CN116628854A - Wing section aerodynamic characteristic prediction method, system, electronic equipment and storage medium - Google Patents
Wing section aerodynamic characteristic prediction method, system, electronic equipment and storage medium Download PDFInfo
- Publication number
- CN116628854A CN116628854A CN202310606863.2A CN202310606863A CN116628854A CN 116628854 A CN116628854 A CN 116628854A CN 202310606863 A CN202310606863 A CN 202310606863A CN 116628854 A CN116628854 A CN 116628854A
- Authority
- CN
- China
- Prior art keywords
- fidelity
- aerodynamic
- task
- model
- expert
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
- 238000000034 method Methods 0.000 title claims abstract description 57
- 230000007246 mechanism Effects 0.000 claims abstract description 8
- 238000012549 training Methods 0.000 claims description 26
- 238000004590 computer program Methods 0.000 claims description 9
- 238000005457 optimization Methods 0.000 abstract description 18
- 238000013461 design Methods 0.000 abstract description 12
- 230000000694 effects Effects 0.000 description 10
- 238000004088 simulation Methods 0.000 description 8
- 238000005516 engineering process Methods 0.000 description 6
- 238000004364 calculation method Methods 0.000 description 5
- 238000010586 diagram Methods 0.000 description 5
- 230000008569 process Effects 0.000 description 4
- 238000006243 chemical reaction Methods 0.000 description 3
- 238000012360 testing method Methods 0.000 description 3
- 230000003213 activating effect Effects 0.000 description 2
- 230000006870 function Effects 0.000 description 2
- 238000012795 verification Methods 0.000 description 2
- 230000004913 activation Effects 0.000 description 1
- 238000004458 analytical method Methods 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 238000004422 calculation algorithm Methods 0.000 description 1
- 230000008878 coupling Effects 0.000 description 1
- 238000010168 coupling process Methods 0.000 description 1
- 238000005859 coupling reaction Methods 0.000 description 1
- 230000001419 dependent effect Effects 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 238000002474 experimental method Methods 0.000 description 1
- 239000012530 fluid Substances 0.000 description 1
- 230000004927 fusion Effects 0.000 description 1
- 210000002569 neuron Anatomy 0.000 description 1
- 230000000750 progressive effect Effects 0.000 description 1
- 238000011160 research Methods 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F30/00—Computer-aided design [CAD]
- G06F30/10—Geometric CAD
- G06F30/15—Vehicle, aircraft or watercraft design
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F30/00—Computer-aided design [CAD]
- G06F30/20—Design optimisation, verification or simulation
- G06F30/28—Design optimisation, verification or simulation using fluid dynamics, e.g. using Navier-Stokes equations or computational fluid dynamics [CFD]
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02T—CLIMATE CHANGE MITIGATION TECHNOLOGIES RELATED TO TRANSPORTATION
- Y02T90/00—Enabling technologies or technologies with a potential or indirect contribution to GHG emissions mitigation
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Geometry (AREA)
- Mathematical Optimization (AREA)
- General Engineering & Computer Science (AREA)
- Mathematical Analysis (AREA)
- Evolutionary Computation (AREA)
- Pure & Applied Mathematics (AREA)
- Computer Hardware Design (AREA)
- Aviation & Aerospace Engineering (AREA)
- Computational Mathematics (AREA)
- Automation & Control Theory (AREA)
- Algebra (AREA)
- Computing Systems (AREA)
- Fluid Mechanics (AREA)
- Mathematical Physics (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The invention discloses a method, a system, electronic equipment and a storage medium for predicting aerofoil aerodynamic characteristics, and relates to the technical field of aerodynamic profile optimization design. The method comprises the following steps: acquiring parameterized variables and flight conditions of the airfoil profile of the target aircraft; utilizing a multi-task multi-fidelity model to predict aerodynamic characteristics of parameterized variables and flight conditions, and determining high-fidelity aerodynamic characteristic data of the airfoil profile of the target aircraft; the multi-task multi-fidelity model comprises a first multi-task expert mixed model and a second multi-task expert mixed model; the first multi-task expert mixed model and the second multi-task expert mixed model are both constructed according to a multi-task learning method and a gating mechanism; the first multitasking expert hybrid model is used to determine low fidelity aerodynamic feature data for the airfoil profile of the target aircraft; the second multitasking expert hybrid model is used to convert low-fidelity aerodynamic data into high-fidelity aerodynamic data. The invention can improve the accuracy and speed of data acquisition.
Description
Technical Field
The invention relates to the technical field of aerodynamic profile optimization design, in particular to an airfoil aerodynamic characteristic prediction method, an airfoil aerodynamic characteristic prediction system, electronic equipment and a storage medium.
