CN107391891B - Large-aspect-ratio wing optimization design method based on model fusion method - Google Patents

Abstract

The invention discloses a high aspect ratio wing optimization design method based on a model fusion method, and belongs to the field of overall optimization design of aircrafts. According to the optimization method, optimization is divided into an optimization model and a system-level optimization model of a structural subject according to requirements, and complex constraints are processed by using a penalty function method; establishing a high-precision pneumatic structure coupling analysis model and a low-precision pneumatic structure coupling analysis model by using a pneumatic structure coupling modeling technology; respectively generating high-precision sample points and low-precision sample points by using a test design method; respectively calling a high-precision structure coupling analysis model and a low-precision structure coupling analysis model to obtain high-precision sample information and low-precision sample information and storing the high-precision sample information and the low-precision sample information; fusing high-precision model information and low-precision model information by using a model fusion method to establish a proxy model; and performing optimization solution by using an optimization method based on the current agent model, judging whether the optimization result is credible according to the difference value between the real response value at the optimal solution and the agent model value based on the model fusion method, returning to the reconstructed fusion model for optimization solution if the optimization result is not credible, and outputting the optimal design result if the optimization result is credible to complete the optimization design.

Description

An Optimal Design Method for Large Aspect Ratio Wings Based on Model Fusion

technical field

The invention relates to a large aspect ratio wing optimization design method based on a model fusion method, and belongs to the technical field of overall optimization design of aircraft.

Background technique

Large aspect ratio wings have the characteristics of large lift-to-drag ratio and large in-wing volume, and are widely used in high-altitude UAVs, solar-powered aircraft, large intercontinental passenger aircraft and other aircraft. During the flight of this type of aircraft, the wings with large aspect ratio are affected by aerodynamic loads, resulting in structural deformation, and the deformation amplitude has a significant impact on the aerodynamic performance. Therefore, the aerodynamic-structure coupling problem needs to be considered when analyzing and designing a wing with a large aspect ratio. For the coupling problem of aerodynamic structure, fluid mechanics and structural mechanics can be solved independently as a single discipline, and the interdisciplinary data interaction can be realized through software scheduling technology, and the coupled analysis can be realized by iterative solution. In order to improve the accuracy of coupled analysis, high-precision analysis methods such as computational fluid dynamics (CFD) and finite element analysis (FEA) are often used to analyze and solve the two single disciplines. However, the high-precision analysis model not only improves the analysis accuracy and reliability, but also brings the problem of time-consuming calculation. Although today's computer software and hardware technology has made great progress, it is still difficult to call the high-precision analysis model to complete an iterative solution. Extremely time consuming. For example, it takes hours or even tens of hours to complete a pneumatic simulation analysis using a CFD model. The optimization design of a wing with a large aspect ratio is also an iterative process. In the optimization process, it often needs to call the high-precision coupled analysis model thousands of times, which further increases the design cost and makes the optimization design efficiency very low.

In order to better illustrate the technical solution of the present invention, the applied aerodynamic structure coupling modeling technology is introduced in detail below.

Aerodynamic structure coupling modeling technology:

As the wing aspect ratio increases, the flexibility of the wing continues to increase, and the coupling phenomenon between its aerodynamic performance and structural performance becomes more obvious. The key to aerodynamic structure coupling modeling technology is the information transfer between aerodynamics and structural disciplines. In the existing mature aero-structure coupled modeling technology, the three-dimensional interpolation method is often used to realize the transfer of aerodynamic analysis results to the structural discipline, and at the same time, the geometric shape of the updated wing is determined according to the coordinates of the control points on the leading and trailing edges of the deformed wing. , re-analyze the aerodynamic discipline, and complete the transfer of the structural discipline analysis results to the aerodynamic discipline. On the other hand, in the coupled modeling of aerodynamic structures, the density of aerodynamic grids often controls the computational cost and model accuracy of the entire coupled analysis model. Increasing the mesh density can improve the accuracy of the analytical model, but it will also increase the computational cost. On the contrary, reducing the mesh density will reduce the calculation accuracy and reduce the calculation cost.

The flow chart of the aerodynamic structure coupling analysis model is shown in Figure 1, and the specific method steps are as follows:

Step 1. Use the model parameterization technology based on UG secondary development to establish/update the wing parameterization model. The parameterized parameters of the geometric model include geometric design variables aspect ratio, root-to-tip ratio, sweep angle, airfoil parameters, and the coordinate information of the position control points of the leading and trailing edges of the wing used to quantitatively represent the deformation of the structural shape. After the parameterization is completed, the output geometry file is used for subsequent analysis, generally in step format or igs format.

Step 2. Build an aerodynamic analysis model using CFD. Input geometry file, aerodynamic analysis working condition information including Mach number, angle of attack, output aerodynamic analysis results including lift, drag information and aerodynamic distribution file. Gambit can be used for mesh drawing and Fluent for aerodynamic analysis.

Step 3. Use the FEA method to establish a structural subject analysis model, use Patran for pre-processing, and Nastran for post-processing. Input geometry files, use PCL language for material properties, element property definitions, and aerodynamic loading and other related analysis and optimization settings. The SQP optimizer that comes with Nastran enables structural discipline optimization. The final output structural analysis structure includes maximum stress and maximum displacement and coordinate information for control points on the leading and trailing edges of the wing.

Step 4. If it is the first analysis, input the deformed control point coordinate information into the model parameterization module, and repeat steps 2, 3, and 4. If it is not the first analysis, calculate the relative displacement. The calculation of the relative displacement η is as follows: Formula (1) is shown. When the relative displacement is less than 0.01, the iteration ends, and the aerodynamic analysis results are output at this time, including the lift-drag ratio, mass, maximum stress, maximum displacement, and optimization results of structural variables; when the relative displacement is greater than 0.01, repeat steps 1 and 1. 2, 3, 4, until the relative displacement is less than 0.01, the iteration ends.

In formula (1), i represents the ith analysis, and i-1 represents the previous analysis of the ith analysis.

SUMMARY OF THE INVENTION

Aiming at the problem that the calculation cost is too high when considering the coupling of aerodynamic structures in the process of optimizing the design of a wing with a large aspect ratio, the invention discloses a method for optimizing the design of a wing with a large aspect ratio based on a model fusion method, and the technical problem to be solved Under the condition of ensuring the accuracy, considering the aerodynamic structure coupling problem to realize the efficient optimization design of the wing with large aspect ratio, it has the following advantages: using the model fusion method to effectively fuse the high- and low-precision analytical model information, making full use of the low-precision model The information ensures the accuracy of the fusion model and reduces the number of calls of the high-precision analysis model, thereby reducing the computational cost and improving the optimization design efficiency of large aspect ratio wings.

The purpose of the present invention is achieved through the following technical solutions.

The invention discloses a large aspect ratio wing optimization design method based on a model fusion method. The initial reference airfoil and the relevant shape parameters of the wing are selected according to the design requirements, and the design working conditions are determined; the optimization model of the structural discipline is established according to the requirements. System-level optimization model, and use penalty function method to deal with complex constraints; use aero-structure coupled modeling technology to establish high- and low-precision high-aspect-ratio wing aero-structure coupled analysis models; use experimental design method to generate high and low-precision sample points respectively ;Invoke the high- and low-precision high-aspect ratio wing aerodynamic structure coupling analysis models to obtain high- and low-precision sample information and store them; use the model fusion method to fuse the high-precision and low-precision model information, and establish a proxy model to achieve model accuracy Comprehensive coordination with computational cost; use the optimization method to optimize the solution based on the current surrogate model, and determine whether the optimization result is credible based on the difference between the real response value at the optimal solution and the surrogate model value based on the model fusion method, and if not, then Return to reconstruct the fusion model for optimization solution. If it is credible, output the optimal design result, that is, complete the efficient optimization design of the wing with large aspect ratio considering the coupling problem of aerodynamic structure.

The invention discloses a large aspect ratio wing optimization design method based on a model fusion method, comprising the following steps:

Step 1: According to the design requirements, select the initial reference airfoil and relevant shape parameters of the airfoil to determine the design conditions.

The design conditions include Mach number and angle of attack.

Step 2: Establish the optimization model and system-level optimization model of structural disciplines according to the requirements.

Step 2.1: Establish an optimization model of structural disciplines according to requirements.

