Generative model-based wide-speed-domain airfoil profile discrimination and generation method

Abstract

The invention provides a wide-speed-range airfoil profile distinguishing and generating method based on a generative model. The wide-speed-range airfoil profile distinguishing and generating method comprises an airfoil profile distinguishing process and an airfoil profile generating process. The method comprises the steps that firstly, airfoils are judged from an existing airfoil library, and the airfoils meeting the wide speed range condition are found; then, airfoils meeting the wide speed range condition are selected from an existing airfoil library, the generated airfoils serve as a training set, and a generative model wing-GAN is trained; by training the generative model wing-GAN, more airfoils conforming to the wide-speed-range aircraft can be obtained, the generation speed is high, the accuracy is high, the airfoils conforming to the performance of the aircraft can be conveniently and rapidly found out, and the combat performance of the aircraft is greatly improved.

Description

Technical Field

[0001] The present invention relates to the field of aircraft design, and particularly to a wide-speed domain airfoil discrimination and generation method based on a generative model. Background Art

[0002] With the continuous development and progress of military technology, humans have imposed stricter control over the aviation territory. It is imperative to develop aircraft that can have good flight performance at high altitudes and high speeds. In the process of globalization, the improvement of the flight speed of aircraft can further gain the initiative and improve the combat efficiency. Therefore, the research on hypersonic vehicles has always attracted much attention.

[0003] A wide-speed domain aircraft refers to an aircraft that can fly autonomously within two or more speed domains. Among them, a full-speed domain aircraft can achieve autonomous flight under low-speed, transonic, supersonic, and hypersonic conditions. As a combat weapon, it has high speed and high flexibility, reducing the probability of being intercepted. Since such wide-speed domain aircraft need to meet the flight ranges of two or more different speed domains, the design of its aerodynamic shape is relatively complex and needs to take into account two or more different speed domains at the same time. Therefore, it is one of the focuses of the competition of the combat strength of aircraft.

[0004] There is little research on the airfoils of such wide-speed domain aircraft in China, and there is no clear regulation on the conditions that the airfoils of wide-speed domain aircraft need to meet. At present, the discrimination of wide-speed domain airfoils is very vague. Since the aerodynamic performance of airfoils varies greatly under different Reynolds numbers and different Mach numbers, it is very difficult to determine whether an airfoil can maintain good aerodynamic performance in multiple speed domains. In addition, the existing airfoil library is small in quantity, and it is impossible to quickly obtain airfoils that meet the performance of the aircraft. There is an urgent need to develop a database of such airfoils. Summary of the Invention

[0005] In order to solve the problems of the prior art, the present invention provides a wide-speed domain airfoil discrimination and generation method based on a generative model. By training the generative model wing-GAN, more airfoils that meet the wide-speed domain aircraft can be obtained, with a faster generation speed and high accuracy, which is convenient for quickly finding airfoils that meet the performance of the aircraft and greatly improving the combat performance of the aircraft.

[0006] The present invention includes an airfoil discrimination process and an airfoil generation process.

[0007] Step 1) Airfoil discrimination process: From the existing airfoil library, discriminate the airfoils to find the airfoils that meet the wide-speed domain conditions; the specific process is as follows: Step 1.1) Select several airfoils from the airfoil library, model them and draw a curve graph; Step 1.2) Divide the drawn curve graph into computational grids and set boundary conditions; Step 1.3) Perform aerodynamic calculations on the divided computational grid file; Step 1.4) Calculate the four sonic conditions to obtain the lift coefficient and drag coefficient; Step 1.5) Normalize the lift-to-drag ratio in Step 1.4) based on a certain standard wide-speed domain airfoil (such as NPU_hyper_04), and determine whether two or more sonic conditions satisfy that the normalized lift-to-drag ratio is greater than or equal to 0.8. If satisfied, it is considered that the airfoil belongs to the wide-speed domain airfoil.

[0008] Step 2) Airfoil generation process: Step 2.1) Select airfoils that meet the wide-speed domain conditions from the existing airfoil library; Step 2.2) Perform CST interpolation on the airfoils that meet the wide-speed domain conditions as the training set to train the generative model wing-GAN: This generative model wing-GAN consists of a generator and a discriminator. When generating, there is a latent space and random noise as inputs. In the latent space, there are three variables, namely the leading-edge angle variable of the wing, the trailing-edge angle variable of the wing, and the thickness variable. These three variables change and combine continuously to generate airfoils. The random noise is used to fine-tune the geometry to prevent the generated airfoils from being too similar. During the generation process, in order to prevent the generated curve from being too strange, the constraint of the Bezier curve is specifically added to make the generated curve conform to the airfoil shape. In the discriminator, the real wide-speed domain airfoil training set and the generated fake wide-speed domain airfoils are used as inputs, continuously comparing the coordinate information of the real and fake airfoils, determining whether the generated fake wide-speed domain airfoils also conform to the airfoil characteristics of the real training set, giving a true / false judgment, and feeding it back to the generator. After receiving the feedback, the generator continues to generate; Step 2.3) Through training the generator and the discriminator, the generator and the discriminator play against each other. The generator continuously generates wide-speed domain airfoils for the discriminator, and the discriminator continuously feeds back to the generator until a balance is reached, obtaining a complete wide-speed domain airfoil model, and using the wide-speed domain airfoil model to output wide-speed domain airfoils that meet the conditions.

[0009] Step 1) The airfoil discrimination process For further improvement, in Step 1.1), the modeling process uses the 3D modeling software catia to model the airfoil.

[0010] For further improvement, in Step 1.2), the process of dividing the computational grid is to import the drawn curve graph into the computational grid drawing software pointwise for computational grid division.

[0011] For further improvement, in Step 1.3), the aerodynamic calculation process is to import the divided computational grid file into the CFD calculation software FLUENT and use this software to perform relevant aerodynamic calculations.

[0012] The beneficial effects of the present invention are as follows: 1. By training the generative model wing-GAN, more airfoils suitable for wide-speed-range aircraft can be obtained, with a fast generation speed and high accuracy, facilitating the quick finding of airfoils that meet the performance of the aircraft and greatly enhancing the combat performance of the aircraft.

[0013] 2. In the airfoil discrimination process, taking a certain reference airfoil such as NPU_hyper_04 as the benchmark, it can be determined whether any airfoil can have good aerodynamic performance at multiple speed ranges. BRIEF DESCRIPTION OF THE DRAWINGS

[0014] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the following will briefly introduce the drawings required in the embodiments. Obviously, the drawings in the following description are only some embodiments of the present invention. For those of ordinary skill in the art, without creative efforts, other drawings can also be obtained based on these drawings.

[0015] Figure 1 It is the airfoil diagram of NACA64A204.

[0016] Figure 2 It is the airfoil diagram of NACA16006.

[0017] Figure 3 It is the airfoil diagram of NACA64206.

[0018] Figure 4 It is the global mesh division diagram of NACA64A204.

[0019] Figure 5 It is the partial enlarged view of the mesh center of NACA64A204.

[0020] Figure 6 It is the example diagram of wide-speed-range airfoil discrimination.

[0021] Figure 7 It is the schematic diagram of the wing-GAN training model.

[0022] Figure 8 It is the wide-speed-range airfoil diagram generated by wing-GAN. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0023] The following will clearly and completely describe the technical solutions in the embodiments of the present invention with reference to the drawings in the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, rather than all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts belong to the scope of protection of the present invention.

[0024] A specific implementation manner of the present invention is as follows: 1. In the UIUC airfoil library, select airfoils for mesh generation. Here, three airfoils are selected as examples, namely NACA64A-201, NACA16006, and NACA64206; see Figures 1-5 .

[0025] 2. Perform CFD calculations on the three airfoils NACA64A-201, NACA16006, and NACA64206 with generated meshes to obtain their lift coefficients and drag coefficients.

[0026] 3. Normalize the three airfoils NACA64A-201, NACA16006, and NACA64206. It can be seen that at the low subsonic Mach number of 0.2 and the transonic Mach number of 0.8, the lift-to-drag ratio is greater than or equal to 0.8. That is, these three airfoils are all wide-speed airfoils. See Figure 6 .

[0027] 4. Use the selected wide-speed airfoils of the aircraft as the training set to train the wing-GAN model;.

[0028] 5. Train the generator to enable wing-GAN to learn a low-dimensional latent space, which contains the main shape change characteristics of wide-speed airfoils. Additionally, input a noise space to learn the secondary characteristics of wide-speed airfoils. Compared with the ordinary GAN model, wing-GAN adds a Bessel curve constraint, making the generated airfoils smoother and more continuous.

[0029] 6. Train the discriminator to determine whether the airfoils generated by the generator conform to the input characteristics and feedback to the generator.

