Improved Deep Learning Algorithm for Reproduction of Airfoil Flow Field

doi:10.7652/xjtuxb202103003

您当前的位置：

首页 >

文章列表页 >

Improved Deep Learning Algorithm for Reproduction of Airfoil Flow Field

更新时间：2025-07-09

- Improved Deep Learning Algorithm for Reproduction of Airfoil Flow Field
- Vol. 55, Issue 3, Pages: 20-28(2021)
- 作者机构：
  
  1. 兰州理工大学能源与动力工程学院,兰州,730050
  2. 中国核动力研究设计院,成都,610000
- 作者简介：
- 基金信息：
- DOI：10.7652/xjtuxb202103003
  CLC： TP183;V211.3
- Online First：10 March 2021，
  
  Published：2021
- 稿件说明：
移动端阅览
Improved Deep Learning Algorithm for Reproduction of Airfoil Flow Field[J]. 2021, 55(3): 20-28.
DOI：

Improved Deep Learning Algorithm for Reproduction of Airfoil Flow Field[J]. 2021, 55(3): 20-28. DOI： 10.7652/xjtuxb202103003.

摘要

为克服计算流体力学(CFD)方法计算成本高及无法重复利用计算结果的缺陷

基于深度学习方法

利用132组二维流场数据建立NACA0018翼型α=2°～8°、Re=0.1×10

～1.6×10

下压力场、速度场的神经网络定常预测模型。以此为基础

将低速不可压缩流动能量守恒方程作为约束条件

考虑到翼型升阻力与表面压力的相关性及流场静压与动压的关系

提出一种能间接约束翼型压力场与速度场之间关系的激活函数。结果表明

传统激活函数与改进激活函数下的神经网络在压力场预测平均误差均在2.77%左右

但传统神经网络在速度场预测中平均误差达到11%

最大达到26.993%

而改进激活函数平均误差只有2.77%。与传统的激活函数相比

改进的激活函数的神经网络因存在内部间接约束

故在翼型速度场的预测中更准确

流场过渡更均匀

并且神经网络方法可以通过保存模型以达到重复利用数值模拟结果目的

相比传统CFD方法数小时的计算

训练完成的神经网络只需要数秒即可得到计算结果

可大大减少计算时间。

Abstract

In order to overcome the disadvantages of the computational fluid dynamics(CFD)method such as high computational cost and the inability to reuse the computational results

a steady prediction model of pressure and velocity fields for NACA0018 airfoil in α=2°-8°

Re=0.1×10

-1.6×10

is established based on deep-learning method using 132 sets of two-dimensional flow data. The energy conservation equation of incompressible flow at low velocity is used as the constraint condition. Considering the correlation between lift drag and surface pressure

an activation function is proposed. The results show that for pressure field prediction the average e

rror of the traditional neural network is about 2.77%

but for velocity field prediction that of the traditional neural network is 11% and the maximum is 26.993%

while the average error of the improved neural network is only 2.77%. Compared with the traditional activation function

the improved activation function neural network is more accurate in predicting airfoil velocity field and the flow field transition is more uniform. Compared with the traditional CFD method

the neural network can obtain the flow field in a few seconds

which can greatly reduce the calculation time.

关键词

Keywords

references

朱志斌, 刘强, 白鹏. 低雷诺数翼型层流分离现象大涡模拟方法 [J]. 空气动力学学报, 2019, 37(6): 915-923.

ZHU Zhibin, LIU Qiang, BAI Peng. Large eddy simulation method for the laminar separation phenomenon on low Reynolds number airfoils [J]. Acta Aerodynamica Sinica, 2019, 37(6): 915-923.

时北极, 何国威, 王士召. 基于滑移速度壁模型的复杂边界湍流大涡模拟 [J]. 力学学报, 2019, 51(3): 754-766.

SHI Beiji, HE Guowei, WANG Shizhao. Large-eddy simulation of flows with complex geometries by using the slip-wall model [J]. Chinese Journal of Theoretical and Applied Mechanics, 2019, 51(3): 754-766.

赵松原, 黄明恪. 模拟退火算法和POD降阶模态计算在翼型反设计中的应用 [J]. 空气动力学学报, 2007, 25(2): 236-240.

ZHAO Songyuan, HUANG Mingke. Application of simulated annealing method and reduced order models based on POD to airfoil inverse design problems [J]. Acta Aerodynamica Sinica, 2007, 25(2): 236-240.

刘浩, 徐敏, 叶茂. 基于特征正交分解的跨声速流场重构和翼型反设计方法研究 [J]. 空气动力学学报, 2012, 30(4): 539-545.

LIU Hao, XU Min, YE Mao. Investigations of transonic flow field reconstruction and inverse airfoil design based on proper orthogonal decomposition ROM [J]. Acta Aerodynamica Sinica, 2012, 30(4): 539-545.

SILVA W A. Simultaneous excitation of multiple-input/multiple-output CFD-based unsteady aerodynamic systems [J]. Journal of Aircraft, 2008, 45(4): 1267-1274.

LEE S, YOU D. Prediction of laminar vortex shedding over a cylinder using deep learning [EB/OL]. [2020-07-25]. https: ∥arxiv.org/abs/1712.07854.

何磊, 钱炜祺, 汪清, 等. 机器学习方法在气动特性建模中的应用 [J]. 空气动力学学报, 2019, 37(3): 470-479.

HE Lei, QIAN Weiqi, WANG Qing, et al. Applications of machine learning for aerodynamic characteristics modeling [J]. Acta Aerodynamica Sinica, 2019, 37(3): 470-479.

惠心雨, 袁泽龙, 白俊强, 等. 基于深度学习的非定常周期性流动预测方法 [J]. 空气动力学学报, 2019, 37(3): 462-469.

HUI Xinyu, YUAN Zelong, BAI Junqiang, et al. A method of unsteady periodic flow field prediction based on the deep learning [J]. Acta Aerodynamica Sinica, 2019, 37(3): 462-469.

NAIR V, HINTON G E. Rectified linear units improve restricted Boltzmann machines [C]∥Proceedings of the 27th ICML. Haifa, Israel: [s.n.], 2010: 807-814.

MAAS A L, HANNUN A Y, NG A Y. Rectifier nonlinearities improve neural network acoustic models [C/OL]∥Proceedings of 30th International Conference on Machine Learning.[2020-07-25]. https: ∥arxiv. org/abs/1706.02515.

KLAMBAUER G, UNTERTHINER T, MAYR A, et al. Self-normalizing neural networks [C]∥Proceedings of the 31st NIPS. Long Beach, CA, USA: [s.n.], 2017: 972-981

李航. 统计学习方法 [M]. 2版. 北京: 清华大学出版社, 2019: 4-8.

徐洋, 陈燚, 黄磊, 等. 基于多层BP神经网络和无参数微调的人群计数方法 [J]. 计算机科学, 2018, 45(10): 235-239.

XU Yang, CHEN Yi, HUANG Lei, et al. Crowd counting method based on multilayer BP neural networks and non-parameter tuning [J]. Computer Science, 2018, 45(10): 235-239.

杨从新, 姬永魁. 关于对称翼型气动特性数值模拟的研究 [J]. 新技术新工艺, 2012(7): 47-50.

YANG Congxin, JI Yongkui. Numerical simulation on aerodynamic performance of symmetric airfoil [J]. New Technology New Process, 2012(7): 47-50.

KIM H J, LEE S, FUJISAWA N. Computation of unsteady flow and aerodynamic noise of NACA0018 airfoil using large-eddy simulation [J]. International Journal of Heat and Fluid Flow, 2006, 27(2): 229-242.

汪泉, 陈晓田, 黄攀, 等. 垂直轴风力机两种翼型气动性能比较研究 [J]. 可再生能源, 2020, 38(2): 199-204.

WANG Quan, CHEN Xiaotian, HUANG Pan, et al. Comparative study on the aerodynamic performance of two airfoils of vertical axis wind turbine [J]. Renewable Energy Resources, 2020, 38(2): 199-204.

高飞, 李广雪, 乔璐璐. 山东半岛近海潮汐及潮汐、潮流能的数值评估 [J]. 中国海洋大学学报(自然科学版), 2012, 42(12): 91-96.

GAO Fei, LI Guangxue, QIAO Lulu. Resource assessment of the tidal energy around the Shandong peninsula [J]. Periodical of Ocean University of China, 2012, 42(12): 91-96.

帕拉斯基沃尤. 垂直轴风力机原理与设计 [M]. 李春, 等译. 上海: 上海科学技术出版社, 2013: 334-345.

罗惕乾. 流体力学 [M]. 2版. 北京: 机械工业出版社, 2003: 129-136.

KINGMA D P, BA J. Adam: a method for stochastic optimization [EB/OL]. [2020-07-25]. https: ∥arxiv.org/abs/1412.6980.

卓丛山, 钟诚文, 李凯, 等. 高雷诺数下的翼型绕流LBM数值模拟 [J]. 航空学报, 2010, 31(2): 238-243.

ZHUO Congshan, ZHONG Chengwen, LI Kai, et al. Simulation of high Reynolds number flow around airfoil by lattice Boltzmann method [J]. Acta Aeronautica et Astronautica Sinica, 2010, 31(2): 238-243.

张士杰, 袁新, 叶大均. 非线性双方程湍流模型用于翼型分离流动模拟 [J]. 太阳能学报, 2001, 22(2): 162-166.

ZHANG Shijie, YUAN Xin, YE Dajun. Simulation of separated flows for the nrel airfoil using anisotropic two-equation turbulence models [J]. Acta Energiae Solaris Sinica, 2001, 22(2): 162-166.

POPOV A V, BOTEZ R M, LABIB M. Transition point detection from the surface pressure distribution for controller design [J]. Journal of Aircraft, 2008, 45(1): 23-28.

Views

下载量

CSCD

Alert me when the article has been cited

提交

Tools

Publicity Resources

Dynamic Flow Field Prediction Method for Compressed Air Storage Tanks Using Dynamic Mode Decomposition and Long Short-Term Memory Network

Study on Dynamic Prediction Method of Flow Field in Compressed Air Storage Device Based on Dynamic Mode Decomposition and LSTM Neural Network

Cavitation State Recognition of Centrifugal Pump Based on Deep Learning

Reduced-Order Prediction Method for the Dynamic Thermal Response of Phase-Change Cooling Heat Exchangers

Intelligent Quantitative Non-Destructive Evaluation Method for Delamination of Carbon Fiber Composite Tube with Mirror-Reflection Infrared Thermography

Related Author

SU Hao

WANG Tongsheng

ZOU Hansen

YU Keyu

XI Guang

YAO Erren

SU Hao

WANG Tonsheng

Related Institution

School of Future Technology, Xi’an Jiaotong University, Xi’an 710049, China

School of Energy and Power Engineering, Xi’an Jiaotong University, Xi’an 710049, China

School of Energy and Power Engineering, Xi’an Jiaotong University

School of Future Technology, Xi’an Jiaotong University

College of Power Engineering, Naval University of Engineering

⁰