Background
The aerodynamic profile optimization design is to research how to combine the modern CFD technology with a numerical optimization algorithm, and automatically search the discipline direction of the optimal profile meeting the performance requirement through a computer. The traditional trial-and-error method is to rely on theoretical deduction and experimental verification, and the aerodynamic shape optimization is carried out by a method for continuously adjusting the shape to judge whether the optimization target is met, wherein the method is seriously dependent on experience of a designer, the design period is very long, and the optimal shape under multiple constraints and multiple design states is often not obtained under normal conditions.
With the continuous development of aircraft technology and computer technology, CFD technology has greatly advanced in theory of computational accuracy and grid generation, and high-fidelity simulation based on Computational Fluid Dynamics (CFD) has gradually become a main method for evaluating aerodynamic performance, however, aerodynamic profile optimization based on CFD technology still takes a long time, and one analysis may take hours or even days, and thousands of times of CFD calls may cause "computational disasters". To improve the optimization efficiency and reduce the design cycle, many practitioners use xf oil software instead of CFD simulation to perform pneumatic optimization, which can be done once in a few seconds. The optimization process is quickened, but as the optimization design requirement of the fine aerodynamic shape of the aircraft is higher and higher, the precision of XFOIL calculation often cannot meet the engineering design requirement.
Disclosure of Invention
The invention aims to provide an airfoil aerodynamic characteristic prediction method, an airfoil aerodynamic characteristic prediction system, electronic equipment and a storage medium, which can convert low-precision data into high-precision data and improve data acquisition precision and speed.
In order to achieve the above object, the present invention provides the following solutions:
a method of airfoil aerodynamic property prediction comprising:
acquiring parameterized variables and flight conditions of the airfoil profile of the target aircraft;
utilizing a multi-task multi-fidelity model to predict aerodynamic characteristics of the parameterized variable and the flight condition, and determining high-fidelity aerodynamic characteristic data of the airfoil profile of the target aircraft; the multi-task multi-fidelity model comprises a first multi-task expert mixed model and a second multi-task expert mixed model; the first multi-task expert mixed model and the second multi-task expert mixed model are both constructed according to a multi-task learning method and a gating mechanism; the first multitasking expert hybrid model is used to determine low fidelity aerodynamic feature data for the target aircraft airfoil profile; the second multitasking expert hybrid model is used to convert the low-fidelity aerodynamic data into high-fidelity aerodynamic data.
Optionally, the flight condition is a working condition of the aircraft in the flight process; the operating conditions include angle of attack, mach number and Reynolds number.
Optionally, the method for determining the parameterized variable of the airfoil profile of the target aircraft comprises the following steps: potential variables of the target aircraft airfoil profile that are generated against the network are determined as parameterized variables.
Optionally, the determining, by using a multi-task multi-fidelity model, high-fidelity aerodynamic characteristic data of the target aircraft airfoil profile by predicting aerodynamic characteristics of the parameterized variable and the flight condition specifically includes:
inputting the parameterized variables and the flight conditions into the first multitasking expert hybrid model to determine low-fidelity aerodynamic data for the target aircraft airfoil profile;
and inputting the low-fidelity aerodynamic feature data into the second multitasking expert hybrid model, converting the low-fidelity aerodynamic feature data, and determining the high-fidelity aerodynamic feature data of the airfoil profile of the target aircraft.
Optionally, the determining method of the multi-task multi-fidelity model is as follows:
acquiring training data; the training data comprises training parameter variables, flight training conditions and corresponding high-fidelity aerodynamic characteristic prediction results;
and inputting the training data into the first multi-task expert mixed model, taking the low-fidelity pneumatic characteristic data output by the first multi-task expert mixed model as the input of the second multi-task expert mixed model, and training by using a back propagation method and taking the minimum loss between the high-fidelity pneumatic characteristic data output by the second multi-task expert mixed model and the high-fidelity pneumatic characteristic prediction result as a target to obtain the multi-task multi-fidelity model.
The invention also provides an airfoil aerodynamic characteristic prediction system, comprising:
the data acquisition module is used for acquiring parameterized variables and flight conditions of the airfoil profile of the target aircraft;
the characteristic prediction module is used for predicting aerodynamic characteristics of the parameterized variable and the flight condition by utilizing a multi-task multi-fidelity model and determining high-fidelity aerodynamic characteristic data of the airfoil profile of the target aircraft; the multi-task multi-fidelity model comprises a first multi-task expert mixed model and a second multi-task expert mixed model; the first multi-task expert mixed model and the second multi-task expert mixed model are both constructed according to a multi-task learning method and a gating mechanism; the first multitasking expert hybrid model is used to determine low fidelity aerodynamic feature data for the target aircraft airfoil profile; the second multitasking expert hybrid model is used to convert the low-fidelity aerodynamic data into high-fidelity aerodynamic data.
The invention also provides electronic equipment, which comprises a memory and a processor, wherein the memory is used for storing a computer program, and the processor runs the computer program to enable the electronic equipment to execute the airfoil aerodynamic characteristic prediction method.
The present invention also provides a computer readable storage medium storing a computer program which when executed by a processor implements the aerofoil aerodynamic characteristic prediction method as described above.
According to the specific embodiment provided by the invention, the invention discloses the following technical effects:
the invention discloses an airfoil aerodynamic characteristic prediction method, a system, electronic equipment and a storage medium, wherein the method comprises the steps of predicting the aerodynamic characteristic of an obtained parameterized variable and a flight condition of an airfoil profile of a target aircraft by utilizing a multi-task multi-fidelity model, and determining high-fidelity aerodynamic characteristic data of the airfoil profile of the target aircraft, wherein the multi-task multi-fidelity model comprises a first multi-task expert mixed model and a second multi-task expert mixed model, the first multi-task expert mixed model is used for determining low-fidelity aerodynamic characteristic data of the airfoil profile of the target aircraft, the second multi-task expert mixed model is used for converting the low-fidelity aerodynamic characteristic data into high-fidelity aerodynamic characteristic data, and the two mixed models can be converted from the low-precision data to the high-precision data, so that the data acquisition precision and speed are improved.
Drawings
In order to more clearly illustrate the embodiments of the present invention or the technical solutions of the prior art, the drawings that are needed in the embodiments will be briefly described below, it being obvious that the drawings in the following description are only some embodiments of the present invention, and that other drawings may be obtained according to these drawings without inventive effort for a person skilled in the art.
FIG. 1 is a flow chart of a method of aerodynamics prediction of the present invention;
FIG. 2 is a schematic diagram of the result of XFOIL solver solution in this embodiment;
FIG. 3 is a diagram showing the effects of mesh division required to obtain high-fidelity simulation in the present embodiment;
FIG. 4 is a graph showing the comparison of XFOIL calculation and high fidelity simulation with the pressure distribution of a wind tunnel test under the same working condition in the present embodiment;
FIG. 5 is a schematic diagram of the MMOE network structure of the present embodiment;
FIG. 6 is a schematic diagram of the training process of MMOE-MF in the present embodiment;
FIG. 7 is a graph showing the predicted effects of MMOE-MF in the sub-sonic conditions for predicting lift, drag, and pitching moment in this embodiment; wherein (a) is C l Predicting an effect graph; (b) Is C d Predicting an effect graph; (c) Is C m Predicting an effect graph;
FIG. 8 is the presentIn the embodiment, the MMOE-MF predicts a predicted effect diagram of lift force, resistance and pitching moment under a transonic working condition; wherein (a) is C l Predicting an effect graph; (b) Is C d Predicting an effect graph; (c) Is C m And predicting an effect graph.
Detailed Description
The following description of the embodiments of the present invention will be made clearly and completely with reference to the accompanying drawings, in which it is apparent that the embodiments described are only some embodiments of the present invention, but not all embodiments. All other embodiments, which can be made by those skilled in the art based on the embodiments of the invention without making any inventive effort, are intended to be within the scope of the invention.
The invention aims to provide an airfoil aerodynamic characteristic prediction method, an airfoil aerodynamic characteristic prediction system, electronic equipment and a storage medium, which can convert low-precision data into high-precision data and improve data acquisition precision and speed.
In order that the above-recited objects, features and advantages of the present invention will become more readily apparent, a more particular description of the invention will be rendered by reference to the appended drawings and appended detailed description.
As shown in fig. 1, the present invention provides a method for predicting aerodynamic characteristics of an airfoil, comprising:
step 100: and obtaining parameterized variables and flight conditions of the airfoil profile of the target aircraft. The method for determining the parameterized variable of the airfoil profile of the target aircraft comprises the following steps: potential variables of the target aircraft airfoil profile that are generated against the network are determined as parameterized variables. The flight condition is the working condition of the aircraft in the flight process; the operating conditions include angle of attack, mach number and Reynolds number.
Step 200: utilizing a multi-task multi-fidelity model to predict aerodynamic characteristics of the parameterized variable and the flight condition, and determining high-fidelity aerodynamic characteristic data of the airfoil profile of the target aircraft; the multi-task multi-fidelity model comprises a first multi-task expert mixed model and a second multi-task expert mixed model; the first multi-task expert mixed model and the second multi-task expert mixed model are both constructed according to a multi-task learning method and a gating mechanism; the first multitasking expert hybrid model is used to determine low fidelity aerodynamic feature data for the target aircraft airfoil profile; the second multitasking expert hybrid model is used to convert the low-fidelity aerodynamic data into high-fidelity aerodynamic data.
As a specific embodiment of step 200, it includes:
inputting the parameterized variables and the flight conditions into the first multitasking expert hybrid model to determine low-fidelity aerodynamic data for the target aircraft airfoil profile; and inputting the low-fidelity aerodynamic feature data into the second multitasking expert hybrid model, converting the low-fidelity aerodynamic feature data, and determining the high-fidelity aerodynamic feature data of the airfoil profile of the target aircraft.
