航空学报 > 2023, Vol. 44 Issue (6): 126962-126962   doi: 10.7527/S1000-6893.2022.26962

基于新型多可信度代理模型的多目标优化方法

赵欢, 高正红, 夏露()   

  1. 西北工业大学 航空学院,西安 710072
  • 收稿日期:2022-01-18 修回日期:2022-04-24 接受日期:2022-05-10 出版日期:2022-05-20 发布日期:2022-05-19
  • 通讯作者: 夏露 E-mail:xialu@nwpu.edu.cn
  • 基金资助:
    国家自然科学基金(12102489);翼型、叶栅空气动力学重点实验室基金(614220121010126)

Novel multi-fidelity surrogate model assisted many-objective optimization method

Huan ZHAO, Zhenghong GAO, Lu XIA()   

  1. School of Aeronautics,Northwestern Polytechnical University,Xi’an,710072,China
  • Received:2022-01-18 Revised:2022-04-24 Accepted:2022-05-10 Online:2022-05-20 Published:2022-05-19
  • Contact: Lu XIA E-mail:xialu@nwpu.edu.cn
  • Supported by:
    National Natural Science Foundation of China(12102489);National Key Laboratory of Science and Technology on Aerodynamic Design and Research(614220121010126)

摘要:

相对于常规直升机,基于前行桨叶方案(ABC)速度更高的刚性共轴双旋翼直升机对旋翼翼型的气动性能要求更加苛刻,包括更高马赫数下的低阻和高阻力发散特性,并维持良好的中、低马赫数下的升阻特性以及所有状态下的力矩特性等10余项指标要求,面临高维多目标优化难题。针对该类问题,首先发展了基于核主成分分析(KPCA)的非线性目标降维技术,然后建立了基于新型多可信度多项式混沌-Kriging(AMF-PCK)代理模型的高效多目标稳健优化方法,并提出了变可信度伪期望改进矩阵(VF-PEIM)并行样本填充方法,使多目标优化设计效率和能力显著提高。利用所建立的基于新型多可信度代理模型的多目标优化方法对高速直升机7%厚度旋翼翼型进行了优化设计。设计翼型的气动表现分别与经典的OA407翼型在高、中、低马赫数下进行了全面比较,对比结果验证了提出的新型多目标优化设计方法的有效性,设计翼型高速气动特性得到了显著提升。

关键词: 多目标全局优化, 旋翼翼型, 多可信度代理模型, 非线性降维, 变可信度伪期望改进矩阵

Abstract:

Compared with a conventional helicopter, a compound helicopter with its rigid coaxial rotor, famous as an Advancing Blade Concept (ABC) rotor, has more stringent requirements for the aerodynamic performance of the rotor airfoil. And the 10 more index requirements, e.g., the low-drag and high drag-divergence characteristics at higher Mach numbers, the high lift-drag characteristics at medium and low Mach numbers, and good pitching moment characteristics at all conditions, face the problem of many-objective optimization. To solve this issue, this paper first develops a novel nonlinear dimension-reduction technology based on Kernel Principal Component Analysis (KPCA), then establishes an efficient many-objective robust optimization framework based on a novel Adaptive Multi-Fidelity Polynomial Chaos-Kriging (AMF-PCK) surrogate model. Moreover, the paper proposes a Variable-Fidelity Pseudo Expected Improvement Matrix (VF-PEIM) parallel in-filling method, which significantly improves the efficiency and ability of many-objective optimization. The novel AMF-PCK assisted multi-objective optimization method is used to optimize the 7% thickness rotor airfoil of such compound helicopter. The aerodynamic performances of the designed airfoils are compared with those of the classical OA407 airfoil at high, medium, and low Mach numbers comprehensively. Results demonstrate the effectiveness of the proposed many-objective global optimization method and a significant improvement of high-speed aerodynamic characteristics of the designed rotor airfoils.

Key words: many-objective global optimization, rotor airfoil, multi-fidelity surrogate model, nonlinear dimension-reduction, variable-fidelity pseudo expected improvement matrix

中图分类号: