Hypersonic vehicle aerodynamic prediction method, system, electronic equipment and medium

Abstract

The present invention discloses a method, system, electronic device and medium for predicting aerodynamic force of a hypersonic aircraft. The method comprises: establishing an aerodynamic force prediction model based on random forest; selecting a plurality of flight operating points with different flight parameters within the flight envelope of the hypersonic aircraft; performing CFD calculation on the selected flight operating points to obtain aerodynamic force parameters of different flight operating points; generating a data set, the data set comprising a plurality of data points, each data point comprising a flight parameter of a flight operating point and a corresponding aerodynamic force parameter; using some data points in the data set as a training set, and using the training set to train an aerodynamic force prediction model, the trained aerodynamic force prediction model can output corresponding aerodynamic force parameter prediction values ​​according to the input real-time flight parameters, so as to realize rapid prediction of the aerodynamic performance of the aircraft according to the flight parameters.

Description

Technical Field

[0001] The present invention belongs to the field of hypersonic aircraft, and more specifically, relates to a method, system, electronic equipment and medium for predicting aerodynamic force of a hypersonic aircraft. Background Art

[0002] Aerodynamic modeling is the basis for aircraft flight performance evaluation and stability control, and it affects the success or failure of aircraft design, and is a basic design element. If you want to obtain the aerodynamic characteristics within the entire flight envelope, you need to conduct huge tests and calculations. On the one hand, due to the complex environment of hypersonic aircraft, it is impossible to measure all the aerodynamic data required for aircraft aerodynamic performance evaluation under flight conditions or in a wind tunnel environment. On the other hand, the accuracy of engineering algorithms is low and sometimes cannot meet design requirements. The existing CFD calculation accuracy can meet the requirements, but it takes a long time and requires a lot of computing resources. Summary of the invention

[0003] The purpose of the present invention is to provide a method, system, electronic equipment and medium for predicting the aerodynamic force of a hypersonic aircraft, so as to realize rapid prediction of the aerodynamic performance of the aircraft according to flight parameters.

[0004] In a first aspect, the present invention provides a method for predicting aerodynamic force of a hypersonic aircraft, comprising:

[0005] Establish an aerodynamic prediction model based on random forest;

[0006] Selecting a plurality of flight operating points with different flight parameters within the flight envelope of the hypersonic vehicle;

[0007] Perform CFD calculations on the selected flight operating points to obtain the aerodynamic parameters at different flight operating points;

[0008] Generate a data set, the data set comprising a plurality of data points, each of the data points comprising a flight parameter and a corresponding aerodynamic parameter of a flight operating point;

[0009] Some data points in the data set are used as a training set, and the aerodynamic prediction model is trained using the training set. The trained aerodynamic prediction model can output corresponding aerodynamic parameter prediction values ​​according to the input real-time flight parameters.

[0010] Optionally, the flight parameters include: altitude, Mach number, angle of attack and sideslip angle;

[0011] The aerodynamic parameters include: axial force, normal force, lateral force, pitch moment, yaw moment and roll moment.

[0012] Optionally, selecting a plurality of operating points with different flight parameters within the flight envelope of the hypersonic aircraft includes:

[0013] According to the flight envelope, a plurality of flight operating points covering the entire flight envelope are selected within the variation range of altitude, Mach number, angle of attack and sideslip angle.

[0014] Optionally, selecting a plurality of operating points with different flight parameters within the flight envelope of the hypersonic aircraft further includes:

[0015] According to the changing trend of aerodynamic force with angle of attack under different flight parameters, additional flight operating points that require CFD calculation are selected at locations where the changing trend is abrupt or drastic.

[0016] Optionally, the aerodynamic force prediction model is a combined classifier including multiple decision trees;

[0017] In the process of training the aerodynamic prediction model, mse is used as the evaluation standard of the feature, and the maximum number of features of the sample selected by a single decision tree is used as the maximum number of features when establishing the decision tree;

[0018] The minimum number of samples for subtree division is 2, and the minimum number of samples for leaf nodes is 1;

[0019] Bootstrap sampling is used when building decision trees.