The method for determining the multi-task multi-fidelity model comprises the following steps:
acquiring training data; the training data comprises training parameter variables, flight training conditions and corresponding high-fidelity aerodynamic characteristic prediction results; and inputting the training data into the first multi-task expert mixed model, taking the low-fidelity pneumatic characteristic data output by the first multi-task expert mixed model as the input of the second multi-task expert mixed model, and training by using a back propagation method and taking the minimum loss between the high-fidelity pneumatic characteristic data output by the second multi-task expert mixed model and the high-fidelity pneumatic characteristic prediction result as a target to obtain the multi-task multi-fidelity model.
On the basis of the technical scheme, the following embodiment is also provided.
In the embodiment, the method for converting the aerofoil aerodynamic characteristics into high precision with low precision mainly comprises an aerofoil profile parameterized variable, a flight condition, a low-fidelity aerodynamic database and a high-fidelity aerodynamic database, and an MMOE-MF multi-fidelity model is built, and a result output layer is formed.
Wherein, airfoil profile parameterized variable: in aerodynamic profile optimization design, the parameterization of the airfoil profile is the most important content of the aerodynamic profile optimization design, determining the optimized quality and the optimized efficiency. The invention uses advanced intelligent parameterization technology of the generation countermeasure network, and potential variables of the generation countermeasure network are used as parameterized variables.
Flight conditions: the aircraft conditions during flight, such as angle of attack, mach number, reynolds number, etc.
The low-fidelity pneumatic database is used for obtaining the airfoil profile and inputting the airfoil profile into the XFOIL solver, and inputting corresponding flight conditions, so that the low-fidelity pneumatic characteristics including lift force, resistance, pitching moment and the like can be obtained. The aerofoil shape parameterized variable is used for fusing the flight condition and the low-fidelity aerodynamic characteristic is added to form the low-fidelity aerodynamic database.
High fidelity pneumatic database: after the airfoil profile is obtained, high-fidelity aerodynamic characteristic data are obtained through high-fidelity simulation, the profile is firstly input into Pointwise software for grid division, and then the generated grid is input into Fluent software for high-fidelity simulation to obtain the high-fidelity aerodynamic characteristic data. Similarly, the airfoil profile parameterized variable fuses flight conditions, plus high-fidelity aerodynamic characteristics, to form a high-fidelity aerodynamic database.
Constructing a multi-task multi-fidelity model (MMOE-MF): the method comprises the steps of coupling two multi-tasking expert mixed models to establish a multi-fidelity proxy model to predict multi-fidelity aerodynamic characteristics of an airfoil, wherein a first multi-tasking expert mixed model is used for predicting low-precision aerodynamic data, then fusing the characteristics of high-fidelity aerodynamic data to be input into a second multi-tasking expert mixed model, and once the model is trained, the second multi-tasking expert mixed model is used for excavating the relation between the low-precision and the high-precision, and once the model is trained, high-fidelity aerodynamic data can be obtained in a few seconds only by fusing flight conditions through airfoil profile parameterized variables.
Output of results: for each profile, its high fidelity aerodynamic characteristics are predicted.
Based on the components, the specific application steps are as follows:
step one: inputting the pneumatic shape into an XFOIL solver, and obtaining a large amount of cheap low-fidelity pneumatic data by setting flight conditions;
step two: the high-fidelity pneumatic characteristics of a small quantity and high cost are obtained by carrying out grid division and high-fidelity simulation on the appearance;
step three: training a first MMOE model using the low-fidelity dataset to predict low-fidelity aerodynamic characteristics;
step four: training a second MMOE model to predict high-fidelity aerodynamic feature data using the feature fusion low-fidelity aerodynamic feature of the high-fidelity dataset;
step five: once the multi-fidelity proxy model training is completed, we can input the high-fidelity aerodynamic properties obtained by the MMOE-MF model into the high-fidelity database.
Wherein the MMOE-MF model is constructed according to the following formula:
y k ＝h k (f k (x))
g k (x)＝softmax(W·x)
where k is the number of tasks, E i (x) For the ith expert network,activating input features for the ith expert kth gating network using softmax, f k (x) Weighted sum for all experts, y k And outputting a result for each task. And x is input data, and wing profile parameterized variables and flight conditions are fused. For different tasks, the model selection is different, the output of a specific Gate K represents the probability of being selected by different experts, and a plurality of experts are weighted and summed to obtain f k (x) And output to a specific Tower model, and finally output the result.
The low-precision conversion of the airfoil aerodynamic data into high-precision data is realized through the MMOE-MF model. And, after constructing the MMOE-MF model, the method further comprises: parameters of the MMOE-MF model are updated optimally by back-propagation.
Optimizing and updating parameters of the MMOE-MF model through the following formula:
wherein n is the number of training samples, y pred For model predictive value, y true Is the true value of the training sample.
Therefore, outputting high-precision data requires the following steps:
step one: inputting aerodynamic profile parameterized variables and flight conditions into an MMOE-MF model, and predicting low-fidelity aerodynamic characteristics by using a first MMOE model;
step two: merging high-fidelity characteristics and flight conditions and merging low-fidelity aerodynamic characteristics to obtain a merging layer;
step three: the merge layer is input into the MMOE-MF model and a second MMOE model is used to obtain high-precision aerodynamic characteristics.
