改进的物理融合神经网络在瑞利-泰勒不稳定性问题中的应用
THE APPLICATION OF MODIFIED PHYSICS-INFORMED NEURAL NETWORKS IN RAYLEIGH-TAYLOR INSTABILITY
查看参考文献33篇
|
文摘
|
基于相场法的物理融合神经网络PF-PINNs被成功用于两相流动的建模,为两相流动的高精度直接数值模拟提供了全新的技术手段.相场法作为一种新兴的界面捕捉方法,其引入确保了界面的质量守恒,显著提高了相界面的捕捉精度;但是相场法中高阶导数的存在也降低了神经网络的训练速度.为了提升计算训练过程的效率,本文在PF-PINNS框架下,参考深度混合残差方法MIM,将化学能作为辅助变量以及神经网络的输出之一,并修改了物理约束项的形式,使辅助变量与相分数的关系式由硬约束转为了软约束.上述两点改进显著降低了自动微分过程中计算图的规模,节约了求导过程中的计算开销.同时,为了评估建立的PF-PINNS在雷诺数较高、计算量较大的场景中的建模能力,本文将瑞利-泰勒RT不稳定性问题作为验证算例.与高精度谱元法的定性与定量对比结果表明,改进PF-PINNs有能力捕捉到两相界面的强非线性演化过程,且计算精度接近传统算法,计算结果符合物理规律.改进前后的对比结果表明,深度混合残差方法能够显著降低PF-PINNS的训练用时.本文所述方法是进一步提升神经网络训练速度的重要参考资料,并为探索高精度智能建模方法提供了全新的见解. |
|
其他语种文摘
|
Physics-informed neural networks for phase-field method (PF-PINNs) are successfully applied to the modeling of two-phase flow,and it provides a brand-new way for the high-accuracy direct numerical simulation of twophase flow.As an emerging interface capturing method,the introduction of the phase-field method in two phase flow ensures mass conservation near the interface and significantly enhances the interface capturing accuracy.However,the high-order derivate introduced by the phase-field method decreases the efficiency of network training.To enhance the efficiency of the training process,this paper regards the chemical energy as an auxiliary parameter and one of the outputs of the proposed neural network and revises the loss functions of physics constrain in the PF-PINNs framework based on the deep mixed residual method (MIM),which transforms the relationship between the auxiliary parameter and the phase-field variable from hard constrain to soft constrain.The proposed improvements decrease the size of the computational graph generated in automatic differentiation significantly and the computational cost of calculating highorder derivates in automatic differentiation is reduced.Meanwhile,the Rayleigh-Taylor (RT) instability is tested to assess the modeling ability of the proposed PF-PINNs when the Reynolds number is high and an enormous amount of calculation is needed.Compared with the spectral element-based phase-field method qualitatively and quantitatively,modified PF-PINNs can capture the strong non-linear evolution process of the interface,and the accuracy of modified PF-PINNs reaches the accuracy of traditional numerical solver.The result of the proposed neural network fits the characteristic of RT instability well.Compared the modified PF-PINNs with the original PF-PINNs,the results indicate that the deep mixed residual method can reduce the training time of the original PF-PINNs notably.The proposed method in this paper is a valuable reference for improving the training speed of the neural network and gives a new insight into exploring the intelligent modeling method with high accuracy. |
|
来源
|
力学学报
,2022,54(8):2224-2234 【核心库】
|
|
DOI
|
10.6052/0459-1879-22-253
|
|
关键词
|
瑞利-泰勒不稳定性
;
深度混合残差方法
;
物理融合神经网络
;
两相流
|
|
地址
|
1.
中国科学院力学研究所, 中国科学院流固耦合系统力学重点实验室, 北京, 100190
2.
中国科学院大学未来技术学院, 北京, 100049
3.
中国科学院大学工程科学学院, 北京, 100049
4.
中国科学院合肥物质科学研究院, 合肥, 230031
|
|
语种
|
中文 |
|
文献类型
|
研究性论文 |
|
ISSN
|
0459-1879 |
|
学科
|
力学 |
|
基金
|
国家自然科学基金
;
中国科学院青年创新促进会项目
|
|
文献收藏号
|
CSCD:7292456
|
参考文献 共
33
共2页
|
|
1.
Magnaudet J. Particles, drops, and bubbles moving across sharp interfaces and stratified layers.
Annual Review of Fluid Mechanics,2020,52(1):61-91
|
CSCD被引
2
次
|
|
|
|
|
2.
Ruspini L C. Two-phase flow instabilities: A review.
International Journal of Heat and Mass Transfer,2014,71:521-548
|
CSCD被引
30
次
|
|
|
|
|
3.
Sui Y. Numerical simulations of flows with moving contact lines.
Annual Review of Fluid Mechanics,2014,46(1):97-119
|
CSCD被引
11
次
|
|
|
|
|
4.
Huang B. A review of transient flow structure and unsteady mechanism of cavitating flow.
Journal of Hydrodynamics,2019,31(3):429-444
|
CSCD被引
25
次
|
|
|
|
|
5.
郭子漪. 界面张力梯度驱动对流向湍流转捩的研究.
力学进展,2021,51(1):1-28
|
CSCD被引
3
次
|
|
|
|
|
6.
李康. 一种基于双界面函数的界面捕捉方法.
力学学报,2017,49(6):1290-1300
|
CSCD被引
6
次
|
|
|
|
|
7.
Hirt C W. Volume of fluid (VOF) method for the dynamics of free boundaries.
Journal of Computational Physics,1981,39(1):201-225
|
CSCD被引
1084
次
|
|
|
|
|
8.
Sethian J A. Level set methods for fluid interfaces.
Annual Review of Fluid Mechanics,2003,35(1):341-372
|
CSCD被引
23
次
|
|
|
|
|
9.
Sussman M. A level set approach for computing solutions to incompressible two-phase flow.
Journal of Computational Physics,1994,114(1):146-159
|
CSCD被引
183
次
|
|
|
|
|
10.
Anderson D M. Diffuse-interface methods in fluid mechanics.
Annual Review of Fluid Mechanics,1998,30(1):139-165
|
CSCD被引
25
次
|
|
|
|
|
11.
Gurtin M E. Two-phase binary fluids and immiscible fluids described by an order parameter.
Mathematical Models and Methods in Applied Sciences,1996,6:815-831
|
CSCD被引
11
次
|
|
|
|
|
12.
Qiao L. Modeling thermocapillary migration of interfacial droplets by a hybrid lattice Boltzmann finite difference scheme.
Applied Thermal Engineering,2018,131:910-919
|
CSCD被引
3
次
|
|
|
|
|
13.
Yue P T. Thermodynamically consistent phase-field modelling of contact angle hysteresis.
Journal of Fluid Mechanics,2020,899:A15
|
CSCD被引
2
次
|
|
|
|
|
14.
Taylor G I. The instability of liquid surfaces when accelerated in a direction perpendicular to their planes. I. Proceedings of the Royal Society of London. Series A.
Mathematical and Physical Sciences,1950,201(1065):192-196
|
CSCD被引
1
次
|
|
|
|
|
15.
Rayleigh R. Investigation of the character of the equilibrium of an incompressible heavy fluid of variable density.
Proceedings of the London Mathematical Society,1882,1(1):170-177
|
CSCD被引
22
次
|
|
|
|
|
16.
Zhao Z. Analytical model of nonlinear evolution of single-mode Rayleigh-Taylor instability in cylindrical geometry.
Journal of Fluid Mechanics,2020,900:A24
|
CSCD被引
1
次
|
|
|
|
|
17.
Zhang T W. An interface-compressed diffuse interface method and its application for multiphase flows.
Physics of Fluids,2019,31(12):122102
|
CSCD被引
2
次
|
|
|
|
|
18.
赵暾. 基于内嵌物理机理神经网络的热传导方程的正问题及逆问题求解.
空气动力学学报,2021,39(5):19-26
|
CSCD被引
8
次
|
|
|
|
|
19.
陆志彬. 基于物理信息神经网络的传热过程物理场代理模型的构建.
化工学报,2021,72(3):1496-1503
|
CSCD被引
1
次
|
|
|
|
|
20.
金晓威. 物理增强的流场深度学习建模与模拟方法.
力学学报,2021,53(10):2616-2629
|
CSCD被引
13
次
|
|
|
|
|
了解气象卫星更多信息，请访问