program stored thereon, characterized in that, when the computer program is executed by a processor, it implements the aforementioned airfoil aerodynamic coefficient prediction method based on attribute reduction.

[0039] The beneficial effects of this invention are:

[0040] The present invention discloses an airfoil aerodynamic coefficient prediction method based on attribute reduction. By performing attribute reduction on airfoil CST parameter data, valuable low-dimensional structures are extracted from high-dimensional data, thereby effectively eliminating the influence of data redundancy, reducing the training difficulty of neural networks, helping to improve training efficiency, and thus improving the prediction accuracy of aerodynamic coefficients.

[0041] The airfoil aerodynamic coefficient prediction method based on attribute reduction described in this invention can effectively reduce the number of wind tunnel tests and CFD numerical simulations required in traditional methods, thereby reducing testing costs, computational costs, and R&D costs, saving design time, and effectively improving R&D progress. It can also provide a reference for the aerodynamic design of other missile, aircraft, and other models. Attached Figure Description

[0042] Figure 1 This is a flowchart of an airfoil aerodynamic coefficient prediction method based on attribute reduction, as described in this invention.

[0043] Figure 2 This is a simplified schematic diagram of the properties of the airfoil aerodynamic coefficient prediction method based on property reduction described in this invention. Detailed Implementation

[0044] 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 and specific embodiments. It should be understood that the specific embodiments described herein are only for explaining the invention and are not intended to limit the invention; that is, the described specific embodiments are merely a part of the embodiments of the invention, and not all of them. The components of the specific embodiments of the invention described and shown in the accompanying drawings can generally be arranged and designed in various different configurations, and the invention may also have other embodiments.

[0045] Therefore, the following detailed description of specific embodiments of the invention provided in the accompanying drawings is not intended to limit the scope of the claimed invention, but merely to illustrate selected specific embodiments of the invention. All other specific embodiments obtained by those skilled in the art based on these specific embodiments without inventive effort are within the scope of protection of this invention.

[0046] To further understand the invention's content, features, and effects, the following specific embodiments are provided, along with accompanying drawings. Figure 1 - Appendix Figure 2 Detailed explanation is as follows: Specific implementation method one:

[0048] A method for predicting airfoil aerodynamic coefficients based on attribute reduction includes the following steps:

[0049] S1. Perform CST parameterization on the airfoil to obtain the airfoil CST parameters that characterize the airfoil's shape geometry.

[0050] Furthermore, the specific implementation method of step S1 includes the following steps:

[0051] S1.1 The CST parameterization method describes the airfoil using a category function and a shape function, expressed as follows:

[0052]

[0053] in, The dimensionless longitudinal coordinates of the airfoil after CST parameterization; The horizontal coordinate is dimensionless. x is the lateral coordinate of the airfoil curve, and C is the airfoil chord length; For category functions, For shape function, τ is used to control the thickness at the trailing edge of the airfoil. T τ is the dimensionless trailing edge thickness. T =Δz TE / C, Δz TE The thickness of the trailing edge;

[0054] S1.2, Set the expression for the category function as follows:

[0055]

[0056] Wherein, N1 and N2 are coefficients that control the shape of the leading edge and trailing edge of the airfoil, respectively, and the values ​​of N1 and N2 range from 0 to 1;

[0057] S1.3, Set the expression for the shape function as follows:

[0058]

[0059] Where i is the i-th term, n is the n-th term, and A i As a weighting factor, For shape function components, Replace with Bernstein polynomials, the expression for the nth term of the Bernstein polynomial. for:

[0060]

[0061] but:

[0062]

[0063] S1.4, By controlling the weighting factor A i To achieve control over the airfoil geometry, airfoil CST parameters characterizing the airfoil's shape geometry are obtained;

[0064] Furthermore, the higher the order of the Bernstein polynomial, the smaller the reconstruction error between the CST-parameterized airfoil and the original airfoil. The order of the Bernstein polynomial n is greater than 6 to ensure the reconstruction accuracy of the airfoil.

[0065] S2. Perform attribute reduction on the airfoil CST parameters obtained in step S1 to obtain the reduced airfoil attributes;

[0066] Furthermore, step S2 performs attribute reduction on the airfoil CST parameters obtained in step S1. The attribute reduction method is one of principal component analysis, multidimensional scaling, or t-distribution random neighborhood embedding. The CST parameter data obtained in step S1 is set to be weighted by the upper surface weighting factor A. ui and lower surface weighting factor A li The CST parameter dataset S1 obtained in step S1 is: S1 = {A} u1 A u2 ,…,A u(n+1) A l1 A l2 ,…,A l(n+1)}

[0067] The reduced dataset is as follows:

[0068] S2={B1,B2,…B m}, m≤2(n+1)

[0069] Among them, B i Represents the reduced parameter data;

[0070] Furthermore, the specific implementation method of principal component analysis in step S2 is as follows: For the CST parameter dataset S1 = {A} obtained in step S1...u1 A u2 ,…,A u(n+1) A l1 A l2 ,…,A l(n+1)}, calculate the pulsation quantity:

[0071]

[0072] Among them, A j For the j-th item in dataset S1, j = 1, ..., 2(n+1), we obtain matrix A. j Then calculate the autocorrelation matrix A'A'. T A' is all A j The matrix formed by 'A' and the autocorrelation matrix A'A' T By performing eigenvalue decomposition to obtain the eigenvector W and its eigenvalues, the low-dimensional output data B = WA' is obtained.

[0073] Furthermore, the specific implementation process of the multidimensional scaling method in step S2 is as follows: Data points are distinguished by calculating the Euclidean distance. The original 2(n+1)-dimensional CST parameter dataset S1 is input, and the distance B between any two samples after dimensionality reduction is calculated. m This is equal to the distance in 2(n+1)-dimensional space, resulting in the reduced dataset S2;

[0074] Furthermore, a cost function with symmetry in the SNE is used to simplify the gradient formula, and the Gaussian distribution is replaced by the t-distribution. The original 2(n+1)-dimensional CST parameter dataset S1 is input, and the number of iterations and learning rate parameters are set to calculate the probability distribution p. ij The distance between two points is expressed by replacing the Gaussian distribution with the t-distribution. The probability distribution after dimensionality reduction is compared with the probability distribution of the original space to obtain the reduced dataset S2.

[0075] Furthermore, given the large design space for aerodynamic geometry parameters, using methods such as Principal Components Analysis (PCA), Multidimensional Scaling (MDS), and t-Stochastic Neighbor Embedding (t-SNE) for attribute reduction can represent high-dimensional data with low-dimensional data. This effectively eliminates the impact of data redundancy, alleviates the curse of dimensionality, reduces the training difficulty of neural networks, helps improve training efficiency, and thus improves the prediction accuracy of aerodynamic characteristics.

[0076] S3. Perform wind tunnel tests or CFD numerical simulations on the airfoil reduced properties in step S2 to obtain the aerodynamic data after airfoil reduction.

[0077] Furthermore, the aerodynamic data obtained in step S3 includes: target independent variable data and target dependent variable data. The target independent variable data includes: reduced CST parameters of the airfoil, as well as Mach number Ma, Reynolds number Re, and angle of attack state parameter data α. The target dependent variable data includes six-dimensional aerodynamic coefficients, used to characterize the aerodynamic performance of the model. The target dependent variable data is obtained by adjusting the target independent variable data through wind tunnel tests on a vehicle-by-vehicle basis or through CFD numerical simulations on a parameter-by-parameter basis.

[0078] The six-element aerodynamic coefficient is specifically the wind shaft lift coefficient C. L Wind axis drag coefficient C D Wind axis lateral force coefficient C C Body axis pitching moment coefficient C m Body axis yaw moment coefficient C n and body axis rolling moment coefficient C l This constitutes the dataset Y;

[0079] The airfoil aerodynamic data can be obtained through existing wind tunnel tests, computer CFD numerical simulations, or Xfoil's dedicated airfoil aerodynamic calculation tool.

[0080] Furthermore, the methods for obtaining airfoil aerodynamic data are as follows:

[0081] Step A: Organize wind tunnel test data, including test task information table and train schedule;

[0082] Step B: Compile different standard templates for conventional force measurement tests for different wind tunnel data storage formats;

[0083] Step C: Develop a data extraction interface to transform the raw data;

[0084] Step D: Unlike wind tunnel test data, CFD numerical simulation data does not require conversion of the original data. Extract the shape geometry parameters and state parameters, such as Mach number, Reynolds number, and angle of attack, from the organized wind tunnel test data and CFD numerical simulation data.

[0085] Step E: Extract the target dependent variable data, including the force and moment coefficients, including the lift coefficient, drag coefficient, lateral force coefficient, pitching moment coefficient, yaw moment coefficient, and roll moment coefficient;

[0086] Step F: Clean all target independent and dependent variable data, remove duplicate and missing data, and normalize the obtained data to obtain data with consistent dimensions.

[0087] Furthermore, the acquired target independent variable data and target dependent variable data are in one-to-one correspondence, specifically, each set of target independent variable data corresponds to a set of target dependent variable data;

[0088] S4. Use an adaptive boosting method to build a prediction model for the aerodynamic data after airfoil reduction obtained in step S3 and train it until the target evaluation conditions are met.

[0089] Furthermore, step S4 uses the Adaboost algorithm to process the reduced target independent variable data X = {B1, B2, ..., B...} m {Ma,Re,α} and the target dependent variable data Y={C L C D C C C m C n C l The established prediction model is trained until the target evaluation conditions are met; the target evaluation conditions compare the mean square error and the mean absolute error as prediction errors.