[0030] 7. The generator and the discriminator play against each other. The generator continuously generates wide-speed airfoils for the discriminator, and the discriminator continuously feeds back to the generator until balance is achieved; (see the training schematic diagram of wing-GAN in Figure 7 ).

[0031] 8. After training the wing-GAN model, a large number of high-quality and fast wide-speed airfoil diagrams can be output. (See the generated wide-speed airfoil diagrams in Figure 8 ).

[0032] Each embodiment in this specification is described in a progressive manner. For the same or similar parts among the embodiments, reference can be made to each other, and the key point of each embodiment is to illustrate the differences from other embodiments. In particular, for the device embodiments, the above description is only the preferred embodiment of the present invention. Since it is basically similar to the method embodiments, the description is relatively simple, and reference can be made to the relevant parts of the method embodiments for the relevant content. As described above, these are only the specific embodiments of the present invention, but the protection scope of the present invention is not limited thereto. For any person skilled in the art within the technical scope disclosed by the present invention, for those of ordinary skill in the art in this technical field, any changes or substitutions that can be easily thought of without departing from the principle of the present invention should be covered within the protection scope of the present invention. Therefore, the protection scope of the present invention should be subject to the protection scope of the claims.

Claims

1. A discriminant and generation method for wide-speed-range airfoils based on a generative model, characterized in that: It includes an airfoil discrimination process and an airfoil generation process; Step 1) Airfoil discrimination process: Discriminate the airfoils from the existing airfoil library to find the airfoils that meet the wide-speed range conditions; Step 2) Airfoil generation process: Step 2.1) Select the airfoils that meet the wide-speed range conditions from the existing airfoil library; Step 2.2) Perform CST interpolation on the airfoils that meet the wide-speed range conditions above as the training set to train the generative model wing-GAN. This generative model wing-GAN consists of a generator and a discriminator. The specific generation process is as follows: Step 2.21) In the generator, use the latent space and random noise as inputs; the latent space includes three variables, the wing leading-edge angle variable, the wing trailing-edge angle variable, and the thickness variable. Generate airfoils by continuously changing and combining the three variables, and fine-tune the geometry with random noise; Step 2.22) During the generation process, the constraint of the Bezier curve is added to make the generated curve conform to the airfoil shape; Step 2.23) In the discriminator, use the real wide-speed range airfoil training set and the generated fake wide-speed range airfoils as inputs, continuously compare the coordinate information of the real and fake airfoils, judge whether the generated fake wide-speed range airfoils also conform to the airfoil characteristics of the real training set, give a true / false judgment, and feedback it to the generator. After receiving the feedback, the generator continues to generate; Step 2.3) By training the generator and the discriminator, the generator and the discriminator play against each other. The generator continuously generates wide-speed range airfoils for the discriminator, and the discriminator continuously feedbacks to the generator until a balance is reached, obtaining a complete wide-speed range airfoil model. Use the wide-speed range airfoil model to output the wide-speed range airfoils that meet the conditions.

2. The wide-speed-range airfoil discrimination and generation method based on a generative model according to claim 1, characterized in that: The specific process of step 1) the airfoil discrimination process is as follows: Step 1.1) Select several airfoils from the airfoil library, perform modeling and draw the curve graph; Step 1.2) Divide the drawn curve graph into computational grids and set boundary conditions; Step 1.3) Perform aerodynamic calculations on the divided computational grid files; Step 1.4) Calculate for four sonic conditions to obtain the lift coefficient, drag coefficient, and lift-to-drag ratio; Step 1.5) Normalize the lift-to-drag ratio in step 1.4) based on a certain standard wide-speed range airfoil, and judge whether two or more sonic conditions satisfy that the normalized lift-to-drag ratio is greater than or equal to 0.

8. If it is satisfied, it is considered that the airfoil belongs to the wide-speed range airfoil.

3. The wide-speed-range airfoil discrimination and generation method based on the generative model according to claim 1, characterized in that: In step 1.1), the modeling process uses the 3D modeling software catia to model the airfoil.

4. The wide-speed-range airfoil discrimination and generation method based on a generative model according to claim 1, characterized in that: In step 1.2), the process of dividing the computational grid is to import the drawn curve graph into the computational grid drawing software pointwise for computational grid division.

5. The wide-speed-range airfoil discrimination and generation method based on a generative model according to claim 1, characterized in that: In step 1.3), the aerodynamic calculation process is to import the divided computational grid files into the CFD calculation software FLUENT and use this software to perform relevant aerodynamic calculations.

Images