Hidden fluid mechanics: Learning velocity and pressure fields from flow visualizations
By Maizar Raissi, Alireza Yazdani, G. Em Karniadakis. 2020
2,068 citations recorded by the source corpus.
Machine-learning fluid flow Quantifying fluid flow is relevant to disciplines ranging from geophysics to medicine. Flow can be experimentally visualized using, for example, smoke or contrast agents, but extracting velocity and pressure fields from this information is tricky. Raissi et al. developed a machine-learning approach to tackle this problem. Their method exploits the knowledge of Navier-Stokes equations, which govern the dynamics of fluid flow in many scientifically relevant situations. The authors illustrate their approach using examples such as blood flow in an aneurysm. Science, this issue p. 1026 A machine learning approach exploiting the knowledge of Navier-Stokes equations can extract detailed fluid flow information. For centuries, flow visualization has been the art of making fluid motion visible in physical and biological systems. Although such flow patterns can be, in principle, described by the Navier-Stokes equations, extracting the velocity and pressure fields directly from the images is challenging. We addressed this problem by developing hidden fluid mechanics (HFM), a physics-informed deep-learning framework capable of encoding the Navier-Stokes equations into the neural networks while being agnostic to the geometry or the initial and boundary conditions. We demonstrate HFM for several physical and biomedical problems by extracting quantitative information for which direct measurements may not be possible. HFM is robust to low resolution and substantial noise in the observation data, which is important for potential applications.
Subfields: Computer Vision and Pattern Recognition, Statistical and Nonlinear Physics, AI for Physical Sciences
How researchers see this paper
The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.
In Computer Vision and Pattern Recognition
- G. Em Karniadakis
- Paris Perdikaris
- Sifan Wang
- Jian-Xun Wang
- Luning Sun
- Yang Liu
- Andrew Markham
- T. Nissen‐Meyer
In Statistical and Nonlinear Physics
- G. Em Karniadakis
- Maizar Raissi
- Lu Lu
- Paris Perdikaris
- Jian-Xun Wang
- Zhiping Mao
- Shengze Cai
- Han Gao
In AI for Physical Sciences
- G. Em Karniadakis
- Maizar Raissi
- Paris Perdikaris
- Lu Lu
- Ameya Dilip Jagtap
- Sifan Wang
- Yang Liu
- Xuhui Meng
Citations in the corpus
- Interpretable Machine Learning: Fundamental Principles and 10 Grand Challenges (2021, methodology): “They have been used in many scientific applications, including fluid mechanics modeling (Raissi et al., 2020), cardiac activation mapping (Sahli Costabal et al., 2020), stochastic systems modeling (Yang and Perdikaris, 2019), and discovery of differential equations (Raissi et al., 2018; Raissi,…”
- Physics informed deep learning for computational elastodynamics without labeled data (2020, background): “[9, 10, 15, 16] has inherited and extended this concept, leveraged the strong expressibility of DNNs, and developed the general PINN framework to solve the forward and inverse problems involving the system of nonlinear PDEs with small datasets or even without any labeled data.”
- Stiff-PINN: Physics-Informed Neural Network for Stiff Chemical Kinetics (2020, methodology): “[11] M.”
- DeepXDE: A Deep Learning Library for Solving Differential Equations (2019, background): “An attractive feature of PINNs is that it can be used to solve inverse problems with minimum change of the code for forward problems [47, 48, 51, 21, 13].”
- Physics-Informed Neural Operator for Learning Partial Differential Equations (2021, positive): “For example, in challenging problems such as turbulence, PINNs are only able to solve the PDE on a relatively small domain [44], or otherwise, require extra observational data which is not always available in practice [31, 45].”
- CAN-PINN: A Fast Physics-Informed Neural Network Based on Coupled-Automatic-Numerical Differentiation Method (2021, methodology): “Interesting applications include quantification of cardiovascular flows from visualization or sensor data [32–34], metamaterials design [35], and nondestructive quantification of cracks [36].”
- PhyGeoNet: Physics-informed geometry-adaptive convolutional neural networks for solving parameterized steady-state PDEs on irregular domain (2020, methodology): “This weakly-supervised method has been applied to solve various PDEs with limited training data in many scientific problems, including subsurface flows [20], vortex-induced vibrations [21], turbulent flows [22, 23], cardiovascular systems [14, 24–26], metamaterial design [27–29], geostatistical mode”
- NSFnets (Navier-Stokes flow nets): Physics-informed neural networks for the incompressible Navier-Stokes equations (2020, methodology): “This approach has also been used to simulate vortex induced vibrations in [16] and also to tackle ill-posed inverse fluid mechanics problems, a framework called “hidden fluid mechanics” presented in [17].”
- Analyses of internal structures and defects in materials using physics-informed neural networks (2022, neutral): “L D (, ) = 1 ─ N D ∑ i=1 N D ∣ ̃ rD ( X D (i) ( geo ) ; ) ∣ 2 (13)”
- A comprehensive study of non-adaptive and residual-based adaptive sampling for physics-informed neural networks (2022, methodology): “Physics-informed neural networks (PINNs) [1] have emerged in recent years and quickly became a powerful tool for solving both forward and inverse problems of partial differential equations (PDEs) via deep neural networks (DNNs) [2, 3, 4].”