基于深度神经网络的粒子图像测速算法

蔡声泽; 许超; 高琪; 魏润杰

doi:10.7638/kqdlxxb-2019.0042

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 >

基于深度神经网络的粒子图像测速算法

专栏——智能空气动力学研究探索 | 更新时间：2026-05-28

- 基于深度神经网络的粒子图像测速算法
- Particle image velocimetry based on a deep neural network
- 空气动力学学报 2019年37卷第3期页码：455-461
- 作者机构：
  
  1. 浙江大学控制科学与工程学院,杭州,310027
  2. 浙江大学航空航天学院,杭州,310027
  3. 北京立方天地科技发展有限责任公司,北京,100083
- 作者简介：
  
  蔡声泽(1991-), 男, 广东潮州人, 博士, 研究方向:流体可视化计算与机器学习.E-mail:szcai@zju.edu.cn
  许超(1980-), 教授, 研究方向:机器人与流动控制.E-mail:cxu@zju.edu.cn
- 基金信息：
  
  国家自然科学基金（61473253），中央高校基本科研业务费专项资金（2018XZZX001-09，2019QNA4056）
- DOI：10.7638/kqdlxxb-2019.0042
  中图分类号： TP183;V211.7
- 纸质出版：2019
- 稿件说明：
移动端阅览
蔡声泽, 许超, 高琪, 魏润杰. 基于深度神经网络的粒子图像测速算法[J]. 空气动力学学报, 2019, 37(3): 455-461.

CAI Shengze, XU Chao, GAO Qi, et al. Particle image velocimetry based on a deep neural network[J]. 2019, 37(3): 455-461.
蔡声泽, 许超, 高琪, 魏润杰. 基于深度神经网络的粒子图像测速算法[J]. 空气动力学学报, 2019, 37(3): 455-461. DOI： 10.7638/kqdlxxb-2019.0042.

CAI Shengze, XU Chao, GAO Qi, et al. Particle image velocimetry based on a deep neural network[J]. 2019, 37(3): 455-461. DOI： 10.7638/kqdlxxb-2019.0042.

摘要

粒子图像测速（PIV）作为一种流体力学实验技术，能够从流体图像中获取全局、定量的速度场信息。随着人工智能技术的发展，设计用于粒子图像测速的深度学习技术具有广泛的应用前景和研究价值。借鉴在计算机视觉领域用于运动估计的光流神经网络，采用人工合成的粒子图像数据集进行监督学习训练，从而获得适用于流体运动估计的深度神经网络模型，并且能够高效地提供单像素级别分辨率的速度场。文中采用人工合成的湍流流场粒子图像进行初步实验评估，并讨论PIV神经网络的隐藏层输出和内在原理，同时将训练而成的深度神经网络模型与传统的相关分析法、光流法对比；随后进行射流流场测速实验，验证深度神经网络PIV的实用性。实验结果表明，文中提出的基于深度神经网络的粒子图像测速在精度、分辨率、计算效率上具有优势。

Abstract

As an experimental technique for fluid mechanics

particle image velocimetry (PIV) can extract global and quantitative velocity field from images. With the development of artificial intelligence

designing PIV method based on deep learning is quite promising and worth studying. First

the authors in this paper introduce the optical flow neural network which is proposed in the computer vision community. Second

a dataset including particle images and the ground-truth fluid motions is generated to train the parameters of the networks. This leads to a deep neural network for particle image velocimetry which can provide dense motion estimation (one vector for one pixel) efficiently. The featuring of particle image extracted by the neural network is also investigated in this paper. It is found that feature matching improves the accuracy of estimation. The proposed network model is firstly evaluated by a synthetic image sequence of turbulent flow. A jet flow experiment is also carried out in this paper to validate the practicability. The experimental results indicate that compared with the traditional cross-correlation and optical flow methods

the proposed deep neural network has advantages in accuracy

spatial resolution as well as efficiency.

关键词

Keywords

references

RAFFEL M, SCARANO F, WERELEY S, et al. Particle image velocimetry: a practical guide[M]. Springer, 2007.

WESTERWEEL J. Digital particle image velocimetry: Theory and application[D]. Delft University of Technology, 1993.

HORN B, SCHUNCK B. Determining optical flow[J]. Artificial Intelligence, 1981, 17(1-3):185-203.

SCARANO F. Iterative image deformation methods in PIV[J]. Measurement Science Technology, 2002, 13(1):1-19.

RUHNAU P, KOHLBERGER T, SCHNÖRR C, et al. Variational optical flow estimation for particle image velocimetry[J]. Experiments in Fluids, 2005, 38(1):21-32.

HEITZ D, MÉMIN E, SCHNÖRR C. Variational fluid flow measurements from image sequences:synopsis and perspectives[J]. Experiments in Fluids, 2010, 48(3):369-393.

CORPETTI T, HEITZ D, ARROYO G, et al. Fluid experimental flow estimation based on an optical-flow scheme[J]. Experiments in Fluids, 2006, 40(1):80-97.

LIU T, SHEN L. Fluid flow and optical flow[J]. Journal of Fluid Mechanics, 2008, 614:253-291.

CHEN X, ZILLÉ P, SHAO L, et al. Optical flow for incompressible turbulence motion estimation[J]. Experiments in Fluids, 2015, 56(1):1-14.

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

KRIZHEVSKY A, SUTSKEVER I, HINTON G. ImageNet classification with deep convolutional neural networks[C]//Advances in Neural Information Processing Systems. 2012: 1097-1105.

RABAULT J, KOLAAS J, JENSEN A. Performing particle image velocimetry using artificial neural networks:A proof-of-concept[J]. Measurement Science Technology, 2017, 28(12):125301.

LEE Y, YANG H, YIN Z. PIV-DCNN:cascaded deep convolutional neural networks for particle image velocimetry[J]. Experiments in Fluids, 2017, 58(12):171.

CAI S, ZHOU S, XU C, et al. Dense motion estimation of particle images via a convolutional neural network[J]. Experiments in Fluids, 2019, 60(4):73.

HUI T, TANG X, LOY C. LiteFlowNet: A lightweight convolutional neural network for optical flow estimation[C]//IEEE Conference on Computer Vision and Pattern Recognition (CVPR). 2018.

CARLIER J. Second set of fluid mechanics image sequences-Fluid image analysis and description[R]. University of Meannheim, Technical Report FP-6-513663, 2006.

RESSEGUIER V, MÉMIN E, CHAPRON B. Geophysical flows under location uncertainty, Part Ⅱ:Quasi-geostrophic models and efficient ensemble spreading[J]. Geophysical Astrophysical Fluid Dynamics, 2017, 111(3):177-208.

LI Y, PERLMAN E, WAN M, et al. A public turbulence database cluster and applications to study Lagrangian evolution of velocity increments in turbulence[J]. Journal of Turbulence, 2008, 9:N31.

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

基于深度学习的超分辨率重构方法在CAARC标模绕流流场重构中的应用

基于卷积神经网络的Bolund岛流场重构

基于深度学习的非定常周期性流动预测方法

基于U-Net的磁浮飞行风洞气动力干扰剥离算法

飞行器模型表面三维温度场重建方法