As another example, in aerodynamic profile optimization design, acquiring aerodynamic characteristic data is a very important step. Fig. 2 shows the aerodynamic characteristics calculated using a low-fidelity solver, fig. 3 shows the mixed grid for accelerating convergence, and in order to verify the independence of the grids, the grid quantities of 14577 (M0), 43741 (M1), 83871 (M2) are selected for verification in this embodiment as shown in table 1, and the results are shown in table one, wherein mach numbers are represented, reynolds numbers are represented, and attack angles are represented.
Table 1 subsonic operating conditions, m=0.45, re=6.0×10 6 ，AOA＝1.5901
| Grid quantity | C l (10 -3 ) | C d (10 -4 ) | C m (10 -4 ) | |
| M0 | 14539 | 820.02 | 193.14 | -979.11 |
| M1 | 34359 | 812.84 | 186.83 | -956.48 |
| M2 | 83907 | 826.09 | 191.30 | -984.43 |
In order to verify the accuracy of the calculation in this embodiment, the calculation result is compared with the result of the wind tunnel test (as shown in fig. 4), and the specific comparison result is shown in table 2, where the result of the aerodynamic coefficient of the wind tunnel test is as follows: c (C) l ＝0.802，C d ＝0.0175，C m ＝-0.1。
Table 2 comparison of high and low fidelity aerodynamic coefficients and computation time for rae2822 airfoils
| Solver | C l | ΔC l | C d | ΔC d | C m | ΔC m | CPU time |
| XFOIL | 0.7381 | 0.064 | 0.0083 | 0.0092 | -0.0740 | 0.026 | 0.26s |
| CFD | 0.8200 | 0.018 | 0.0193 | 0.0018 | -0.0979 | 0.002 | 112s |
In consideration of a plurality of aerodynamic characteristics such as lift force, resistance force, pitching moment and the like to be predicted, the embodiment models based on a multi-task learning method, divides a sharing layer into a plurality of experts, introduces a gating mechanism, establishes independent gating weight for each task, and then performs weighted summation on contribution degrees of the experts. However, traditional multi-task learning requires a strong correlation between tasks to obtain a good prediction accuracy.
The method and the device better solve the limitation of strong task correlation, and improve the task prediction precision by fully considering the correlation among the tasks. In the embodiment, on the basis of a multi-gate expert hybrid model (MMOE), two MMOE models (shown in fig. 5) are used for constructing a multi-task multi-fidelity proxy model (MMOE-MF), and a prediction method for converting low-precision aerodynamic characteristics of an aircraft into high-precision aerodynamic characteristics is provided. In the MMOE-MF model, the input is a merge layer, including airfoil profile parameterized variables and flight conditions. The output is three predictive tasks including lift, drag, pitch moment.
As shown in FIG. 4, the MMOE-MF model was constructed according to the following formula.
y k ＝h k (f k (x))
g k (x)＝softmax(W·x)
Where k is the number of tasks, E i (x) For the ith expert network,activating the input feature for the ith expert, kth gating network, using soffmax, f k (x) Weighted sum for all experts, y k And outputting a result for each task. And x is input data, and wing profile parameterized variables and flight conditions are fused. For different tasks, the model selection is different, the output of a specific Gate K represents the probability of being selected by different experts, and a plurality of experts are weighted and summed to obtain f k (x) And output to a specific Tower model, and finally output the result.
Finally, the low-precision conversion of the airfoil aerodynamic data into high-precision data is realized through the MMOE-MF model.
As shown in fig. 6, in some embodiments, parameters of the updated model are optimized by back-propagation and the model is trained, wherein the loss function is as shown in the following equation:
wherein n is the number of training samples, y pred For model predictive value, y true Is the true value of the training sample. Exemplary, specific parameter settings of the MMOE model are shown in Table 3
TABLE 3 MMOE-MF model specific parameter settings
| Network layer | Neurons | Quantity of | Activation function |
| Input | 25 | 1 | - |
| Experts | 25×16 | 10 | Relu |
| Tasks | 25×10 | 3 | Softmax |
| Towers | 512×16 | 3 | Relu |
| Output | 1×3 | 3 | Linear |
As shown in fig. 7-8, in some implementations, separate models were constructed for subsonic and transonic, respectively, and the model of MMOE-MF obtained the predicted results by:
step one: inputting aerodynamic profile parameterized variables and flight conditions into an MMOE-MF model, and predicting low-fidelity aerodynamic characteristics by using a first MMOE model;
step two: merging high-fidelity characteristics and flight conditions and merging low-fidelity aerodynamic characteristics to obtain a merging layer;
step three: the merge layer is input into the MMOE-MF model and a second MMOE model is used to obtain high-precision aerodynamic characteristics.
