上海交通大学学报

上海交通大学学报 >

2025 , Vol. 59 >Issue 5: 684 - 690

DOI: https://doi.org/10.16183/j.cnki.jsjtu.2023.213

机械与动力工程

数据驱动的流场稀疏数据建模

  • 王鸿鑫 ,
  • 徐德刚 ,
  • 周楷文 ,
  • 李林文 ,
  • 温新
展开
  • 1.中南大学 自动化学院,长沙 410083
    2.上海飞机设计研究院,上海 201210
    3.上海交通大学 机械与动力工程学院,上海 200240
王鸿鑫(1979—),正高级工程师,从事民用飞机设计集成、试验验证、建模仿真及数字孪生研究.
徐德刚,教授,博士生导师;E-mail:dgxu@csu.edu.cn.

收稿日期: 2023-05-29

  修回日期: 2023-07-04

  录用日期: 2023-08-22

  网络出版日期: 2023-09-11

收起

Data-Driven Method of Modeling Sparse Flow Field Data

  • WANG Hongxin ,
  • XU Degang ,
  • ZHOU Kaiwen ,
  • LI Linwen ,
  • WEN Xin
Expand
  • 1. School of Automation, Central South University, Changsha 410083, China
    2. Shanghai Aircraft Design and Research Institute, Shanghai 201210, China
    3. School of Mechanical and Power Engineering, Shanghai Jiao Tong University, Shanghai 200240, China

Received date: 2023-05-29

  Revised date: 2023-07-04

  Accepted date: 2023-08-22

  Online published: 2023-09-11

Fold

摘要

流场实时感知与预测在航空航海等领域具有十分重要的应用价值,但是往往面临流场维度高和实时测量信息少等挑战.针对该问题,提出一种数据驱动的流场建模方法,通过线下建立稀疏数据与高维流场映射,实现线上流场实时重构.线下建模中,针对流场高维度挑战,使用本征正交分解等方法对数据进行降维,提取主要流场空间模态.采用正交三角(QR)分解方法,挖掘流场模态敏感性特征,优化测点位置.利用时间延迟的动态模态分解,显著降低测点数量.在线上重构中,基于实时稀疏测量数据与数据驱动模型,实现对当前和未来时刻全场流场的预测.在圆柱尾涡流场测试中,使用该方法并采用20个稀疏测点,得到的全场重构误差可达10%以下.

关键词： 数据驱动; 降阶模型; 动态模态分解; 位置优化; 稀疏数据; 流场重构

本文引用格式

王鸿鑫 , 徐德刚 , 周楷文 , 李林文 , 温新 . 数据驱动的流场稀疏数据建模[J]. 上海交通大学学报, 2025 , 59(5) : 684 -690 . DOI: 10.16183/j.cnki.jsjtu.2023.213

Abstract

Real-time perception and prediction of flow field have very important application value in aviation and navigation, and pose challenges such as high flow field dimension and less real-time measurement information. To solve such problem, a data-driven flow field modeling method framework is proposed, which realizes real-time reconstruction of online flow field by establishing sparse data and high-dimensional flow field mapping offline. In offline modeling, aimed at the high-dimensional challenge of the flow field, the eigenortho decomposition and other methods are used to reduce the dimensionality of the data and extract the spatial mode of the main flow field. The QR decomposition method is used to mine the modal sensitivity characteristics of the flow field and optimize the measurement point position. Dynamic modal decomposition with time delay significantly reduces the number of measurement points. In the online reconstruction, based on real-time sparse measurement data and data-driven models, the prediction of the current and future full-field flow field is realized. In the test of cylinder wake flow, using this method and using 20 sparse measurement points, the full-field reconstruction error obtained can reach less than 10%.

Key words： data-driven; reduced order model; dynamic mode decomposition; location optimization; sparse data; flow field reconstruction

参考文献

[1] VOGEL J M, KELKAR A G. Aircraft control augmentation and health monitoring using flush air data system feedback[C]// AIAA Applied Aerodynamics Conference. Hawaii, USA: AIAA, 2008: 1-16.
[2] TRIANTAFYLLOU M S, WEYMOUTH G D, MIAO J. Biomimetic survival hydrodynamics and flow sensing[J]. Annual Review of Fluid Mechanics, 2016, 48: 1-24.
[3] LIEW J, G??MEN T, LIO W H, et al. Streaming dynamic mode decomposition for short-term forecasting in wind farms[J]. Wind Energy, 2022, 25(4): 719-734.
[4] WILLCOX K. Unsteady flow sensing and estimation via the gappy proper orthogonal decomposition[J]. Computers and Fluids, 2006, 35(2): 208-226.
[5] MANOHAR K, BRUNTON B W, KUTZ J N, et al. Data-driven sparse sensor placement for reconstruction: Demonstrating the benefits of exploiting known patterns[J]. IEEE Control Systems, 2018, 38(3): 63-86.
[6] ZHOU K, ZHOU L, ZHAO S, et al. Data-driven method for flow sensing of aerodynamic parameters using distributed pressure measurements[J]. AIAA Journal, 2021, 59(9): 3504-3516.
[7] DENG Z, CHEN Y, LIU Y, et al. Time-resolved turbulent velocity field reconstruction using a long short-term memory (LSTM)[J]. Physics of Fluids, 2019, 31(075108): 1-13.
[8] SCHMID P J. Dynamic mode decomposition of numerical and experimental data[J]. Journal of Fluid Mechanics, 2010, 656: 5-28.
[9] 刘余丹, 周楷文, 刘应征, 等. 基于卡尔曼滤波的翼型表面压力实时重构方法[J]. 空气动力学学报, 2022, 41(4): 64-72.
  LIU Yudan, ZHOU Kaiwen, LIU Yingzheng, et al. Reconstruction of airfoil surface pressure by Kalman filter[J]. Acta Aerodynamica Sinica, 2022, 41(4): 64-72.
[10] WILLIAMS M O, KEVREKIDIS I G, ROWLEY C W. A data-driven approximation of the Koopman operator: Extending dynamic mode decomposition[J]. Journal of Nonlinear Science, 2015, 25(6): 1307-1346.
[11] WILLIAMS M O, ROWLEY C W, KEVREKIDIS I G. A kernel-based method for data-driven koopman spectral analysis[J]. Journal of Computational Dynamics, 2015, 2(2): 247-265.
[12] YUAN Y, ZHOU K, ZHOU W, et al. Flow prediction using dynamic mode decomposition with time-delay embedding based on local measurement[J]. Physics of Fluids, 2021, 33(9): 1-18.
[13] SHI Z Y, ZHOU K W, QIN C, et al. Experimental study of dynamical airfoil and aerodynamic prediction[J]. Actuators, 2022, 11(2): 1-14.
[14] BERKOOZ G, HOLMES P, LUMLEY J L. The proper orthogonal decomposition in the analysis of turbulent flows[J]. Annual Review of Fluid Mechanics, 1993(25): 539-575.
[15] SARLóS T. Improved approximation algorithms for large matrices via random projections[C]// 2006 47th Annual IEEE Symposium on Foundations of Computer Science. Berkeley, USA: IEEE, 2006:143-152.
[16] MARTINSSON P G, ROKHLIN V, TYGERT M. A randomized algorithm for the decomposition of matrices[J]. Applied and Computational Harmonic Analysis, 2011, 30(1): 47-68.
Options
文章导航

/

〈 〉