In the engineering practice of aircraft aerodynamic design, design specifications are typically proposed by the overall engineering team, while the aerodynamic design department implements the design goals through numerous iterations and extensive numerical simulations, a process that usually consumes significant resources. Generative models show the potential to directly generate design solutions that meet predefined goals, significantly reducing the iterative process in traditional design. This paper proposes a multi-objective generative airfoil design method based on diffusion models. By conditioning on multiple performance metrics such as buffet lift coefficient and cruise drag coefficient, the method generates airfoil designs that simultaneously satisfy these metrics. The use of conditional diffusion models to progressively generate effective airfoils in the design space avoids the complex iterative calculations of traditional optimization methods. Comparative experiments with conditional variational autoencoders demonstrate the advantages of diffusion models in terms of generation accuracy, stability, and diversity. The results indicate that diffusion models not only generate airfoils that meet performance requirements but also offer greater diversity and exploration capability in the design space, providing an efficient new approach for future airfoil design.
[1]韩忠华, 高正红, 宋文萍.翼型研究的历史, 现状与未来发展[J].空气动力学学报, 2021, 39(6):1-36
[2]Barrett T R, Bressloff N W, Keane A J.Airfoil shape design and optimization using multifidelity analysis and embedded inverse design[J].AIAA journal, 2006, 44(9):2051-2060
[3]Sun H.Wind turbine airfoil design using response surface method[J]. Journal of Mechanical Science and Technology, 2011, 25: 1335-1340.[J].Journal of Mechanical Science and Technology, 2011, 25:1335-1340
[4]Xia C-C, Jiang T-T, Chen W-F.Particle swarm optimization of aerodynamic shapes with nonuniform shape parameter–based radial basis function[J].Journal of Aerospace Engineering, 2017, 30(3):04016089-
[5] Lighthill M J.A new method of two-dimensional aerodynamic design[J], 1945.
[6]Takanashi S.Iterative three-dimensional transonic wing design using integral equations[J].Journal of aircraft, 1985, 22(8):655-660
[7]Bui-Thanh T, Damodaran M, Willcox K.Aerodynamic data reconstruction and inverse design using proper orthogonal decomposition[J].AIAA journal, 2004, 42(8):1505-1516
[8]白俊强, 邱亚松, 华俊.改进型 翼型反设计方法[J].航空学报, 2013, 34(04):762-771
[9] Obayashi S.Inverse Optimization Method for Aerodynamic Shape Design[J]. Vieweg+Teubner Verlag, 1999.
[10]Kharal A, Saleem A.Neural networks based airfoil generation for a given C p using Bezier–PARSEC parameterization[J].Aerospace Science and Technology, 2012, 23(1):330-344
[11] Sun G, Sun Y, Wang S.Artificial neural network based inverse design: Airfoils and wings[J]. Aerospace Science and Technology, 2015, 42: 415-428.
[12] Zhang Y, Sung W J, Mavris D N.Application of convolutional neural network to predict airfoil lift coefficient[C]. 2018 AIAA/ASCE/AHS/ASC structures, structural dynamics, and materials conference, 2018: 1903.
[13]Yang Y, Li R, Zhang Y, et al.Flowfield prediction of airfoil off-design conditions based on a modified variational autoencoder[J].AIAA Journal, 2022, 60(10):5805-5820
[14] Vinothkumarsekar, Mengqizhang, Changshu, et al.Inverse Design of Airfoil Using a Deep Convolutional Neural Network[J]. AIAA Journal, 2019, 57(3).
[15] Kingma D P, Mohamed S, Jimenez Rezende D, et al.Semi-supervised learning with deep generative models[J]. Advances in neural information processing systems, 2014, 27.
[16]Mirza M, Osindero S. Conditional generative adversarial nets[J]. arXiv prep.[J].rXiv:1411.1784, 2014., rint, :-
[17] Achour G, Sung W J, Pinon-Fischer O J, et al.Development of a conditional generative adversarial network for airfoil shape optimization[C]. AIAA Scitech 2020 Forum, 2020: 2261.
[18] Bertrand X, Tost F, Champagneux S.Wing airfoil pressure calibration with deep learning[C]. AIAA Aviation 2019 Forum, 2019: 3066.
[19] Du X, He P, Martins J R.A B-spline-based generative adversarial network model for fast interactive airfoil aerodynamic optimization[C]. AIAA scitech 2020 forum, 2020: 2128.
[20]田洁华，孙迪，屈峰，等.基于?的翼型参数化方法[J].航空学报, 2023, 44(18):128280-
[21]Chen W, Fuge M. BezierGAN: Automatic Generation of Smooth Curves from Interpretable Low-Dimensional Parameters[J]. arXiv prep.[J].rXiv:1808.08871, 2018., rint, :-
[22] Heyrani N A, Chen W, Ahmed F.Range-GAN: Range-constrained generative adversarial network for conditioned design synthesis[C]//International Design Engineering Technical Conferences and Computers and Information in Engineering Conference: Vol 85390. American Society of Mechanical Engineers, 2021: V03BT03A039.
[23] XIE H, WANG J, ZHANG M.Parametric generative schemes with geometric constraints for encoding and synthesizing airfoils[J]. Engineering Applications of Artificial Intelligence, 2024, 128: 107505.
[24]Wang J, LI R, HE, et al.An inverse design method for supercritical airfoil based on conditional generative models[J].Chinese Journal of Aeronautics, 2022, 35(3):62-74
[25]Liu J, Wu J, Xie H, et al. AFBench: A Large-scale Benchmark for Airfoil Design[J]. arXiv prep.[J].rXiv:2406.18846, 2024., rint, :-
[26]GRAVES R, FARIMANI A B. Airfoil diffusion: denoising diffusion model for conditional airfoil generat.[J].rXiv:2408.15898[Z/OL] 2024., ion:, :-
[27] Li R, Zhang Y, Chen H.Pressure distribution feature-oriented sampling for statistical analysis of supercritical airfoil aerodynamics[J]. Chinese Journal of Aeronautics, 2022, 35(4), 134-147.
[28] Ho, J., Jain, A., & Abbeel, P.. Denoising diffusion probabilistic models[J]. Advances in neural information processing systems, 2020, 33, 6840-6851.
[29] Van der Maaten L, Hinton G.Visualizing data using t-SNE[J]. Journal of machine learning research, 2008, 9(11).
[30]Jonathan H, Salimans T. Classifier-free diffusion guidance. arXiv prep.[J].rXiv:2207.12598 (2022)., rint, :-
[31] Xie H, Wang J, Zhang M .Knowledge-embedded meta-learning model for lift coefficient prediction of airfoils[J].Expert Systems With Applications, 2023, 233
CJA