利用长短期记忆神经网络的改进POD-Galerkin降阶模型及其在流场预测中的应用

张译文; 王志恒; 邱睿贤; 席光

doi:10.7652/xjtuxb202402002

您当前的位置：

首页 >

文章列表页 >

利用长短期记忆神经网络的改进POD-Galerkin降阶模型及其在流场预测中的应用

更新时间：2025-05-13

- 利用长短期记忆神经网络的改进POD-Galerkin降阶模型及其在流场预测中的应用
- Improved POD-Galerkin Reduced Order Model with Long Short-Term Memory Neural Network and Its Application in Flow Field Prediction
- 西安交通大学学报 2024年58卷第2期页码：12-21
- 作者机构：
  
  西安交通大学能源与动力工程学院,西安,710049
- 作者简介：
  
  : 2023-06-25。作者简介: 张译文(1998—),男,硕士生
  王志恒(通信作者),男,副教授,博士生导师。基金项目: 国家重大科技专项资助项目(2017-Ⅱ-0004-0016)
- 基金信息：
  
  : 2023-06-25。作者简介: 张译文(1998—),男,硕士生;王志恒(通信作者),男,副教授,博士生导师。基金项目: 国家重大科技专项资助项目(2017-Ⅱ-0004-0016);国家自然科学基金资助项目(52176044)。
- DOI：10.7652/xjtuxb202402002
  中图分类号： O357.1
- 网络首发：2024-02-10，
  
  纸质出版：2024
- 稿件说明：
移动端阅览
张译文, 王志恒, 邱睿贤, 等. 利用长短期记忆神经网络的改进POD-Galerkin降阶模型及其在流场预测中的应用[J]. 西安交通大学学报, 2024,58(2):12-21.

Improved POD-Galerkin Reduced Order Model with Long Short-Term Memory Neural Network and Its Application in Flow Field Prediction[J]. 2024, 58(2): 12-21.
张译文, 王志恒, 邱睿贤, 等. 利用长短期记忆神经网络的改进POD-Galerkin降阶模型及其在流场预测中的应用[J]. 西安交通大学学报, 2024,58(2):12-21. DOI： 10.7652/xjtuxb202402002.

Improved POD-Galerkin Reduced Order Model with Long Short-Term Memory Neural Network and Its Application in Flow Field Prediction[J]. 2024, 58(2): 12-21. DOI： 10.7652/xjtuxb202402002.

摘要

针对标准POD-Galerkin降阶模型在流场快速预测中存在误差而导致精度不高的问题

提出了一种利用长短期记忆神经网络的改进POD-Galerkin降阶模型。使用本征正交分解对流场进行降维

投影得到低维降阶模型

引入两个长短期记忆神经网络

建立从POD-Galerkin降阶模型到实际POD模态时间系数之间的修正映射、低阶模态时间系数与高阶模态时间系数之间的扩展映射

分别用于消除标准POD-Galerkin降阶模型的误差累积和扩展降阶模型的阶数

从而实现物理驱动与数据驱动混合的流动降阶模型的构建。将改进POD-Galerkin降阶模型应用于二维圆柱绕流的流场预测

通过与原始标准POD-Galerkin降阶模型的对比

分析了所提模型的精度和计算速度。结果表明:添加神经网络修正项后的降阶模型相较于标准POD-Galerkin降阶模型

有效提升了降阶模型的精度

预测各阶模态时间系数的均方根误差能够减小1～2个数量级

预测的流场更接近原始流场; 在预测相同阶数的情况下

计算时间显著减小

基于4阶和6阶扩展的8阶改进降阶模型相较于原始8阶POD-Galerkin降阶模型预测速度分别提高了约56%和25%。

Abstract

Aiming at the issue of low accuracy in rapid prediction of flow field caused by errors in the standard POD-Galerkin reduced order model

an improved POD-Galerkin model using long short-term memory(LSTM)neural networks is proposed. In this method

based on the reduced order model obtained by dimensionality reduction and projection of the flow field using proper orthogonal decomposition(POD)

two LSTM neural networks are introduced to establish a corrective mapping from the POD-Galerkin model to the actual POD coefficients

and an expansion mapping between the low-order mode time coefficients and the high-order mode time coefficients

so as to eliminate the error accumulation of the standard POD-Galerkin model and extend the order of reduced order model. This method enables the construction of a hybrid reduced order model that combines physics-driven and data-driven approaches. The improved POD-Galerkin model is applied to flow field prediction of a two-dimensional flow around a cylinder. A comparison with the original standard POD-Galerkin model is performed to analyze the model accuracy and computational speed. The results show that