The effectiveness and progress of the method provided by the embodiments of the present invention are verified by comparing the method of the present embodiment with prior art multitasking and multi-fidelity experiments according to the following examples.
Table 4 model comparison
The embodiment has the following beneficial effects:
according to the aerodynamic profile optimization method, the aerodynamic profile optimization device and the aerodynamic profile optimization medium for the aircraft based on the multi-fidelity simulation, which are disclosed by the invention, the aerodynamic characteristics of the wing profile are predicted by constructing two multi-tasking expert mixed models, and the first multi-tasking expert mixed model is used for predicting low-precision aerodynamic data, so that the relation between the low-precision aerodynamic data is fully excavated. The second multi-task expert hybrid model is used to mine the relationship between low accuracy and high accuracy, and the two predictive models combine to achieve a conversion of low-fidelity aerodynamic properties to high-fidelity aerodynamic properties in the Mach number range [0.3-0.74] within a few seconds. The time cost is greatly saved, the calculation resource is saved, and a more efficient method is provided for the pneumatic appearance optimization design.
In addition, the invention also provides an airfoil aerodynamic characteristic prediction system, which comprises:
the data acquisition module is used for acquiring parameterized variables and flight conditions of the airfoil profile of the target aircraft;
the characteristic prediction module is used for predicting aerodynamic characteristics of the parameterized variable and the flight condition by utilizing a multi-task multi-fidelity model and determining high-fidelity aerodynamic characteristic data of the airfoil profile of the target aircraft; the multi-task multi-fidelity model comprises a first multi-task expert mixed model and a second multi-task expert mixed model; the first multi-task expert mixed model and the second multi-task expert mixed model are both constructed according to a multi-task learning method and a gating mechanism; the first multitasking expert hybrid model is used to determine low fidelity aerodynamic feature data for the target aircraft airfoil profile; the second multitasking expert hybrid model is used to convert the low-fidelity aerodynamic data into high-fidelity aerodynamic data.
The invention also provides electronic equipment, which comprises a memory and a processor, wherein the memory is used for storing a computer program, and the processor runs the computer program to enable the electronic equipment to execute the airfoil aerodynamic characteristic prediction method.
The present invention also provides a computer readable storage medium storing a computer program which when executed by a processor implements the aerofoil aerodynamic characteristic prediction method as described above.
In the present specification, each embodiment is described in a progressive manner, and each embodiment is mainly described in a different point from other embodiments, and identical and similar parts between the embodiments are all enough to refer to each other.
The principles and embodiments of the present invention have been described herein with reference to specific examples, the description of which is intended only to assist in understanding the core concept of the invention; also, it is within the scope of the present invention to be modified by those of ordinary skill in the art in light of the present teachings. In view of the foregoing, this description should not be construed as limiting the invention.
Claims (8)
1. A method of aerofoil aerodynamic property prediction, comprising:
acquiring parameterized variables and flight conditions of the airfoil profile of the target aircraft;
utilizing a multi-task multi-fidelity model to predict aerodynamic characteristics of the parameterized variable and the flight condition, and determining high-fidelity aerodynamic characteristic data of the airfoil profile of the target aircraft; the multi-task multi-fidelity model comprises a first multi-task expert mixed model and a second multi-task expert mixed model; the first multi-task expert mixed model and the second multi-task expert mixed model are both constructed according to a multi-task learning method and a gating mechanism; the first multitasking expert hybrid model is used to determine low fidelity aerodynamic feature data for the target aircraft airfoil profile; the second multitasking expert hybrid model is used to convert the low-fidelity aerodynamic data into high-fidelity aerodynamic data.
2. The method for predicting aerodynamic properties of an airfoil according to claim 1, wherein the flight condition is a working condition of an aircraft during flight; the operating conditions include angle of attack, mach number and Reynolds number.
3. The method of claim 1, wherein the method of determining parameterized variables of the target aircraft airfoil profile is: potential variables of the target aircraft airfoil profile that are generated against the network are determined as parameterized variables.
4. The method for predicting aerodynamic properties of an airfoil according to claim 1, wherein said predicting aerodynamic properties of said parameterized variables and said flight conditions using a multitasking multi-fidelity model, determining high-fidelity aerodynamic properties data of said target aircraft airfoil profile, comprises:
inputting the parameterized variables and the flight conditions into the first multitasking expert hybrid model to determine low-fidelity aerodynamic data for the target aircraft airfoil profile;
and inputting the low-fidelity aerodynamic feature data into the second multitasking expert hybrid model, converting the low-fidelity aerodynamic feature data, and determining the high-fidelity aerodynamic feature data of the airfoil profile of the target aircraft.
5. The method of claim 1, wherein the method of determining the multi-tasking multi-fidelity model is:
acquiring training data; the training data comprises training parameter variables, flight training conditions and corresponding high-fidelity aerodynamic characteristic prediction results;
and inputting the training data into the first multi-task expert mixed model, taking the low-fidelity pneumatic characteristic data output by the first multi-task expert mixed model as the input of the second multi-task expert mixed model, and training by using a back propagation method and taking the minimum loss between the high-fidelity pneumatic characteristic data output by the second multi-task expert mixed model and the high-fidelity pneumatic characteristic prediction result as a target to obtain the multi-task multi-fidelity model.
