Multi-body separation modeling method based on wind tunnel model release test data

Abstract

The invention relates to a multi-body separation modeling method based on wind tunnel model release test data. The method comprises the following steps: 1, preparing and preprocessing test data; 2, performing polynomial fitting on the output time sequence data; step 3, fitting precision analysis; 4, input parameters and polynomial coefficients are subjected to normalization processing; 5, analyzing the correlation between input and output parameters; step 6, establishing a regression prediction model; and 7, verifying the model. When the method is used for processing the test data of the wind tunnel launching model, polynomial fitting is carried out on the trajectory and attitude time sequence data, a multi-body separation prediction model is established directly based on a small amount of wind tunnel model launching test data, and the cost-reducing and efficiency-increasing requirements of modern aerospace engineering are met.

Description

Technical Field

[0001] This invention provides a method for predicting the trajectory and attitude of a wind tunnel model deployment test of an aircraft, belonging to the field of multi-body separation technology for aircraft. Background Technology

[0002] Multi-body separation is a critical problem that must be solved in the separation scenarios of aerospace systems. Currently, the most commonly used research methods for multi-body separation include numerical calculation, wind tunnel testing, and flight testing. Wind tunnel testing, as an important experimental method, plays an irreplaceable role in aircraft design and performance optimization. Unlike wind tunnel CTS testing and grid force measurement testing, wind tunnel model deployment testing can simulate the unsteady characteristics of the multi-body separation process of aircraft, and is the unsteady dynamic wind tunnel testing method with the closest reproduction capability to flight testing.

[0003] However, existing wind tunnel model deployment tests primarily utilize disposable models, and wind tunnel operating costs are high. As the number of test conditions increases, the expenses required for model deployment tests far exceed those for numerical computation. Furthermore, existing wind tunnel model deployment test data is mostly used for verification under numerical computation conditions, with little in-depth analysis and processing of the wind tunnel test data itself. Therefore, there is an urgent need for a wind tunnel model deployment test data processing and multibody separation modeling method to directly establish multibody separation prediction models based on wind tunnel model deployment test data, thereby meeting the cost reduction and efficiency improvement needs of modern aerospace engineering. Summary of the Invention

[0004] The problem addressed by this invention is to overcome the shortcomings of existing technologies and provide a multibody separation modeling method based on wind tunnel model deployment test data. This method involves performing polynomial fitting on the model trajectory and attitude data to obtain polynomial coefficients, and then establishing a regression model between the input parameters and the polynomial coefficients. By combining polynomial fitting and the regression model, the wind tunnel model deployment test data can be further analyzed, establishing a rapid prediction method for multibody separation based on this data.

[0005] The technical solution of this invention is:

[0006] A multibody separation modeling method based on wind tunnel model deployment test data includes:

[0007] Prepare and preprocess experimental data;

[0008] Perform polynomial fitting on the output time series data;

[0009] Perform fitting accuracy analysis;

[0010] Perform normalization processing on the input parameters and polynomial coefficients;

[0011] Perform correlation analysis on input and output parameters;

[0012] Establish a regression prediction model;

[0013] Perform model validation.

[0014] Furthermore, the test data includes: input and output parameters of the wind tunnel model deployment test; the input parameters include separation Mach number, separation angle of attack, separation velocity, and separation angular velocity; the output parameters include linear displacement and angular displacement time series, where linear displacement is the amount of translation and angular displacement is the amount of rotation.

[0015] Furthermore, the preprocessing includes processing outliers and missing values ​​in the input and output data, storing the data in a uniform format, and dividing it into a training set and a validation set. The training set is used to build a regression prediction model, and the validation set is used to validate the regression prediction model.

[0016] Furthermore, the step of performing polynomial fitting on the output time series data involves using the time term as the independent variable and the output parameter as the dependent variable. The functional relationship between the dependent and independent variables is described in the form of a polynomial of a certain order. A set of polynomial coefficients representing the output time series data is obtained through polynomial fitting. In subsequent modeling, the polynomial coefficients are used as output parameters to replace the original time series data.

[0017] Furthermore, the fitting accuracy analysis specifically involves: the fitting accuracy is expressed as the statistical index R0. 2 This is a characterization feature; the closer the value is to 1, the higher the fitting accuracy. When the accuracy is insufficient, the polynomial order needs to be adjusted.

[0018] R 2 =SSR / SST

[0019] In the formula: SSR is the residual sum of squares, which is the sum of squares of the differences between the actual value and the fitted value; SST is the total sum of squares, which is the sum of squares of the differences between the actual value and the mean.

[0020] Furthermore, normalization scaling involves uniformly scaling the range of input and output parameters to the range of 0-1, eliminating the influence of data dimensions.

[0021] Furthermore, normalization can be performed using max-min normalization, as shown in the following formula:

[0022]

[0023] Furthermore, the aforementioned input-output parameter correlation analysis specifically involves: correlation analysis assesses the relationship between input parameters and output parameters and the degree of their influence. The analysis results are represented by the correlation coefficient, and the larger the absolute value of the correlation coefficient, the stronger the correlation.