[0020] Optionally, the operating function of the aerodynamic force prediction model is:

[0021]

[0022] Among them, K is the number of decision tree training times, H(x) represents the combined classification model, I is the indicative function, and h i (x) is the classification result of a single decision tree, and y represents the output target variable.

[0023] Optionally, it also includes:

[0024] The data points in the data set other than the training set are used as a test set, and the aerodynamic force prediction model that has completed training is tested using the test set.

[0025] In a second aspect, the present invention provides an electronic device, the electronic device comprising:

[0026] at least one processor; and,

[0027] a memory communicatively connected to the at least one processor; wherein,

[0028] The memory stores instructions that can be executed by the at least one processor, and the instructions are executed by the at least one processor so that the at least one processor can execute the hypersonic aircraft aerodynamic force prediction method described in the first aspect.

[0029] In a third aspect, the present invention proposes a non-transitory computer-readable storage medium, which stores computer instructions for causing a computer to execute the hypersonic aircraft aerodynamic prediction method described in the first aspect.

[0030] In a fourth aspect, the present invention provides a hypersonic vehicle aerodynamic force prediction system, comprising:

[0031] Model building module, used to build aerodynamic prediction model based on random forest;

[0032] A data selection module, used to select a plurality of flight operating points with different flight parameters within the flight envelope of the hypersonic aircraft;

[0033] CFD calculation module, used to perform CFD calculation on the selected flight operating point to obtain the aerodynamic parameters at different flight operating points;

[0034] A data set generation module, used to generate a data set, wherein the data set includes a plurality of data points, each of which includes a flight parameter and a corresponding aerodynamic parameter of a flight operating point;

[0035] The model training module uses some data points in the data set as a training set, and uses the training set to train the aerodynamic prediction model based on random forest. The trained aerodynamic prediction model can output corresponding aerodynamic parameter prediction values ​​according to the input real-time flight parameters.

[0036] The beneficial effects of the present invention are:

[0037] The present invention proposes a prediction method for the aerodynamic characteristics of a hypersonic aircraft. An aerodynamic prediction model based on a random forest is established, and flight operating points are selected within a flight envelope and CFD calculations are performed on the selected flight operating points to obtain aerodynamic parameters of different flight operating points, generate a data set, select some data points in the data set to train the aerodynamic prediction model, and use the trained aerodynamic prediction model to output corresponding aerodynamic parameters according to input flight parameters to achieve evaluation of the aerodynamic performance of the hypersonic aircraft, solve the problem of difficult balance between aerodynamic prediction accuracy and efficiency, and achieve rapid prediction of aerodynamic parameters according to flight parameters. Compared with existing engineering algorithms, the method has higher accuracy and higher calculation efficiency than existing CFD algorithms. At the same time, the method avoids complex physical mechanisms and mathematical and physical modeling based on mechanisms, and the aerodynamic model established based on random forests can approximate any nonlinear function.

[0038] The system of the present invention has other characteristics and advantages, which will be apparent from the drawings incorporated herein and the following detailed description, or will be described in detail in the drawings incorporated herein and the following detailed description, which together serve to explain the specific principles of the present invention. BRIEF DESCRIPTION OF THE DRAWINGS

[0039] The above and other objects, features and advantages of the present invention will become more apparent through a more detailed description of exemplary embodiments of the present invention in conjunction with the accompanying drawings, in which like reference numerals generally represent like components.

[0040] Figure 1 A step diagram of a method for predicting aerodynamic force of a hypersonic aircraft according to Embodiment 1 of the present invention is shown.

[0041] Figure 2 A schematic diagram of a hypersonic vehicle aerodynamic prediction method according to Embodiment 1 of the present invention is shown.

[0042] Figure 3 A comparison chart of the torque predicted by a random forest-based aerodynamic prediction model and the true value in a hypersonic aircraft aerodynamic prediction method according to Example 1 of the present invention is shown. DETAILED DESCRIPTION

[0043] Conventional aerodynamic force is a function of Mach number, angle of attack, sideslip angle and rudder angle. At small angles of attack, a linearized aerodynamic derivative model can be established to obtain aerodynamic characteristics at different angles of attack, Mach numbers and sideslip angles. However, the maneuverability of modern aircraft is increasing, the flow in the aircraft's external field is becoming more and more complex, and the angle of attack and sideslip angle vary widely. At this time, the linear derivative model is no longer applicable, and a nonlinear aerodynamic model needs to be established.

[0044] At present, there are two types of aerodynamic modeling methods for aircraft: one is based on mechanism modeling, and the other is "black box" modeling based on machine learning. Based on mechanism modeling, each variable has a physical meaning and is highly interpretable, but its ability to fit nonlinear data is usually weaker than that of machine learning modeling methods. The aerodynamic model established based on machine learning can not only reduce computing resource requirements and computing time, but also calculate flight operating points that cannot be covered by wind tunnel tests and CFD calculations.

[0045] Therefore, the present invention establishes an aerodynamic prediction model based on random forest, which can realize rapid prediction of aerodynamic forces and moments according to flight parameters, and is a powerful tool for aerodynamic evaluation in the conceptual design and preliminary design stages of aircraft.

[0046] The present invention will be described in more detail below with reference to the accompanying drawings. Although preferred embodiments of the present invention are shown in the accompanying drawings, it should be understood that the present invention can be implemented in various forms and should not be limited by the embodiments set forth herein. On the contrary, these embodiments are provided to make the present invention more thorough and complete, and to fully convey the scope of the present invention to those skilled in the art.

[0047] Example 1

[0048] Figure 1 A step diagram of a method for predicting aerodynamic force of a hypersonic aircraft according to Embodiment 1 of the present invention is shown. Figure 2 A schematic diagram of a hypersonic vehicle aerodynamic prediction method according to Embodiment 1 of the present invention is shown.

[0049] like Figure 1 and Figure 2 As shown, a method for predicting aerodynamic force of a hypersonic aircraft comprises:

[0050] Step S1: Establishing an aerodynamic prediction model based on random forest;

[0051] Specifically, random forest is a statistical learning theory, which is essentially a decision tree consisting of multiple decision trees {h(X,θ k ),k=1,2,…,K},{θ k} is a random vector that follows an independent and identical distribution and determines the form of the decision tree; K is the number of decision trees in the random forest. A relatively simple majority voting method is used, and the category with the most decision tree votes H(x) is taken as the category to which the final sample belongs.

[0052] The operating function of the aerodynamic prediction model is:

[0053]

[0054] Among them, K is the number of decision tree training times, H(x) represents the combined classification model, I is the indicative function, and h i (x) is the classification result of a single decision tree, and y represents the output target variable.

[0055] In this embodiment, it is preferred to use python to write the program, and the training and use of the random forest model are called from the sklearn additional package.

[0056] Step S2: selecting a plurality of flight operating points with different flight parameters within the flight envelope of the hypersonic aircraft; wherein the flight parameters include: altitude, Mach number, angle of attack and sideslip angle;

[0057] Preferably, according to the variation trend of aerodynamic force with angle of attack under different flight parameters, additional flight operating points requiring CFD calculation are selected at locations where the variation trend changes suddenly or drastically.

[0058] Step S3: Perform CFD calculation on the selected flight operating point to obtain aerodynamic parameters of different flight operating points (the calculation results of CFD can be used as true values); wherein the aerodynamic parameters include: axial force, normal force, lateral force, pitch moment, yaw moment and roll moment.

[0059] Step S4: Generate a data set, the data set includes multiple data points, each data point includes a flight parameter of a flight operating point and a corresponding aerodynamic parameter;

[0060] Specifically, first determine the range of flight parameters (altitude, Mach number, angle of attack and sideslip angle) within the flight envelope of the aircraft, select flight operating points for CFD calculation, and the operating points must cover the entire flight envelope. Observe the change trend of aerodynamic force with angle of attack under different flight parameters. For locations where the change trend changes suddenly or violently, additional calculation points are required at that location.

[0061] Then, CFD is used to calculate the axial force, normal force, lateral force, pitch moment, yaw moment and roll moment corresponding to the aircraft at each flight condition point, and a data set is generated, of which 95% of the data points are used as a training set for random forest model training, and 5% of the data points are used as a test set for testing the trained model. It should be noted that the training data set must cover all flight envelope data to ensure the prediction accuracy of the model.

[0062] Step S5: using some data points in the data set as a training set, and using the training set to train the aerodynamic prediction model, the trained aerodynamic prediction model can output corresponding aerodynamic parameters according to the input real-time flight parameters;

[0063] Specifically, the model input is flight parameters (altitude, Mach number, angle of attack and sideslip angle), and the output is axial force, normal force, lateral force, pitch moment, yaw moment and roll moment. After a series of training, the process of establishing the aerodynamic prediction model based on random forest is completed.

[0064] During the model training process: Based on the number of samples, the number of decision trees K in the random forest used is 200, the out-of-bag error is not used to evaluate the model, and the default mse (mean square error) is used as the evaluation standard for the feature. The maximum number of features of the sample selected by a single decision tree is used as the maximum number of features when establishing the decision tree. The maximum depth is not limited. The minimum number of samples for continued subtree division is 2, and the minimum number of samples for leaf nodes is 1. The minimum sample weight of the leaf node is not considered, and the maximum number of leaf nodes is not limited. Bootstrap sampling is used when building the tree (Bootstrap is a method of estimating the population through samples in statistics). In the process of verifying the model parameters, k-fold cross-validation with k=7 is used, and the resulting error does not change significantly.

[0065] Step S6: taking the data points outside the training set in the data set as the test set, and using the test set to test the trained aerodynamic force prediction model.

[0066] Specifically, the test set of the remaining 5% data points was used to test the trained random forest aerodynamic prediction model. The root mean square errors of the six-component aerodynamic forces and moments predicted by the aerodynamic prediction model were 40.91, 220.93, 125.61, 89.85, 25.08, and 36.84, respectively.

[0067] In a specific application example, taking an HTV2-like aircraft as an example, an aerodynamic prediction model based on random forest is established, in which Ma (Mach) and H (altitude) remain unchanged, and the angle of attack and sideslip angle change. CFD calculation is used to generate a data set, and the CFD calculation value is considered to be the true value.

[0068] 1) CFD++ is used to calculate aerodynamic forces. The calculation conditions are Ma=16, H=40km, the angle of attack AoA range is -5°-20°, a total of 11 points are calculated, the sideslip angle Beta range is -10°-10°, a total of 7 points are calculated, and a total of 77 data points. 95% of the data points are used for model training, and 5% of the data points are used to test the trained model.

[0069] 2) The input of the random forest aerodynamic prediction model is the angle of attack and the sideslip angle, and the output is the axial force, normal force, lateral force, pitch moment, yaw moment and roll moment. The yaw moment and pitch moment at the center of mass predicted by the random forest aerodynamic prediction model are compared with the true values. Figure 3 As shown, it can be seen that the error accuracy of the predicted aerodynamic force and moment is within 10%, and the prediction accuracy of the random forest model is higher.

[0070] It should be noted that the hypersonic vehicle aerodynamic force prediction method of the present invention is not limited to the prediction of conventional aerodynamic force, but can be extended to the prediction of unsteady aerodynamic force.

[0071] In summary, the prediction method of this embodiment has the following advantages compared with the prior art:

[0072] 1) It has higher accuracy than engineering algorithms and has a higher calculation effect than CFD algorithms based on NS equations and Euler equations, from hours and minutes to seconds respectively.

[0073] 2) It avoids complex physical mechanisms and mathematical physics modeling based on the mechanisms, and the aerodynamic model established based on random forests can approximate any nonlinear function.

[0074] Example 2

[0075] A hypersonic vehicle aerodynamic prediction system, including:

[0076] Model building module, used to build aerodynamic prediction model based on random forest;

[0077] A data selection module, used to select a plurality of flight operating points with different flight parameters within the flight envelope of the hypersonic aircraft;

[0078] CFD calculation module, used to perform CFD calculation on the selected flight operating point to obtain the aerodynamic parameters at different flight operating points;

[0079] A data set generation module, used to generate a data set, the data set includes a plurality of data points, each data point includes a flight parameter of a flight condition point and a corresponding aerodynamic parameter;

[0080] The model training module uses some data points in the data set as the training set, and uses the training set to train the aerodynamic prediction model based on random forest. The trained aerodynamic prediction model can output the corresponding aerodynamic parameter prediction values ​​according to the input real-time flight parameters.

[0081] Example 3

[0082] This embodiment provides an electronic device, the electronic device comprising:

[0083] at least one processor; and,

[0084] a memory communicatively connected to at least one processor; wherein,

[0085] The memory stores instructions that can be executed by at least one processor, and the instructions are executed by at least one processor so that the at least one processor can execute the hypersonic aircraft aerodynamic force prediction method of embodiment 1.

[0086] The electronic device according to an embodiment of the present disclosure includes a memory and a processor, and the memory is used to store non-temporary computer-readable instructions. Specifically, the memory may include one or more computer program products, and the computer program product may include various forms of computer-readable storage media, such as volatile memory and / or non-volatile memory. The volatile memory may, for example, include a random access memory (RAM) and / or a cache memory (cache), etc. The non-volatile memory may, for example, include a read-only memory (ROM), a hard disk, a flash memory, etc.

[0087] The processor may be a central processing unit (CPU) or other forms of processing units having data processing capabilities and / or instruction execution capabilities, and may control other components in the electronic device to perform desired functions. In one embodiment of the present disclosure, the processor is used to run the computer-readable instructions stored in the memory.

[0088] Those skilled in the art should be able to understand that in order to solve the technical problem of how to obtain a good user experience, the present embodiment may also include well-known structures such as a communication bus and an interface, and these well-known structures should also be included in the protection scope of the present disclosure.

[0089] For detailed description of this embodiment, reference may be made to the corresponding descriptions in the aforementioned embodiments, which will not be repeated here.

[0090] Example 4

[0091] This embodiment provides a non-transitory computer-readable storage medium, which stores computer instructions, and the computer instructions are used to enable a computer to execute the hypersonic aircraft aerodynamic force prediction method of embodiment 1.

[0092] The electronic device according to an embodiment of the present disclosure includes a memory and a processor, and the memory is used to store non-temporary computer-readable instructions. Specifically, the memory may include one or more computer program products, and the computer program product may include various forms of computer-readable storage media, such as volatile memory and / or non-volatile memory. The volatile memory may, for example, include a random access memory (RAM) and / or a cache memory (cache), etc. The non-volatile memory may, for example, include a read-only memory (ROM), a hard disk, a flash memory, etc.

[0093] The processor may be a central processing unit (CPU) or other forms of processing units having data processing capabilities and / or instruction execution capabilities, and may control other components in the electronic device to perform desired functions. In one embodiment of the present disclosure, the processor is used to run the computer-readable instructions stored in the memory.

[0094] Those skilled in the art should be able to understand that in order to solve the technical problem of how to obtain a good user experience, the present embodiment may also include well-known structures such as a communication bus and an interface, and these well-known structures should also be included in the protection scope of the present disclosure.

[0095] For detailed description of this embodiment, reference may be made to the corresponding descriptions in the aforementioned embodiments, which will not be repeated here.

[0096] The embodiments of the present invention have been described above, and the above description is exemplary, not exhaustive, and is not limited to the disclosed embodiments. Many modifications and changes will be apparent to those skilled in the art without departing from the scope and spirit of the described embodiments.

Claims

1. A method for predicting aerodynamic force of a hypersonic vehicle, characterized in that: include: Establish an aerodynamic prediction model based on random forest; Selecting a plurality of flight operating points with different flight parameters within a flight envelope of a hypersonic aircraft, wherein the flight parameters include: altitude, Mach number, angle of attack, and sideslip angle. Selecting a plurality of flight operating points with different flight parameters within the flight envelope of the hypersonic aircraft includes: selecting a plurality of flight operating points covering the entire flight envelope within a variation range of altitude, Mach number, angle of attack, and sideslip angle according to the flight envelope; Perform CFD calculations on the selected flight operating points to obtain the aerodynamic parameters at different flight operating points; Generate a data set, the data set includes a plurality of data points, each of the data points includes a flight parameter of a flight operating point and a corresponding aerodynamic force parameter, the aerodynamic force parameter includes: axial force, normal force, lateral force, pitch moment, yaw moment and roll moment; Some data points in the data set are used as a training set, and the aerodynamic prediction model is trained using the training set. The trained aerodynamic prediction model can output corresponding aerodynamic parameter prediction values ​​according to the input real-time flight parameters.

2. The method for predicting aerodynamic force of a hypersonic vehicle according to claim 1, characterized in that: The step of selecting a plurality of operating points with different flight parameters within the flight envelope of the hypersonic aircraft further includes: According to the changing trend of aerodynamic force with angle of attack under different flight parameters, additional flight operating points that require CFD calculation are selected at locations where the changing trend is abrupt or drastic.

3. The method for predicting aerodynamic force of a hypersonic vehicle according to claim 1, characterized in that: The aerodynamic force prediction model is a combined classifier including multiple decision trees; In the process of training the aerodynamic prediction model, mse is used as the evaluation standard of the feature, and the maximum number of features of the sample selected by a single decision tree is used as the maximum number of features when establishing the decision tree; The minimum number of samples for subtree division is 2, and the minimum number of samples for leaf nodes is 1; Bootstrap sampling is used when building decision trees.

4. The method for predicting aerodynamic force of a hypersonic vehicle according to claim 1 or 3, characterized in that: The operating function of the aerodynamic force prediction model is: Among them, K is the number of decision tree training times, H(x) represents the combined classification model, I is the indicative function, and h i (x) is the classification result of a single decision tree, and y represents the output target variable.

5. The method for predicting aerodynamic force of a hypersonic vehicle according to claim 1, characterized in that: Also includes: The data points in the data set other than the training set are used as a test set, and the aerodynamic force prediction model that has completed training is tested using the test set.

6. An electronic device, characterized in that: The electronic device comprises: at least one processor; and, a memory communicatively connected to the at least one processor; wherein, The memory stores instructions that can be executed by the at least one processor, and the instructions are executed by the at least one processor so that the at least one processor can execute the hypersonic aircraft aerodynamic force prediction method described in any one of claims 1-5.

7. A non-transitory computer-readable storage medium, characterized in that: The non-transitory computer-readable storage medium stores computer instructions, which are used to enable a computer to execute the hypersonic aircraft aerodynamic prediction method described in any one of claims 1-5.

8. A hypersonic vehicle aerodynamic prediction system, characterized in that: include: Model building module, used to build aerodynamic prediction model based on random forest; A data selection module, for selecting a plurality of flight operating points of different flight parameters within a flight envelope of a hypersonic aircraft, wherein the flight parameters include: altitude, Mach number, angle of attack, and sideslip angle, wherein the step of selecting a plurality of operating points of different flight parameters within the flight envelope of the hypersonic aircraft includes: selecting a plurality of flight operating points covering the entire flight envelope within a variation range of the altitude, Mach number, angle of attack, and sideslip angle according to the flight envelope; A CFD calculation module is used to perform CFD calculations on the selected flight operating points to obtain aerodynamic parameters at different flight operating points, wherein the aerodynamic parameters include: axial force, normal force, lateral force, pitch moment, yaw moment and roll moment; A data set generation module, used to generate a data set, wherein the data set includes a plurality of data points, each of which includes a flight parameter and a corresponding aerodynamic parameter of a flight operating point; The model training module uses some data points in the data set as a training set, and uses the training set to train the aerodynamic prediction model based on random forest. The trained aerodynamic prediction model can output corresponding aerodynamic parameter prediction values ​​according to the input real-time flight parameters.

Images