In order to minimize the mass while maintaining the structural strength, the size of each structural component of the wing is optimized during the structural analysis. The design variables include skin thickness, web thickness, flange radius, web thickness, and flange radius; the optimization objective is to minimize the structural mass; the constraints are to satisfy the maximum stress constraint and the maximum displacement deformation constraint. The optimization of the structural discipline is realized in the structural discipline analysis model.

Step 2.2: Build a system-level optimization model based on requirements.

The geometric design parameters are selected as design variables in the system-level optimization. The design variables include aspect ratio, root-to-tip ratio, and sweep angle, and the upper and lower limits are determined according to the requirements; the maximum lift-drag ratio and the minimum structural quality are the optimization goals. Constraints include that the maximum structural stress is less than the allowable stress, the maximum structural displacement is less than the allowable displacement, and the wing area remains unchanged.

Step 3: Use the aerodynamic-structure coupled modeling technology to establish a high- and low-precision high-aspect ratio wing aerodynamic-structure coupled analysis model. The grid density of aerodynamics is the main factor of calculation cost and calculation accuracy. Therefore, in the aerodynamic analysis model, coarse grids are used to establish low-precision analysis models, and fine grids are used to establish high-precision analysis models.

In step 3, the coordination of calculation accuracy and calculation cost is achieved through the density of aerodynamic grids, and a high- and low-accuracy aerodynamic structure coupling analysis model is established.

Step 4: Generate N ^h high-precision sample points and N ^l low-precision sample points respectively using the design of experiments method. The number of sample points is related to the system-level optimization design variable dimension _nv . The low-precision sample points need to include all high-precision sample points.

In order to realize the high efficiency of the high aspect ratio wing optimization design considering the aerodynamic-structure coupling problem, the experimental design method described in step 4 preferably uses the Latin hypercube experimental design method.

The number of sample points is determined according to theoretical analysis, experiments or empirical values, preferably N ^h =(n _v +3)*(n _v +2), 4N ^h ≤N ^l ≤6N ^h .

Step 5: Invoke the high- and low-precision large aspect ratio wing aerodynamic structure coupling analysis model in step 3, obtain the model response values at the N ^h and N ^l sample points in step 4, and store the high and low-precision sample point information.

Step 6: Use the model fusion method to fuse the high- and low-precision sample point information to establish a proxy model. The surrogate model is a fusion model y ^s (x) composed of the surrogate model of the correction model and the surrogate model of the error model.

The specific implementation method of step 6 is as follows:

Step 6.1: According to the high-precision samples and the corresponding low-precision sample information, use the least squares method to obtain the correction factor of the low-precision and large-aspect-ratio wing aerodynamic structure coupling analysis sample points, as shown in formula (2):

Among them: N ^h is the number of sample points for high-precision and large-aspect-ratio wing aerodynamic structure coupling analysis; y ^h ( _xi ) is the response value of the high-precision aerodynamic structure coupling analysis model, and y ^l ( _xi ) is low-precision aerodynamic structure The response value of the structural coupling analysis model, the response value includes the lift-drag ratio, the structural mass, the maximum stress of the structure, and the maximum displacement of the structure; ρ ₀ and ρ ₁ are the correction factors of the sample points of the low-precision aerodynamic structure coupling analysis model, and each The response values have their own corresponding correction factors.

Step 6.2: Use the correction factor of the sample points of the low-precision aerodynamic structure coupling analysis model in step 6.1 to correct all the sample points of the low-precision aerodynamic structure coupling analysis model, and use the Kriging method based on the corrected sample information of the low-precision aerodynamic structure coupling analysis model To construct a proxy model, the revised model y ^l _s (x) of the low-precision aerodynamic-structure coupled analysis model is expressed as:

y ^l _s (x)=ρ ₀ +ρ ₁ y ^l (x) (3)

Among them, y ^l (x) is the response value of the sample point of the low-precision aerodynamic structure coupling analysis model. Use formula (3) to correct all the sample point data of the low-precision aerodynamic structure coupling analysis model to obtain the corrected model of the low-precision aerodynamic structure coupling analysis model. y ^l _s (x). The surrogate model y _s ^s (x) of the modified model y ^l _s (x) is constructed by using the Kriging method.

Step 6.3: Calculate the error value δ(x _i ) between the sample points of the high-precision aerodynamic structure coupling analysis model in step 5 and the corrected model of the low-precision aerodynamic structure coupling analysis model in step 6.2, and the error value δ( _xi ) is calculated by the formula ( 4) Calculate to get:

δ(x _i )=y ^h (x _i )-y ^l _s (x _i )=y ^h (x _i )-[ρ ₀ +ρ ₁ y ^l (x _i )])i=1,2,3… N ^h ) (4)

Based on the error information δ(x _i ), the Kriging method is used to complete the construction of the surrogate model δ ^s (x) of the error model.

Step 6.4: Construct a fusion model y ^s (x) composed of the surrogate model y _s ^s (x) of the corrected model in step 6.2 and the surrogate model δ ^s (x) of the error model in step 6.3, as shown in formula (5) shown:

y ^s (x)=y _s ^s (x)+δ ^s (x) (5)

The fusion model y ^s (x) is a proxy model of the high-precision aerodynamic structure coupling analysis model.

Step 7: Based on the fusion model y ^s (x) established in step 6, use the penalty function to deal with the complex constraints in the problem, use the optimization algorithm to solve the system optimization problem, and obtain the optimal solution based on the current fusion model y ^s (x). untie

The penalty function is shown in formula (6):

F(x)=f(x)+MP(x)M>0

Among them: F(x) is the processed optimization objective, f(x) is the original optimization objective, M is the penalty factor, P(x) is the constraint violation degree, g _i (x) is the inequality constraint, h _i (x) is the equality constraint, m is the number of inequality constraints, and l is the total number of constraints.

In step 7, the optimization algorithm is used to solve the system optimization problem, and the optimal genetic algorithm is used to solve it.

处的真实响应值。计算最优解的真实响应值与代理模型值的差值，根据差值大小判定该优化结果是否可信。若不可信则返回步骤4，增加低精度大展弦比机翼气动结构耦合分析模型样本点数量，重复步骤5、6、7、8，直至获得可信的优化结果，若可信则输出最优设计结果，即完成考虑气动结构耦合问题的大展弦比机翼的高效优化设计。Step 8: Invoke the high-precision large aspect ratio wing aerodynamic structure coupling analysis model in step 3 to obtain the optimal solution of the surrogate model in step 7

the true response value at . Calculate the difference between the real response value of the optimal solution and the surrogate model value, and determine whether the optimization result is credible according to the difference. If it is not credible, go back to step 4, increase the number of sample points of the low-precision large aspect ratio wing aerodynamic structure coupling analysis model, and repeat steps 5, 6, 7, and 8 until credible optimization results are obtained. If credible, output the most The optimal design result is to complete the high-efficiency optimal design of the wing with a large aspect ratio considering the aerodynamic-structure coupling problem.

Beneficial effects:

1. In view of the problem that the aerodynamic structure coupling problem needs to be considered when designing a large aspect ratio wing, which leads to the unbearable calculation cost of the optimization design process, the present invention discloses a large aspect ratio wing optimization design method based on a model fusion method, through The mesh density control realizes the comprehensive coordination of analytical model accuracy and computational cost.

2. A large aspect ratio wing optimization design method based on the model fusion method disclosed in the present invention uses the model fusion method to efficiently fuse high- and low-precision model information, and reduces the number of high-precision analysis models while meeting the design accuracy requirements. The number of calls is reduced, the computational cost is reduced, and the design efficiency of large aspect ratio wings is improved.

3. The present invention discloses a large aspect ratio wing optimization design method based on a model fusion method, which uses a penalty function to deal with complex constraints, thereby realizing the simplicity and convenience of the optimization design process.

4. A large aspect ratio wing optimization design method based on a model fusion method disclosed in the present invention uses the difference between the real response value of the optimal solution and the surrogate model value to determine the reliability of the optimal result to complete the optimization process update and iteration.

5. A large aspect ratio wing optimization design method based on the model fusion method disclosed in the present invention uses the Kriging surrogate model method to efficiently complete the construction of the surrogate model of the correction model and the surrogate model of the error model. Low-precision model information is fused more efficiently.

6. A large aspect ratio wing optimization design method based on the model fusion method disclosed in the present invention uses genetic algorithm for optimization solution, which can avoid the situation of no solution in the optimization solution, and improve the solution feasibility of engineering optimization design problems. .

Description of drawings

Figure 1 is the flow chart of the coupling analysis of the aerodynamic structure of the wing with a large aspect ratio;

Fig. 2 is a flow chart of the optimal design method for a wing with a large aspect ratio considering the aerodynamic structure coupling;

Figure 3 is a comparison diagram of high and low precision grids in aerodynamics.

Among them, Figure 3a is a high-precision analysis model grid, and Figure 3b is a low-precision analysis model grid;

Fig. 4 is the flow chart of model fusion method;

Detailed ways

In order to better illustrate the technical solutions and advantages of the present invention, the present invention is further described below through a specific example of the optimal design of a wing with a large aspect ratio, combined with the accompanying drawings and tables, and the specific embodiments are as follows.

A method for optimal design of a wing with a large aspect ratio based on a model fusion method disclosed in this embodiment, the flowchart is shown in Figure 2, and the specific implementation steps are as follows:

Step 1: Select the laminar airfoil NACA64A816 as the reference initial airfoil, the design condition is the flight Mach number Ma=0.64, the angle of attack α=2° The wing shape is determined by the initial value of the system variable, as shown in Table 2.

Step 2: Establish the optimization model and system-level optimization model of structural disciplines according to the requirements.

Step 2.1: Establish an optimization model of structural disciplines according to requirements.

Each structural component of the wing is dimensionally optimized in the optimization model of the structural discipline. Select the skin thickness for each wing box (T _skin ), the web thickness for each rib (T _rib ), the flange radius for each rib (R _rib ), the web thickness for each spar (T rib ) _spar ), the upper and lower flange radius (R _spar ) of each spar, as the structural discipline optimization design variable. Constraints include that the maximum structural stress σ _max is less than the allowable stress 100MPa, and the maximum structural displacement δ _max is less than the allowable displacement 900mm. The objective of structural optimization is to minimize the structural mass W of the wing. The optimization of the structural discipline is realized in the structural discipline analysis model. The structural optimization model is shown in the following formula (7).

Among them, x _struc is the optimization design variable of the structural discipline, x _struc ^lb and x _struc ^ub are the upper and lower limits of the structural design variable, respectively, and the values of the design variables are shown in Table 1.

Table 1 Structural Design Variables and Variation Scope

Step 2.2: Build a system-level optimization model based on requirements.

The geometric design parameters are selected as design variables in the system-level optimization. The geometric design parameters include aspect ratio, root-to-tip ratio, and sweep angle, and their upper and lower limits are determined according to requirements, as shown in Table 2; the junction lift-to-drag ratio D /L is the largest, and the structural mass W is the smallest as the optimization objective. The constraints include that the maximum structural stress σ _max is less than the allowable stress 100MPa, the maximum structural displacement δ _max is less than the allowable displacement 900mm, and the wing area is constant and constant at 50.17m ² . The system-level optimization model is shown in equation (8).

min F(X)=1/2×C _weight +1/2×C _D/L

stσ _max ≤100Mpa (8)

δmax _≤900mm

X ^lb ≤X≤X ^ub

S=50.17m ²

F(X) is the comprehensive optimization objective, which is obtained by the linear weighting of the structural mass W and the lift-drag ratio D/L of the two optimization objectives. Since the order of magnitude of each target is different, the structural mass W ^baseline of the initial wing and the lift-to-drag ratio (D/L) ^baseline of the initial wing are used to normalize the structural mass of the optimized target and the lift-to-drag ratio, respectively, to obtain Normalized objective function response values C _weight and C _D/L . X is the design variable, X ^lb and X ^ub are the upper and lower limits of the design variable, respectively, and the specific values are shown in Table 2.

Table 2 System-level design variables and their variation ranges

Step 3: Use the aerodynamic structure coupling modeling technology to establish the aerodynamic structure coupling analysis model of the high aspect ratio wing and the aerodynamic structure coupling analysis model of the low-precision large aspect ratio wing. In this step, the distinction between high and low precision analytical models is achieved by adjusting the density of grid rendering. In this embodiment, the high-precision mesh density is twice that of the low-precision grid, as shown in FIG. 3 . In the structural subject analysis model, the element attributes of the finite element model of the structural subject are defined as shown in Table 3. The SQP optimizer that comes with Nastran is used to complete the optimization of the structural subject. The optimization model is described in the optimization model of the structural subject in step 2.1. .

Table 3 Element properties of finite element model

Step 4: Generate high-precision model sample points and low-precision model sample points using the Latin Hypercube Design of Experiments method. In the present invention, the computation cost is calculated in terms of CPU computation time. According to experimental statistics, it takes about 20 minutes for each high-precision analysis model, and the computational cost for each low-precision analysis model is about 3 minutes. Generate 30 high-precision sample points and 130 low-precision sample points for the model fusion method. For efficiency comparison, 50 high-precision sample points (the computation cost is about the sum of 30 high-precision analyses and 130 low-precision analyses) are simultaneously generated for constructing surrogate models using traditional methods.

The above-mentioned use of the traditional method to construct the surrogate model in this example is the surrogate model that directly uses the Kriging method to directly construct a high-accuracy and large aspect ratio wing aerodynamic structure coupled analysis model.

Step 5: Invoke the high- and low-precision high-aspect ratio wing aerodynamic structure coupling analysis model in Step 3, obtain the model response values at the 30 high-precision sample points and 130 low-precision sample points in Step 4, and store the high and low Accurate large aspect ratio wing aero-structure coupled analysis model sample point information. For efficiency comparison, the model response values at 50 high-precision sample points used to construct the surrogate model based on the traditional method should be obtained at the same time.

Step 6: Use the model fusion method to fuse the 30 high-precision sample point information and 130 low-precision sample point information in step 5, and establish the proxy of the high-precision large-span chord wing aerodynamic structure coupling analysis model based on the model fusion method Model. Its specific flow chart is shown in Figure 4. At the same time, the surrogate model of the coupled analysis model of the aerodynamic structure of the high-precision large-span chord wing is directly constructed based on the traditional method using the information of 50 high-precision sample points.

Step 7: Based on the surrogate model established by the model fusion method established in step 6 and the surrogate model constructed by the traditional method, the genetic algorithm is used to optimize. For the complex constraints in the optimization problem, the penalty function is used to deal with it, and the penalty factor is taken as 1000. The optimal solution based on the model fusion method and the optimal solution based on the traditional method are obtained respectively.

处的真实响应值。计算最优解的真实响应值与代理模型值的差值，根据差值大小判定该优化结果是否可信。若不可信则返回步骤4，增加低精度大展弦比机翼气动结构耦合分析模型样本点数量，重复步骤5、6、7、8，直至获得可信的优化结果，若可信则输出最优设计结果，即完成考虑气动结构耦合问题的大展弦比机翼的高效优化设计。Step 8: Invoke the high-precision large aspect ratio wing aerodynamic structure coupling analysis model in step 3 to obtain the optimal solution of the surrogate model in step 7

the true response value at . Calculate the difference between the real response value of the optimal solution and the surrogate model value, and determine whether the optimization result is credible according to the difference. If it is not credible, go back to step 4, increase the number of sample points of the low-precision large aspect ratio wing aerodynamic structure coupling analysis model, and repeat steps 5, 6, 7, and 8 until credible optimization results are obtained. If credible, output the most The optimal design result is to complete the high-efficiency optimal design of the wing with a large aspect ratio considering the aerodynamic-structure coupling problem.

The optimization results of statistical systems are shown in Table 4, and the optimization results of structural disciplines are shown in Table 5.

Table 4 Optimization results of 3D wing system

Observe Table 4 and compare the results of the optimized design using the two methods. It can be found that the lift-to-drag ratio has been slightly improved, but the structural quality has changed significantly. Using the optimization design method in the present invention, the quality is reduced by 54%, while the quality is only reduced by 4% in the optimization result of the traditional method. The normalized comprehensive optimization target value after using the method of the present invention is 0.7258, which is much smaller than the optimization result of the traditional method under the same computational cost. At the same time, when the optimal design method of the present invention is used, the difference between the real target value and the surrogate model value is small, indicating that the optimal design method of the present invention has higher precision. Through the comparison of experimental results, it is concluded that the method of the present invention not only has higher model accuracy, but also has better optimal design results under the same computational cost as compared with the traditional method. Therefore, using the large aspect ratio wing optimization design method based on the model fusion method proposed by the present invention to carry out the large aspect ratio wing optimization design can ensure the accuracy while reducing the call amount of the high-precision analysis model, thereby reducing the calculation. cost and improve the optimal design efficiency of large aspect ratio wings.

Table 5 Optimization results of 3D wing structure

According to the analysis of the above-mentioned specific example of the optimization of a wing with a large aspect ratio, it can be seen that the present invention can achieve the intended purpose of the invention. Wing optimization design results and design quality; on the other hand, in relation to the optimization of high-aspect-ratio wing high-precision analysis models, the present invention can also greatly improve optimization efficiency, reduce optimization design costs, and shorten optimization design cycles.

The above-mentioned specific description further describes the purpose, technical solutions and beneficial effects of the invention in detail. It should be understood that the above-mentioned descriptions are only specific embodiments of the present invention, which are used to explain the present invention and are not intended to be used for The protection scope of the present invention is limited, and any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention shall be included within the protection scope of the present invention.

Claims (6)

1. A high aspect ratio wing optimization design method based on a model fusion method is characterized by comprising the following steps: comprises the following steps of (a) carrying out,

step 1: according to design requirements, selecting initial reference wing profiles and relevant shape parameters of wings, and determining design working conditions;

step 2: establishing an optimization model and a system level optimization model of the structural discipline according to requirements;

and step 3: establishing a high-precision high-aspect-ratio wing aerodynamic structure coupling analysis model by using an aerodynamic structure coupling modeling technology; the density of the grid of the pneumatic discipline is a main factor of the calculation cost and the calculation precision, so that a low-precision analysis model is established by using a coarse grid in a pneumatic analysis model, and a high-precision analysis model is established by using a fine grid;

and 4, step 4: separately generating N Using a design of experiments approach^hA high precision sample point and N^lA low precision sample point; number of sample points and system-level optimization design variable dimension n_vCorrelation; wherein all the high-precision sample points need to be contained in the low-precision sample points;

and 5: calling the high-precision and low-precision high-aspect-ratio wing aerodynamic structure coupling analysis model in the step 3 to obtain N in the step 4^hAnd N^lStoring the high-precision sample point information and the low-precision sample point information according to the model response value at each sample point;

step 6, fusing the high-precision sample point information and the low-precision sample point information by using a model fusion method to establish a proxy model; the proxy model is a fusion model y consisting of a proxy model of a correction model and a proxy model of an error model^s(x)；

The specific implementation method of the step 6 is as follows:

step 6.1: according to the high-precision sample and the corresponding low-precision sample information, a least square method is used for obtaining correction factors of low-precision and high-aspect-ratio wing aerodynamic structure coupling analysis sample points, wherein the correction factors are as shown in formula (1):

wherein: n is a radical of^hAnalyzing the number of sample points for the coupling of the high-precision high-aspect-ratio wing aerodynamic structure; y is^h(x_i) Response value, y, for a high-precision pneumatic structure coupling analysis model^l(x_i) Response values of the coupling analysis model of the low-precision pneumatic structure comprise lift-drag ratio, structure mass, structure maximum stress and structure maximum displacement; rho₀、ρ₁Each response value has a corresponding correction factor for the correction factor of the low-precision pneumatic structure coupling analysis model sample point;

step 6.2: correcting all low-precision pneumatic structure coupling analysis model sample points by using correction factors of the low-precision pneumatic structure coupling analysis model sample points in the step 6.1, constructing an agent model by using a Kriging method based on the corrected low-precision pneumatic structure coupling analysis model sample information, and correcting the model y of the low-precision pneumatic structure coupling analysis model^l _s(x) Expressed as:

y^l _s(x)＝ρ₀+ρ₁y^l(x) (2)

wherein y is^l(x) For the response value of the sample point of the low-precision pneumatic structure coupling analysis model, correcting all sample point data of the low-precision pneumatic structure coupling analysis model by using the formula (2) to obtain a corrected model y of the low-precision pneumatic structure coupling analysis model^l _s(x) (ii) a Method for finishing correction model y of low-precision pneumatic structure coupling analysis model by using Kriging method^l _s(x) Agent model y of_s ^s(x) Constructing;

step 6.3: calculating the error value delta (x) between the high-precision pneumatic coupling analysis model in the step 5 and the correction model of the low-precision pneumatic structure coupling analysis model in the step 6.2_i) Error value delta (x)_i) Obtained by calculation of equation (3):

