自动编码器在流场降阶中的应用

叶舒然; 张珍; 宋旭东; 杜特专; 王一伟; 黄晨光; 陈耀松

doi:10.7638/kqdlxxb-2019.0039

欢迎来到能源电力期刊网！

English Version 期刊登录

您当前的位置：

首页 >

文章列表页 >

自动编码器在流场降阶中的应用

专栏——智能空气动力学研究探索 | 更新时间：2026-05-28

- 自动编码器在流场降阶中的应用
- Applications of autoencoder in reduced-order modeling of flow field
- 空气动力学学报 2019年37卷第3期页码：498-504
- 作者机构：
  
  1. 中国科学院力学研究所流固耦合系统力学重点实验室,北京,100190
  2. 中国科学院大学工程科学学院,北京,100049
  3. 北京大学工学院,北京,100871
- 作者简介：
  
  叶舒然(1995-), 女, 安徽合肥人, 硕士研究生, 研究方向:水动力学.E-mail:yeshuran@imech.ac.cn
  王一伟(1983-), 男, 研究员, 研究方向:水动力学.E-mail:wangyw@imech.ac.cn
- 基金信息：
  
  国家重点研发计划（2016YFC0300600，2016YFC0301601）
- DOI：10.7638/kqdlxxb-2019.0039
  中图分类号： TP18;V211.3
- 纸质出版：2019
- 稿件说明：
移动端阅览
叶舒然, 张珍, 宋旭东, 杜特专, 王一伟, 黄晨光, 陈耀松. 自动编码器在流场降阶中的应用[J]. 空气动力学学报, 2019, 37(3): 498-504.

YE Shuran, ZHANG Zhen, SONG Xudong, et al. Applications of autoencoder in reduced-order modeling of flow field[J]. 2019, 37(3): 498-504.
叶舒然, 张珍, 宋旭东, 杜特专, 王一伟, 黄晨光, 陈耀松. 自动编码器在流场降阶中的应用[J]. 空气动力学学报, 2019, 37(3): 498-504. DOI： 10.7638/kqdlxxb-2019.0039.

YE Shuran, ZHANG Zhen, SONG Xudong, et al. Applications of autoencoder in reduced-order modeling of flow field[J]. 2019, 37(3): 498-504. DOI： 10.7638/kqdlxxb-2019.0039.

摘要

自动编码器作为一种压缩算法，在数据降维和去噪等方面有着广泛实践，有条件作为一种降阶方法在流场识别与数据处理方面得到应用。文章中以圆柱绕流为例，首先对圆柱后速度场建立了编码模型，用来对原始数据进行降维和特征提取，之后将编码后的数据与流场特征量相关联，建立了由流场编码回归圆柱表面压力系数的神经网络，探索了降维后数据的应用。结果表明，自动编码得到的结果能够承载原始速度场的主要信息，解码后速度场与原速度场测试均方根误差小于0.02，压力回归测试均方根误差可小于0.1。说明自动编码器能够作为一种流场的特征提取和降阶方法，在未来得到更广泛的应用。

Abstract

Known as a compression algorithm

autoencoder is widely used for dimensional reduction and image denoising. It can be used in flow field identification and data processing as a reduced-order method. Moreover

massive labeled flow field data makes it promising to apply machine learning in fluid dynamics. Taking the flow around a cylinder as an example

the autoencoder model of the velocity field behind the cylinder is established to reduce the order and extract features from the original data. This model encodes the 1394 velocity components into 32-dimensional data

and trains them through self-supervised learning. For well-trained autoencoder models

the basis for evaluating them is to discuss whether the original flow field could be reconstructed. The application of the low dimensional encoded data is explored by correlating it with the flow field sensitive outputs

and a neural network for the regression of surface pressure of cylinder based on encoded data is established. It is verified that the result of the autoencoder has inherited the main information in the original velocity field. The root mean square error of the decoded velocity field compared with the original field is less than 0.02

and the root mean square error of the pressure coefficient regression network can be less than 0.1. The above results indicate that the autoencoder can be used in the future as a feature extraction and order reduction method of flow field.

关键词

Keywords

references

LECUN Y, BENGIO Y, HINTON G. Deep learning.[J]. Nature, 2015, 521:436-444.

TRANCY B D, DURAISAMY K, ALONSO J J. A machine learning strategy to assist turbulence model development[R]. AIAA 2015-1287, 2015.

LING J, KURZAWSKI A, TEMPLETON J. Reynolds averaged turbulence modelling using deep neural networks with embedded invariance[J]. Journal of Fluid Mechanics, 2016, 807:155-166.

STROFER, CARLOS M, WU J, et al. Data-driven, physics-based feature extraction from fluid flow fields[J]. Communications in Computational Physics, 2018, 25(3):625-650.

JIN X, CHENG P, CHEN W L, et al. Prediction model of velocity field around circular cylinder over various Reynolds numbers by fusion convolutional neural networks based on pressure on the cylinder[J]. Physics of Fluids, 2018, 30(4):047105.

FARAHAT A K, GHODSI A, KAMEL M S. Efficient greedy feature selection for unsupervised learning[J]. Knowledge and Information Systems, 2013, 35(2):285-310.

KAISER E, NOACK B R, CORDIER L, et al. Cluster-based reduced-order modelling of a mixing layer[J]. Journal of Fluid Mechanics, 2014, 754:365-414.

O'SHEA T J, HOYDIS J. An introduction to deep learning for the physical layer[J]. IEEE Transactions on Cognitive Communications Networking, 2017, 3(4):563-575.

LIOU C Y, HUANG J C, YANG W C. Autoencoder for words[J]. Neurocomputering, 2014, 139:84-96.

ILIADIS M, SPINOULAS L, KATSAGGELOS A K. Deep fully-connected networks for video compressive sensing[J]. Digital Signal Processing, 2017:S1051200417302130.

ZHANG Z, SONG X D, YE S R, et al. Application of deep learning method to Reynolds stress models of channel flow based on reduced-order modeling of DNS data[J]. Journal of Hydrodynamics, 2018, 31(11):58-65.

NAIR V, HINTON G E. Rectified linear units improve restricted Boltzmann machines[C]//Proceedings of the 27th International Conference on Machine Learning. Haifa, ILMS: 2010: 807-814.

CHAI T, DRAXLER R R. Root mean square error (RMSE) or mean absolute error (MAE)?-Arguments against avoiding RMSE in the literature[J]. Geoscientific Model Development, 2014, (7):1247-1250.

KRIZHEVSKY A, SUTSKEVER I, HINTON G E. Image net classification with deep convolutional neural networks[C]//Neural Information Processing Systems-Volume 1(NIPS'12), 2012: 1097-1105.

CHEN H, QIAN W Q, HE L. Aerodynamic coefficient prediction of airfoils based on deep learning[J]. Acta Aerodynamica Sinica, 2018, 36(2):294-299. (in Chinese)

SUN W, SHAO S, ZHAO R, et al. A sparse auto-encoder-based deep neural network approach for induction motor faults classification[J]. Measurement, 2016, 89(ISFA):171-178.

SCHAFER M, TUREK S, DURST F, et al. Benchmark computations of laminar flow around a cylinder[J]. Notes on Numerical Fluid Mechanics, 1996, 48:547-566.

HINTON G E, SALAKHUUTDINOV R R. Reducing the dimensionality of data with neural networks[J]. Science, 2006, 313:504-507.

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

基于条件扩散模型的火箭气动外形快速设计方法

基于机器学习的极端风场短时预测研究进展与思考

大型结构风效应流固耦合机器学习研究进展

基于机器学习的房颤病人左心耳血流动力学研究

航空气动噪声机器学习研究进展