with the addition of neural network correction terms

the accuracy of the reduced order model is effectively improved compared with the standard POD-Galerkin model. The root mean square error of the predicted mode time coefficients is reduced by one to two orders of magnitude compared with the original model

and the predicted flow field is closer to the original flow field. The improved model achieves a significant reduction in computation time while predicting the same order. The improved 8-order improved reduced order model based on 4- and 6-order expansions improves the prediction speed by approximately 56% and 25%

respectively

compared with the original 8-order POD-Galerkin model.

关键词

Keywords

references

DOWELL E H. Eigenmode analysis in unsteady aerodynamics: reduced-order models [J]. AIAA Journal, 1996, 34(8): 1578-1583

VENEZIANI A. Certified reduced basis methods for parametrized partial differential equations [J]. SIAM Review, 2017, 59(1): 219-223

WYNN A, PEARSON D S, GANAPATHISUBRAMANI B, et al. Optimal mode decomposition for unsteady flows [J]. Journal of Fluid Mechanics, 2013, 733: 473-503

康伟, 张家忠, 李凯伦. 利用本征正交分解的非线性Galerkin降维方法 [J]. 西安交通大学学报, 2011, 45(11): 58-62, 67.KANG Wei, ZHANG Jiazhong, LI Kailun. Nonlinear Galerkin method for dimension reduction using proper orthogonal decomposition [J]. Journal of Xi'an Jiaotong University, 2011, 45(11): 58-62, 67

AUBRY N, HOLMES P, LUMLEY J L, et al. The dynamics of coherent structures in the wall region of a turbulent boundary layer [J]. Journal of Fluid Mechanics, 1988, 192: 115-173

DEANE A E, KEVREKIDIS I G, KARNIADAKIS G E, et al. Low-dimensional models for complex geometry flows: application to grooved channels and circular cylinders [J]. Physics of Fluids: A Fluid Dynamics, 1991, 3(10): 2337-2354

NOACK B R, AFANASIEV K, MORZYN' SKI M, et al. A hierarchy of low-dimensional models for the transient and post-transient cylinder wake [J]. Journal of Fluid Mechanics, 2003, 497: 335-363

SCHLEGEL M, NOACK B R. On long-term boundedness of Galerkin models [J]. Journal of Fluid Mechanics, 2015, 765: 325-352

张鸿志, 周强, 陈刚, 等. 气动弹性系统本征正交分解降阶模型精度的参数影响研究 [J]. 西安交通大学学报, 2016, 50(11): 104-109.ZHANG Hongzhi, ZHOU Qiang, CHEN Gang, et al. Effects of some parameters on the accuracy of aeroelastic proper orthogonal decomposition reduced order model [J]. Journal of Xi'an Jiaotong University, 2016, 50(11): 104-109

李静, 张伟伟, 李新涛. 失稳初期的低雷诺数圆柱绕流POD-Galerkin 建模方法研究 [J]. 西北工业大学学报, 2015, 33(4): 596-602.LI Jing, ZHANG Weiwei, LI Xintao. Researching method of building flow field of initial development stage of low Reynolds number flow past a circular cylinder [J]. Journal of Northwestern Polytechnical University,2015, 33(4): 596-602

冯俞楷. POD降维算法在传热与流动数值模拟中的应用 [D]. 北京: 华北电力大学(北京), 2017

梁钰. 基于自由单元法的瞬态非线性热传导问题模型降阶及参数辨识方法研究 [D]. 大连: 大连理工大学, 2021

KING R, BECKER R, FEUERBACH G, et al. Adaptive flow control using slope seeking [C]//2006 14th Mediterranean Conference on Control and Automation. Piscataway, NJ, USA: IEEE, 2006: 1-6

TIMME S, MARQUES S, BADCOCK K J. Transonic aeroelastic stability analysis using a Kriging-based Schur complement formulation [J]. AIAA Journal, 2011, 49(6): 1202-1213

LUI H F S, WOLF W R. Construction of reduced-order models for fluid flows using deep feedforward neural networks [J]. Journal of Fluid Mechanics, 2019, 872: 963-994

BRUNTON S L, PROCTOR J L, KUTZ J N. Discovering governing equations from data by sparse identification of nonlinear dynamical systems [J]. Proceedings of the National Academy of Sciences of the United States of America, 2016, 113(15): 3932-3937

AHMED S E, RAHMAN S M, SAN O, et al. Memory embedded non-intrusive reduced order modeling of non-ergodic flows [J]. Physics of Fluids, 2019, 31(12): 126602

BHATNAGAR S, AFSHAR Y, PAN Shaowu, et al. Prediction of aerodynamic flow fields using convolutional neural networks [J]. Computational Mechanics, 2019, 64(2): 525-545

SAEED A, FAROOQ H, AKHTAR I, et al. Deep learning-based reduced-order model for turbulent flows [C]//2022 19th International Bhurban Conference on Applied Sciences and Technology(IBCAST). Piscataway, NJ, USA: IEEE, 2022: 821-829

BUKKA S R, GUPTA R, MAGEE A R, et al. Assessment of unsteady flow predictions using hybrid deep learning based reduced-order models [J]. Physics of Fluids, 2021, 33(1): 013601

SHAMAI M, DAWSON S T M, MEZIC I, et al. Unsteady dynamics in the streamwise-oscillating cylinder wake for forcing frequencies below lock-on [J]. Physical Review Fluids, 2021, 6(7): 074702

MENDONÇA G, AFONSO F, LAU F. Model order reduction in aerodynamics: review and applications [J]. Proceedings of the Institution of Mechanical Engineers: Part G Journal of Aerospace Engineering, 2019, 233(15): 5816-5836

LASSILA T, MANZONI A, QUARTERONI A, et al. Model order reduction in fluid dynamics: challenges and perspectives [M]//QUARTERONI A, ROZZA G. Reduced Order Methods for Modeling and Computational Reduction. Cham, Germany: Springer International Publishing, 2014: 235-273

RAHMAN S M, AHMED S E, SAN O. A dynamic closure modeling framework for model order reduction of geophysical flows [J]. Physics of Fluids, 2019, 31(4): 046602

KANI J N, ELSHEIKH A H. Reduced-order modeling of subsurface multi-phase flow models using deep residual recurrent neural networks [J]. Transport in Porous Media, 2019, 126(3): 713-741

BRUNTON S L, NOACK B R, KOUMOUTSAKOS P. Machine learning for fluid mechanics [J]. Annual Review of Fluid Mechanics, 2020, 52(1): 477-508

LUMLEY J L. Stochastic tools in turbulence [M]. Burlington, VT, USA: Elsevier Science, 1970

PAWAR S, SAN O, NAIR A, et al. Model fusion with physics-guided machine learning: projection-based reduced-order modeling [J]. Physics of Fluids, 2021, 33(6): 067123

PARK J, KWON K, CHOI H. Numerical solutions of flow past a circular cylinder at Reynolds numbers up to 160 [J]. KSME International Journal, 1998, 12(6): 1200-1205

SADOVNICHII V A, DUBROVSKII V V, KADCHENKO S I, et al. The calculation of eigenvalues of problem for hydrodynamic stability of current of viscous fluid between two rotational cylinders with little numbers of Reinolds [J]. Doklady Akademii Nauk SSSR, 1998, 363(6): 748-750

ZHANG Chaoying, SHI Juan, TAN Huili, et al. Sedimentation of a single charged elliptic cylinder in a Newtonian fluid by lattice Boltzmann method [J]. Chinese Physics Letters, 2004, 21(6): 1108

MITTAL S. Excitation of shear layer instability in flow past a cylinder at low Reynolds number [J]. International Journal for Numerical Methods in Fluids, 2005, 49(10): 1147-1167

STÅLBERG E, BRÜGER A, LÖTSTEDT P, et al. High order accurate solution of flow past a circular cylinder [J]. Journal of Scientific Computing, 2006, 27(1): 431-441

POSDZIECH O, GRUNDMANN R. A systematic approach to the numerical calculation of fundamental quantities of the two-dimensional flow over a circular cylinder [J]. Journal of Fluids and Structures, 2007, 23(3): 479-499.

浏览量

下载量

CSCD

文章被引用时，请邮件提醒。

提交

工具集

关联资源

气动弹性系统本征正交分解降阶模型精度的参数影响研究

相变冷却换热器动态热响应降阶预测方法研究

采用动力学模态分解和长短期记忆网络的压缩空气储气装置流场动态预测方法

新能源汽车热管理系统协同优化策略

采用偏心端头多几何参数耦合的商用质子交换膜燃料电池电堆阴极气体分配优化