δ(x_i)＝y^h(x_i)-y^l _s(x_i)＝y^h(x_i)-[ρ₀+ρ₁y^l(x_i)](i＝1,2,3…N^h) (3)

based on error information delta (x)_i) Using Kriging method to complete the proxy model delta of error model^s(x) The structure of (1);

step 6.4: constructing a proxy model y from the modified model in step 6.2_s ^s(x) Surrogate model delta to the error model in step 6.3^s(x) Composed fusion model y^s(x) As shown in formula (4):

y^s(x)＝y_s ^s(x)+δ^s(x) (4)

the fusion model y^s(x) A proxy model for the high-precision pneumatic structure coupling analysis model;

and 7: based on the fusion model y established in step 6^s(x) Using complex constraint in penalty function processing problem, using optimization algorithm to solve system optimization problem to obtain current fusion model y^s(x) Of (2) an optimal solution

The penalty function is shown in equation (5):

F(x)＝f(x)+M*P(x)M＞0

wherein: f (x) is the processed optimization target, f (x) is the original optimization target, M is the penalty factor, P (x) is the constraint violation degree, g_i(x) Is inequality constraint, h_i(x) Is equality constraint, m is inequality constraint number, l is constraint total number;

and 8: calling the high-precision high-aspect-ratio wing aerodynamic structure coupling analysis model in the step 3 to obtain the optimal solution of the fusion model in the step 7

The true response value of (d); calculating a difference value between a real response value of the optimal solution and the fusion model, and judging whether the optimization result is credible according to the difference value; and if the model is not credible, returning to the step 4, increasing the number of sample points of the aerodynamic structure coupling analysis model of the low-precision high-aspect-ratio wing, repeating the steps 5, 6, 7 and 8 until a credible optimization result is obtained, and if the model is credible, outputting an optimal design result, namely completing the high-efficiency optimization design of the high-aspect-ratio wing considering the aerodynamic structure coupling problem.

2. The optimization design method of the high aspect ratio wing based on the model fusion method as claimed in claim 1, wherein: the specific implementation method of the step 2 is that,

step 2.1: establishing an optimization model of the structural discipline according to requirements;

in order to reduce the mass to the maximum extent while ensuring the structural strength, the size of each structural component of the wing is optimized in the structural analysis process; design variables include skin thickness, web thickness, flange radius; the optimization goal is that the structure quality is minimum; the constraint condition is that the maximum stress constraint and the maximum displacement deformation constraint are satisfied; the optimization of the structural disciplines is realized in a structural discipline analysis model;

step 2.2: establishing a system-level optimization model according to requirements;

selecting geometric design parameters as design variables in system level optimization, wherein the design variables comprise an aspect ratio, a root-tip ratio and a sweep angle, and determining upper and lower limits according to requirements; the maximum lift-drag ratio and the minimum structure mass are taken as optimization targets, and the constraint conditions comprise that the maximum structure stress is smaller than the allowable stress, the maximum structure displacement is smaller than the allowable displacement and the wing area is unchanged.

3. The optimization design method of the high aspect ratio wing based on the model fusion method as claimed in claim 1 or 2, wherein: and 3, realizing coordination of calculation precision and calculation cost through the density degree of the grid density of the pneumatic discipline, and establishing a high-precision and low-precision pneumatic structure coupling analysis model.

4. The high aspect ratio wing optimization design method based on the model fusion method as claimed in claim 3, wherein: in order to realize the high efficiency of the optimized design of the high-aspect-ratio wing considering the coupling problem of the aerodynamic structure, the experimental design method in the step 4 adopts a Latin hypercube experimental design method.

5. The method of claim 4, wherein the method comprises a step of optimizing the design of the high aspect ratio wing based on a model fusion methodIs characterized in that: the number of sample points in step 4 is determined according to theoretical analysis, experiment or empirical value, and N is taken^h＝(n_v+3)*(n_v+2)，4N^h≤N^l≤6N^h。

6. The optimization design method of the high aspect ratio wing based on the model fusion method as claimed in claim 5, wherein: and 7, using an optimization algorithm to solve the system optimization problem and select a genetic algorithm to solve.

Images