[0024] Furthermore, the establishment of the regression prediction model specifically includes a linear regression model and a nonlinear regression model, which are trained using training set data. First, a linear regression model is established as the baseline for the regression model. When the linear regression model is insufficient to characterize the input-output relationship, a nonlinear regression model is established. The input parameters in the training set data are the separation Mach number, separation angle of attack, separation velocity, and separation angular velocity, and the output parameters are the polynomial coefficients obtained by polynomial fitting of the linear displacement and angular displacement time series.

[0025] Furthermore, by substituting the validation set data input parameters generated during experimental data preparation and preprocessing into the regression prediction model, the polynomial coefficients representing the linear and angular displacement time series are output. These coefficients are then substituted into the polynomial function to restore the linear and angular displacement time series. The model is then compared with the linear and angular displacement time series restored from the polynomial coefficients of the validation set to complete the model validation.

[0026] The beneficial technical effects of this invention are as follows:

[0027] (1) The present invention proposes a multibody separation modeling method based on wind tunnel model deployment test data. This method performs polynomial fitting on the model trajectory and attitude data to obtain polynomial coefficients, and then establishes a regression model between the input parameters and the polynomial coefficients. By combining polynomial fitting and regression modeling, the wind tunnel model deployment test data can be further analyzed.

[0028] (2) The trajectory and attitude data of the wind tunnel deployment model test are time series data, which are difficult to establish a one-to-one mapping relationship with the input parameters, such as separation Mach number, separation angle of attack, separation speed, etc. By performing polynomial fitting on the trajectory and attitude time series data, it is easy to establish the regression relationship between the input parameters and the polynomial coefficients, thereby establishing a deployment test prediction model.

[0029] (3) By establishing a regression model, the uncertainty of wind tunnel model deployment test data can be quantitatively assessed, effectively improving the practicality of test data.

[0030] (4) By establishing a regression model, the influence of input parameters on output parameters can be quantitatively evaluated, which facilitates the optimization of the separation scheme. Attached Figure Description

[0031] Figure 1 This is a flowchart of the multibody separation modeling method of the present invention. Detailed Implementation

[0032] The present application will now be described in further detail with reference to the accompanying drawings and specific embodiments, so that those skilled in the art can implement it based on the description.

[0033] This invention discloses a multibody separation modeling method based on wind tunnel model deployment test data, such as... Figure 1 As shown, the method includes the following steps:

[0034] Step 1: Preparation and preprocessing of experimental data.

[0035] The experimental data includes input and output parameters of the wind tunnel model deployment test. Input parameters include separation Mach number, separation angle of attack, separation velocity, and separation angular velocity. Output parameters include linear displacement and angular displacement time series, where linear displacement is translation and angular displacement is rotation. The preprocessing includes handling outliers and missing values ​​in the input and output data, storing the data in a unified format, and dividing it into training and validation sets. The training set is used to build a regression model, and the validation set is used to validate the regression model.

[0036] Step 2: Output time series data for polynomial fitting.

[0037] The polynomial fitting of the output time series data uses the time term as the independent variable and the output parameter as the dependent variable. The functional relationship between the dependent and independent variables is described in the form of a polynomial of a certain order, such as a fourth-order polynomial, in the following format:

[0038] y = At 4 +Bt 3 +Ct 2 +Dt+E (1)

[0039] In the formula, y is the output parameter, t is time, and A, B, C, D, and E are the five coefficients of the fourth-order polynomial.

[0040] The polynomial fitting yields a set of polynomial coefficients that characterize the output time series data. These polynomial coefficients are then used as output parameters in subsequent modeling to replace the original time series data.

[0041] Step 3: Fitting accuracy analysis. The fitting accuracy is expressed as the statistical index R0. 2 The value of R represents the accuracy of the fit; the closer it is to 1, the higher the fitting accuracy. When the accuracy is insufficient, the polynomial order needs to be adjusted. 2 The calculation formula is:

[0042] R 2 =SSR / SST (2)

[0043] In the formula: SSR is the residual sum of squares, which is the sum of squares of the differences between the true value and the fitted value; SST is the total sum of squares, which is the sum of squares of the differences between the actual value and the mean.

[0044] Step 4: Input Parameter and Polynomial Coefficient Normalization. This normalization process scales the input and output parameter ranges to a uniform 0-1 range, eliminating the influence of data dimensions. Max-min normalization can be used, as shown in the following formula:

[0045]

[0046] Step 5: Input-output parameter correlation analysis. This correlation analysis assesses the relationship between input and output parameters and their degree of influence. The results are represented by correlation coefficients; the larger the absolute value of the correlation coefficient, the stronger the correlation, and the more it should be prioritized in the design.

[0047] Step Six: Establish a regression prediction model. The regression model includes a linear regression model and a nonlinear regression model. First, a linear regression model is established as the baseline. When the linear regression model is insufficient to represent the input-output relationship, a nonlinear regression model is established. Both the linear and nonlinear regression models are trained using training set data. The input parameters in the training set data are the separation Mach number, separation angle of attack, separation velocity, and separation angular velocity. The output parameters are the polynomial coefficients obtained by polynomial fitting of the linear displacement and angular displacement time series.

[0048] Step 7: Model Validation. The validation set data generated during experimental data preparation and preprocessing is substituted into the regression prediction model. The output polynomial coefficients representing the linear and angular displacement time series are then substituted into the polynomial function to reconstruct the linear and angular displacement time series. These are then compared with the linear and angular displacement time series reconstructed from the polynomial coefficients of the validation set to complete model validation.

[0049] The trajectory and attitude data of wind tunnel deployment tests are time-series data, making it difficult to establish a one-to-one mapping relationship with input parameters such as separation Mach number, separation angle of attack, and separation velocity. This invention, by performing polynomial fitting on the trajectory and attitude time-series data, can easily establish a regression relationship between input parameters and polynomial coefficients, thereby establishing a deployment test prediction model. By establishing a regression model, the uncertainty of wind tunnel model deployment test data can be quantitatively evaluated, effectively improving the practicality of the test data. Simultaneously, the influence of input parameters on output parameters can also be quantitatively evaluated, facilitating the optimization of separation schemes.

[0050] The parts of this invention not described in detail are common knowledge to those skilled in the art.

Claims

1. A multibody separation modeling method based on wind tunnel model deployment test data, characterized in that, include: Prepare and preprocess experimental data; Perform polynomial fitting on the output time series data; Perform fitting accuracy analysis; Perform normalization processing on the input parameters and polynomial coefficients; Perform correlation analysis on input and output parameters; Establish a regression prediction model; Perform model validation.

2. The multibody separation modeling method based on wind tunnel model deployment test data according to claim 1, characterized in that: The test data includes: input and output parameters of the wind tunnel model deployment test; the input parameters include separation Mach number, separation angle of attack, separation velocity, and separation angular velocity; the output parameters include linear displacement and angular displacement time series, where linear displacement is the amount of translation and angular displacement is the amount of rotation.

3. The multibody separation modeling method based on wind tunnel model deployment test data according to claim 2, characterized in that: The preprocessing includes handling outliers and missing values ​​in the input and output data, storing the data in a uniform format, and dividing it into a training set and a validation set. The training set is used to build a regression prediction model, and the validation set is used to validate the regression prediction model.

4. The multibody separation modeling method based on wind tunnel model deployment test data according to claim 1, characterized in that: The process of performing polynomial fitting on the output time series data involves using the time term as the independent variable and the output parameter as the dependent variable. The functional relationship between the dependent and independent variables is described in the form of a polynomial of a certain order. A set of polynomial coefficients representing the output time series data is obtained through polynomial fitting. In subsequent modeling, the polynomial coefficients are used as output parameters to replace the original time series data.

5. The multibody separation modeling method based on wind tunnel model deployment test data according to claim 4, characterized in that: The fitting accuracy analysis is specifically performed as follows: fitting accuracy is expressed as the statistical index R0. 2 This is a characterization feature; the closer the value is to 1, the higher the fitting accuracy. When the accuracy is insufficient, the polynomial order needs to be adjusted. R 2 =SSR / SST In the formula: SSR is the residual sum of squares, which is the sum of squares of the differences between the actual value and the fitted value; SST is the total sum of squares, which is the sum of squares of the differences between the actual value and the mean.

6. The multibody separation modeling method based on wind tunnel model deployment test data according to claim 1, characterized in that: Normalization scaling involves uniformly scaling the range of input and output parameters to the range of 0-1, eliminating the influence of data dimensions.

7. The multibody separation modeling method based on wind tunnel model deployment test data according to claim 6, characterized in that: Normalization can be achieved using max-min normalization, as shown in the following formula:

8. The multibody separation modeling method based on wind tunnel model deployment test data according to claim 1, characterized in that: The aforementioned input-output parameter correlation analysis specifically involves evaluating the relationship between input and output parameters and their degree of influence. The analysis results are represented by the correlation coefficient, with a larger absolute value of the correlation coefficient indicating a stronger correlation.

9. The multibody separation modeling method based on wind tunnel model deployment test data according to claim 1, characterized in that: The establishment of the regression prediction model specifically includes a linear regression model and a nonlinear regression model, which are trained using training set data. First, a linear regression model is established as the baseline of the regression model. When the linear regression model is insufficient to represent the input-output relationship, a nonlinear regression model is established. The input parameters in the training set data are the separation Mach number, separation angle of attack, separation velocity, and separation angular velocity, and the output parameters are the polynomial coefficients obtained by polynomial fitting of the linear displacement and angular displacement time series.

10. The multibody separation modeling method based on wind tunnel model deployment test data according to claim 9, characterized in that: By substituting the validation set data generated during experimental data preparation and preprocessing into the regression prediction model, the output polynomial coefficients representing the linear and angular displacement time series are used to restore the linear and angular displacement time series. The model is then compared with the linear and angular displacement time series restored by the polynomial coefficients of the validation set to complete the model validation.

Images