[0090] Preferably, after determining the target evaluation conditions, a target termination threshold is set as the training termination condition. That is, when the target evaluation conditions of the trained prediction model are less than or equal to the target termination threshold, the training terminates and the model is used as the target prediction model.

[0091] Preferably, the samples during training can be divided into the following three sets: training set, test set, and validation set. The training set is used for initial training of the model; the test set is used for parameter tuning of the initially trained model and for secondary training of the initially trained model; the validation set is used to validate the model trained in the secondary training using data from the validation set, and the target prediction model is obtained after successful validation.

[0092] Preferably, the training sample allocation ratio is 6:2:2 (training set:test set:validation set); or 6:3:1 (training set:test set:validation set); or 7:2:1 (training set:test set:validation set); or 7:1.5:1.5 (training set:test set:validation set).

[0093] To prevent overfitting, i.e., to prevent the phenomenon that the training error is very low but the test error is relatively high, the preferred method is to correct the model by adjusting the distribution ratio of the training set, test set and validation set.

[0094] The extraction of independent variables through the attribute reduction stage reduces the training difficulty of the prediction model, thereby helping to improve training results, reduce training time, and improve the accuracy of aerodynamic characteristic prediction. Specific Implementation Method Two:

[0096] An electronic device includes a memory and a processor. The memory stores a computer program, and the processor executes the computer program to implement the steps of the airfoil aerodynamic coefficient prediction method based on attribute reduction as described in Specific Embodiment 1.

[0097] The computer device of the present invention may include a processor and a memory, such as a microcontroller containing a central processing unit. Furthermore, when the processor executes the computer program stored in the memory, it implements the steps of the aforementioned recommendation method for modifyable relationship-driven recommendation data based on CREO software.

[0098] The processor referred to can be a Central Processing Unit (CPU), or other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. A general-purpose processor can be a microprocessor or any conventional processor.

[0099] The memory may primarily include a program storage area and a data storage area. The program storage area may store the operating system and at least one application program required for a function (such as sound playback, image playback, etc.); the data storage area may store data created based on the use of the mobile phone (such as audio data, phonebook, etc.). Furthermore, the memory may include high-speed random access memory, and may also include non-volatile memory, such as hard disks, RAM, plug-in hard disks, smart media cards (SMC), secure digital cards (SD cards), flash cards, at least one disk storage device, flash memory device, or other volatile solid-state storage devices. Specific implementation method three:

[0101] A computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the airfoil aerodynamic coefficient prediction method based on attribute reduction as described in Specific Embodiment 1.

[0102] The computer-readable storage medium of the present invention can be any form of storage medium that can be read by the processor of a computer device, including but not limited to non-volatile memory, volatile memory, ferroelectric memory, etc. The computer-readable storage medium stores a computer program. When the processor of the computer device reads and executes the computer program stored in the memory, the steps of the above-described modeling method for modifyable relation-driven modeling data based on CREO software can be implemented.

[0103] The computer program includes computer program code, which may be in the form of source code, object code, executable file, or some intermediate form. The computer-readable medium may include: any entity or device capable of carrying the computer program code, recording media, USB flash drive, portable hard drive, magnetic disk, optical disk, computer memory, read-only memory (ROM), random access memory (RAM), electrical carrier signals, telecommunication signals, and software distribution media, etc. It should be noted that the content included in the computer-readable medium may be appropriately added to or subtracted according to the requirements of legislation and patent practice in the jurisdiction. For example, in some jurisdictions, according to legislation and patent practice, computer-readable media may not include electrical carrier signals and telecommunication signals.

[0104] It should be noted that relational terms such as "first" and "second" are used merely to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitations, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes said element.

[0105] Although this application has been described above with reference to specific embodiments, various modifications can be made and components can be replaced with equivalents without departing from the scope of this application. In particular, as long as there is no structural conflict, the features in the specific embodiments disclosed in this application can be combined with each other in any way. The lack of an exhaustive description of these combinations in this specification is merely for the sake of brevity and resource conservation. Therefore, this application is not limited to the specific embodiments disclosed herein, but includes all technical solutions falling within the scope of the claims.

Claims

1. A method for predicting airfoil aerodynamic coefficients based on attribute reduction, characterized in that: Includes the following steps: S1. Perform CST parameterization on the airfoil to obtain the airfoil CST parameters that characterize the airfoil's shape geometry. S2. Perform attribute reduction on the airfoil CST parameters obtained in step S1 to obtain the reduced airfoil attributes; Step S2 performs attribute reduction on the airfoil CST parameters obtained in step S1. The attribute reduction method is one of principal component analysis, multidimensional scaling, or t-distribution random neighborhood embedding. The CST parameter data obtained in step S1 is set to be weighted by the upper surface weight factor. and lower surface weighting factor The CST parameter dataset S1 obtained in step S1 is composed of: ; The reduced dataset S2 is: ； B m represent the reduced parameter data; S3. Perform wind tunnel tests or CFD numerical simulations on the airfoil reduced properties in step S2 to obtain the aerodynamic data after airfoil reduction. The aerodynamic data obtained in step S3 includes: target independent variable data and target dependent variable data. The target independent variable data includes: reduced CST parameters of the airfoil, as well as Mach number Ma, Reynolds number Re, and angle of attack state parameter data α. The target dependent variable data includes six-dimensional aerodynamic coefficients, used to characterize the aerodynamic performance of the model. The target dependent variable data is obtained by adjusting the target independent variable data through wind tunnel tests on a vehicle-by-vehicle basis or through CFD numerical simulations on a parameter-by-parameter basis. The six-element aerodynamic coefficient is specifically the wind shaft lift coefficient. Wind axis drag coefficient lateral force coefficient of wind axis Body axis pitching moment coefficient Body axis yaw moment coefficient and body axis rolling moment coefficient This constitutes the dataset Y; The airfoil aerodynamic data can be obtained through existing wind tunnel tests, computer CFD numerical simulations, or Xfoil's dedicated airfoil aerodynamic calculation tool. S4. Use an adaptive boosting method to build a prediction model for the aerodynamic data after airfoil reduction obtained in step S3 and train it until the target evaluation conditions are met. Step S4 uses the Adaboost algorithm to process the reduced target variable data. and target dependent variable data The established prediction model is trained until the target evaluation conditions are met; the target evaluation conditions compare the mean square error and the mean absolute error as prediction errors.

2. The method for predicting airfoil aerodynamic coefficients based on attribute reduction according to claim 1, characterized in that: The specific implementation method of step S1 includes the following steps: S1.1 The CST parameterization method describes the airfoil using a category function and a shape function, expressed as follows: ； in, The dimensionless longitudinal coordinates of the airfoil after CST parameterization; The horizontal coordinate is dimensionless. x is the lateral coordinate of the airfoil curve, and C is the airfoil chord length; For category functions, For shape function, This is used to control the thickness at the airfoil's trailing edge. The dimensionless trailing edge thickness, , The thickness of the trailing edge; S1.2, Set the expression for the category function as follows: ； Wherein, N1 and N2 are coefficients that control the shape of the leading edge and trailing edge of the airfoil, respectively, and the values ​​of N1 and N2 range from 0 to 1; S1.3, Set the expression for the shape function as follows: ； Where i is the i-th term and n is the n-th term. As a weighting factor, For shape function components, Replace with Bernstein polynomials, the expression for the nth term of the Bernstein polynomial. for: ； but: ； S1.4, By controlling the weighting factor To achieve control over the airfoil geometry, airfoil CST parameters, which characterize the airfoil's shape geometry, are obtained.

3. The method for predicting airfoil aerodynamic coefficients based on attribute reduction according to claim 2, characterized in that: The specific implementation process of principal component analysis in step S2 is as follows: For the CST parameter dataset obtained in step S1, Calculate the pulsation quantity: ； in, For the j-th item in dataset S1, , to obtain the matrix Then calculate the autocorrelation matrix. , For all The constructed matrix, and the autocorrelation matrix By performing eigenvalue decomposition to obtain the eigenvector W and its eigenvalues, the low-dimensional output data B=WA' is obtained.

4. The method for predicting airfoil aerodynamic coefficients based on attribute reduction according to claim 3, characterized in that: The specific implementation process of the multidimensional scaling method in step S2 is as follows: Data points are distinguished by calculating the Euclidean distance. The original 2(n+1)-dimensional CST parameter dataset S1 is input, and the distance B between any two samples after dimensionality reduction is calculated. m The distance is equal to the distance in 2(n+1)-dimensional space, resulting in the reduced dataset S2.

5. The method for predicting airfoil aerodynamic coefficients based on attribute reduction according to claim 4, characterized in that: The gradient formula is simplified using a cost function with symmetry in the SNE, and the Gaussian distribution is replaced by a t-distribution. The original 2(n+1)-dimensional CST parameter dataset S1 is input, and the number of iterations and learning rate parameters are set. The probability distribution p is then calculated. ij The distance between two points is expressed by replacing the Gaussian distribution with the t-distribution. By comparing the probability distribution after dimensionality reduction with the probability distribution of the original space, the reduced dataset S2 is obtained.

6. An electronic device, characterized in that, It includes a memory and a processor, the memory storing a computer program, and the processor executing the computer program to implement the steps of the airfoil aerodynamic coefficient prediction method based on attribute reduction as described in any one of claims 1-5.

7. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by the processor, it implements the airfoil aerodynamic coefficient prediction method based on attribute reduction as described in any one of claims 1-5.

Images