航空学报 > 2014, Vol. 35 Issue (11): 2949-2957   doi: 10.7527/S1000-6893.2014.0086

空气动力二阶核函数辨识方法

王云海1, 韩景龙1, 张兵2, 员海玮1   

  1. 1. 南京航空航天大学 机械结构力学与控制国家重点实验室, 江苏 南京 210016;
    2. 合肥工业大学 机械与汽车工程学院, 安徽 合肥 230009
  • 收稿日期:2014-02-17 修回日期:2014-05-05 出版日期:2014-11-25 发布日期:2014-05-12
  • 通讯作者: 韩景龙,Tel.: 025-84896484 E-mail: hjlae@nuaa.edu.cn E-mail:hjlae@nuaa.edu.cn
  • 作者简介:王云海 男, 博士研究生.主要研究方向: 飞行器气动弹性力学, 系统辨识. Tel: 025-89846484 E-mail: yantaicity@163.com; 韩景龙 男, 博士, 教授, 博士生导师.主要研究方向: 气动弹性力学, 非线性动力学, 复杂结构动力学. Tel: 025-89846484 E-mail: hjlae@nuaa.edu.cn; 张兵 男, 博士, 讲师.主要研究方向: 流固耦合力学. Tel: 025-89846484 E-mail: zhangbing_end@163.com; 员海玮 男, 博士, 副教授.主要研究方向: 飞行器气动弹性力学. Tel: 025-89846484 E-mail: zhangbing_end@163.com
  • 基金资助:

    国家自然科学基金(11102085); 江苏高校优势学科建设工程资助项目

Identification Method of Second-order Kernels in Aerodynamics

WANG Yunhai1, HAN Jinglong1, ZHANG Bing2, Yuan Haiwei1   

  1. 1. State Key Lab of Mechanics and Control for Mechanical Structures, Nanjing University of Aeronautics and Astronautics, Nanjing 210016, China;
    2. School of Mechanical and Automotive Engineering, Hefei University of Technology, Hefei 230009, China
  • Received:2014-02-17 Revised:2014-05-05 Online:2014-11-25 Published:2014-05-12
  • Supported by:

    National Natural Science Foundation of China (11102085); Project Funded by the Priority Academic Program Development of Jiangsu Higher Education Institutions

摘要:

采用截断三阶Volterra级数模型来研究空气动力二阶核函数的辨识问题,选取一簇正交化的切比雪夫多项式对二阶核函数进行参数化处理,并将非参数辨识问题转化成参数辨识问题.相比于其他方法,本文模型能有效降低对激励信号幅值的敏感程度,保证辨识出的核函数具有较好的保真度;只针对三阶Volterra降阶模型中的一阶、二阶核函数进行辨识,即可提升原系统一阶、二阶核函数的辨识精度,却不会显著增加辨识过程的工作量;参数化辨识方法属于整体性辨识,根据已有的部分数据对(输入、输出数据)就能完成系统辨识,且能达到较好的辨识精度,从而有效地减少了执行计算流体力学(CFD)代码程序的总次数,节约了大量的时间成本.算例表明,与目前流行的非参数化方法相比,本文提出的切比雪夫函数辨识方法,精度上达到预期要求,辨识过程消耗的CFD总时间量至少可降低一个数量级.

关键词: 系统辨识, Volterra 级数, 降阶模型, 切比雪夫展式方法, 参数化方法

Abstract:

Based on the truncated third-order Volterra series model, this paper investigates the identification of second-order Volterra kernels in aerodynamics. In this model, the second-order Volterra kernels can be parameter treated by employing a cluster of Chebyshev orthogonal polynomials. And then the problem of non-parameter is converted to the problem of parameter. In comparison with other treatments, it has several advantages. First, it ensures that the second-order Volterra kernels can be identified with better fidelity due to the reduced signal amplitude sensitivity. Then, in this model, the accuracy of first- and second-order Volterra kernels can be improved and the third-order Volterra kernels do not need to be identified. That means the cost for identification is reasonable. Last, the proposed parametric methods treat the identification as a whole and the partial data is available to be used for identification. That means the total number of executing computational fluid dynamics (CFD) code can be reduced and the identification cost is highly time-saving. The numerical results show the efficiency and high accuracy compared to those non-parameter methods and the total amount of CFD time can be reduced by at least one order of magnitude.

Key words: system identification, Volterra series, reduced-order model, Chebyshev expansion method, parametric approach

中图分类号: