Volume 40, Number 4, August 2022
|Page(s)||837 - 844|
|Published online||30 September 2022|
Inverse design of aerodynamic configuration using generative topographic mapping
Computational Aerodynamics Institute, China Aerodynamics Research and Development Center, Mianyang 621000, China
The inverse design method of aerodynamic configuration is hard to give a reasonable pressure distribution, and strongly rely on experience of designers. The method has been difficult to adapt to the needs of modern aircraft design. Aiming at the shortcoming of the method, an efficient and robust aerodynamic configuration inverse design method is developed, employing knowledge of machine learning methods and optimization methods. The present method establishes the mapping between the high dimensional data obtained from aerodynamic shape and pressure distribution and the variables in the latent space. Then, the global optimization is carried out in the latent space by using the genetic algorithm. The optimum pressure distribution and the corresponding shape can be obtained. Through the the GTM model with high precision, there is not necessary for the flow solver in the whole iterative process, thus the design efficiency can be enhanced. Besides, by taking the advantage of optimization method, the target pressure distribution can be given in a very flexible way, and does not need to be physically meaningful. This feature can reduce reliance on the design experience. Airfoils in low speed and transonic flow and a three-dimensional laminar nacelle design cases are carried out. It is shown that the method robustly and efficiently converges to the target pressure, and has good engineering application potential.
气动外形反设计方法存在难以给定合理目标压力分布, 强烈依赖设计经验等问题, 难以适应现代工程设计需求。针对反设计方法的不足, 结合机器学习与全局优化方法, 发展了一种高效鲁棒的气动外形反设计方法。该方法利用生成拓扑映射(generative topographic mapping, GTM)模型建立气动外形及其压力分布组合数据和低维隐空间变量的映射关系, 并利用遗传算法在隐空间寻优, 同时得到最优压力分布与对应气动外形。GTM能够建立高精度的映射关系, 设计过程不需要流场解算参与迭代, 极大提高了设计效率。所提方法充分利用了遗传算法与GTM方法的特点, 不要求目标压力分布具有实际物理意义, 减小了对设计经验的依赖程度。分别针对低速翼型、跨声速翼型、三维层流短舱开展了反设计研究, 设计算例表明, 该方法设计鲁棒性好, 能够高效率收敛到目标压力分布, 具有良好的工程应用价值。
Key words: inverse design / generative topographic mapping / machine learning / genetic algorithm / pressure distribution
关键字 : 反设计 / 生成拓扑映射 / 机器学习 / 遗传算法 / 压力分布
© 2022 Journal of Northwestern Polytechnical University. All rights reserved.
This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/4.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
Current usage metrics show cumulative count of Article Views (full-text article views including HTML views, PDF and ePub downloads, according to the available data) and Abstracts Views on Vision4Press platform.
Data correspond to usage on the plateform after 2015. The current usage metrics is available 48-96 hours after online publication and is updated daily on week days.
Initial download of the metrics may take a while.