6. An airfoil aerodynamic characteristic prediction system, comprising:
the data acquisition module is used for acquiring parameterized variables and flight conditions of the airfoil profile of the target aircraft;
the characteristic prediction module is used for predicting aerodynamic characteristics of the parameterized variable and the flight condition by utilizing a multi-task multi-fidelity model and determining high-fidelity aerodynamic characteristic data of the airfoil profile of the target aircraft; the multi-task multi-fidelity model comprises a first multi-task expert mixed model and a second multi-task expert mixed model; the first multi-task expert mixed model and the second multi-task expert mixed model are both constructed according to a multi-task learning method and a gating mechanism; the first multitasking expert hybrid model is used to determine low fidelity aerodynamic feature data for the target aircraft airfoil profile; the second multitasking expert hybrid model is used to convert the low-fidelity aerodynamic data into high-fidelity aerodynamic data.
7. An electronic device comprising a memory for storing a computer program and a processor that runs the computer program to cause the electronic device to perform the airfoil aerodynamic feature prediction method according to claims 1-5.
8. A computer readable storage medium, characterized in that it stores a computer program which, when executed by a processor, implements the aerofoil aerodynamic characteristic prediction method as claimed in claims 1-5.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310606863.2A CN116628854A (en) | 2023-05-26 | 2023-05-26 | Wing section aerodynamic characteristic prediction method, system, electronic equipment and storage medium |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310606863.2A CN116628854A (en) | 2023-05-26 | 2023-05-26 | Wing section aerodynamic characteristic prediction method, system, electronic equipment and storage medium |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN116628854A true CN116628854A (en) | 2023-08-22 |
Family
ID=87609396
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202310606863.2A Pending CN116628854A (en) | 2023-05-26 | 2023-05-26 | Wing section aerodynamic characteristic prediction method, system, electronic equipment and storage medium |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN116628854A (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117540489A (en) * | 2023-11-13 | 2024-02-09 | 重庆大学 | Airfoil pneumatic data calculation method and system based on multitask learning |
Citations (12)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20040019469A1 (en) * | 2002-07-24 | 2004-01-29 | Leary Stephen J. | Method of generating a multifidelity model of a system |
| CN113609596A (en) * | 2021-09-29 | 2021-11-05 | 中国空气动力研究与发展中心计算空气动力研究所 | Aircraft aerodynamic characteristic prediction method based on neural network |
| US20210357692A1 (en) * | 2020-05-15 | 2021-11-18 | Microsoft Technology Licensing, Llc | Multi-fidelity simulated data for machine learning |
| CN114676639A (en) * | 2022-03-31 | 2022-06-28 | 上海大学 | Aircraft aerodynamic shape optimization method, device and medium based on neural network |
| CN114692501A (en) * | 2022-04-01 | 2022-07-01 | 西安交通大学 | Pneumatic data fusion method and device based on multi-precision deep neural network |
| CN114692488A (en) * | 2022-03-15 | 2022-07-01 | 南方科技大学 | Method and device for generating supercritical airfoil, electronic equipment and storage medium |
| CN114722498A (en) * | 2022-04-02 | 2022-07-08 | 浙江大学 | Wing section design method based on generation of countermeasure network |
| CN114818477A (en) * | 2022-03-31 | 2022-07-29 | 上海大学 | Aircraft aerodynamic characteristic prediction method, device and medium |
| CN115310209A (en) * | 2022-09-15 | 2022-11-08 | 西安交通大学 | VAE-based pneumatic shape migration optimization method and related device |
| CN115438584A (en) * | 2022-09-16 | 2022-12-06 | 西北工业大学 | Wing profile aerodynamic force prediction method based on deep learning |
| CN115525980A (en) * | 2022-11-04 | 2022-12-27 | 北京航空航天大学 | Optimization method and optimization device for aerodynamic configuration of reentry vehicle |
| CN115659823A (en) * | 2022-11-02 | 2023-01-31 | 中国航空工业集团公司沈阳空气动力研究所 | Wing profile aerodynamic coefficient prediction method based on attribute reduction, electronic device and storage medium |
-
2023
- 2023-05-26 CN CN202310606863.2A patent/CN116628854A/en active Pending
Patent Citations (12)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20040019469A1 (en) * | 2002-07-24 | 2004-01-29 | Leary Stephen J. | Method of generating a multifidelity model of a system |
| US20210357692A1 (en) * | 2020-05-15 | 2021-11-18 | Microsoft Technology Licensing, Llc | Multi-fidelity simulated data for machine learning |
| CN113609596A (en) * | 2021-09-29 | 2021-11-05 | 中国空气动力研究与发展中心计算空气动力研究所 | Aircraft aerodynamic characteristic prediction method based on neural network |
| CN114692488A (en) * | 2022-03-15 | 2022-07-01 | 南方科技大学 | Method and device for generating supercritical airfoil, electronic equipment and storage medium |
| CN114676639A (en) * | 2022-03-31 | 2022-06-28 | 上海大学 | Aircraft aerodynamic shape optimization method, device and medium based on neural network |
| CN114818477A (en) * | 2022-03-31 | 2022-07-29 | 上海大学 | Aircraft aerodynamic characteristic prediction method, device and medium |
| CN114692501A (en) * | 2022-04-01 | 2022-07-01 | 西安交通大学 | Pneumatic data fusion method and device based on multi-precision deep neural network |
| CN114722498A (en) * | 2022-04-02 | 2022-07-08 | 浙江大学 | Wing section design method based on generation of countermeasure network |
| CN115310209A (en) * | 2022-09-15 | 2022-11-08 | 西安交通大学 | VAE-based pneumatic shape migration optimization method and related device |
| CN115438584A (en) * | 2022-09-16 | 2022-12-06 | 西北工业大学 | Wing profile aerodynamic force prediction method based on deep learning |
| CN115659823A (en) * | 2022-11-02 | 2023-01-31 | 中国航空工业集团公司沈阳空气动力研究所 | Wing profile aerodynamic coefficient prediction method based on attribute reduction, electronic device and storage medium |
| CN115525980A (en) * | 2022-11-04 | 2022-12-27 | 北京航空航天大学 | Optimization method and optimization device for aerodynamic configuration of reentry vehicle |
Non-Patent Citations (3)
| Title |
|---|
| XU WANG等: "Unsteady aerodynamic prediction for iced airfoil based on Multi-Task Learning", PHYSICS OF FLUIDS, pages 087117 - 1 * |
| 宋保维;王新晶;王鹏;: "基于变保真度模型的AUV流体动力参数预测", 机械工程学报, no. 18, pages 190 - 196 * |
| 韩少强;宋文萍;韩忠华;王乐;: "基于梯度增强型Kriging模型的气动反设计方法", 航空学报, no. 07, pages 138 - 152 * |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117540489A (en) * | 2023-11-13 | 2024-02-09 | 重庆大学 | Airfoil pneumatic data calculation method and system based on multitask learning |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN112016167B (en) | Aircraft aerodynamic shape design method and system based on simulation and optimization coupling | |
| Preidikman et al. | Time-domain simulations of linear and nonlinear aeroelastic behavior | |
| CN116628854A (en) | Wing section aerodynamic characteristic prediction method, system, electronic equipment and storage medium | |
| CN114065662A (en) | Method suitable for rapidly predicting airfoil flow field with variable grid topology | |
| CN115495965A (en) | Method for analyzing time-varying reliability of complex aviation structure under mixed uncertainty | |
| CN116090191A (en) | Simulation method and system of offshore wind turbine under comprehensive airflow factors | |
| CN114564787A (en) | Bayesian optimization method, device and storage medium for target-related airfoil design | |
| CN111027250A (en) | Special-shaped curved surface reinforced shell modeling method based on grid deformation technology | |
| Raul et al. | Aerodynamic shape optimization for delaying dynamic stall of airfoils using cokriging regression | |
| CN112414668B (en) | Wind tunnel test data static bomb correction method, device, equipment and medium | |
| Zhu et al. | Design of an RBF surrogate model for low Reynolds number airfoil based on transfer learning | |
| Piperni et al. | Singlepoint and multipoint robust design of airfoils using CST functions | |
| CN116167220A (en) | Simulation method and system of offshore wind turbine under comprehensive airflow factors | |
| CN114282310A (en) | Aeroelastic structure coupling optimization method based on self-adaptive point-adding proxy model | |
| CN114676639A (en) | Aircraft aerodynamic shape optimization method, device and medium based on neural network | |
| CN110826182B (en) | Aeroelastic design method of aircraft structure based on vertex method and sequential optimization strategy | |
| CN111177855B (en) | Pneumatic structure solving method and system in global aeroelasticity optimization | |
| Amrit et al. | Efficient multi-objective aerodynamic optimization by design space dimension reduction and co-kriging | |
| CN114117950B (en) | Flutter judgment method for shuttle aircraft based on acting principle | |
| CN112733259B (en) | Pneumatic servo elasticity rapid iterative analysis and design method | |
| Sridharan et al. | Accelerating Aerodynamic Design of Rotors using a Multi-Fidelity Approach in TORC: Tool for Optimization of Rotorcraft Concepts | |
| Zuo et al. | Efficient aeroelastic design optimization based on the discrete adjoint method | |
| Ma et al. | Airfoil optimization design based on a combined optimization strategy | |
| CN117874925A (en) | Gradient type pneumatic shape optimization method, system and equipment based on deep learning | |
| CN116414044A (en) | Aeroengine control method and system based on self-adaptive composition |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination |