Reconstructing Rayleigh-Benard flows out of temperature-only measurements using Physics-Informed Neural Networks

被引:9
|
作者
Di Leoni, Patricio Clark [1 ]
Agasthya, Lokahith [2 ,3 ,4 ]
Buzzicotti, Michele [2 ,3 ]
Biferale, Luca [2 ,3 ]
机构
[1] Univ San Andres, Dept Ingn, Buenos Aires, Argentina
[2] Univ Roma Tor Vergata, Dept Phys, Rome, Italy
[3] Univ Roma Tor Vergata, Ist Nazl Fis Nucl, Rome, Italy
[4] IST Austria, Campus 1, A-3400 Klosterneuburg, Austria
来源
EUROPEAN PHYSICAL JOURNAL E | 2023年 / 46卷 / 03期
基金
欧洲研究理事会;
关键词
DEEP LEARNING FRAMEWORK; CONVECTION; ASSIMILATION;
D O I
10.1140/epje/s10189-023-00276-9
中图分类号
O64 [物理化学(理论化学)、化学物理学];
学科分类号
070304 ; 081704 ;
摘要
We investigate the capabilities of Physics-Informed Neural Networks (PINNs) to reconstruct turbulent Rayleigh-Benard flows using only temperature information. We perform a quantitative analysis of the quality of the reconstructions at various amounts of low-passed-filtered information and turbulent intensities. We compare our results with those obtained via nudging, a classical equation-informed data assimilation technique. At low Rayleigh numbers, PINNs are able to reconstruct with high precision, comparable to the one achieved with nudging. At high Rayleigh numbers, PINNs outperform nudging and are able to achieve satisfactory reconstruction of the velocity fields only when data for temperature is provided with high spatial and temporal density. When data becomes sparse, the PINNs performance worsens, not only in a point-to-point error sense but also, and contrary to nudging, in a statistical sense, as can be seen in the probability density functions and energy spectra.
引用
收藏
页数:8
相关论文
共 50 条
  • [1] Reconstructing Rayleigh–Bénard flows out of temperature-only measurements using Physics-Informed Neural Networks
    Patricio Clark Di Leoni
    Lokahith Agasthya
    Michele Buzzicotti
    Luca Biferale
    The European Physical Journal E, 2023, 46
  • [2] Reconstructing Rayleigh-Benard flows out of temperature-only measurements using nudging
    Agasthya, Lokahith
    Clark Di Leoni, Patricio
    Biferale, Luca
    PHYSICS OF FLUIDS, 2022, 34 (01)
  • [3] Periodically activated physics-informed neural networks for assimilation tasks for three-dimensional Rayleigh-Benard convection
    Mommert, Michael
    Barta, Robin
    Bauer, Christian
    Volk, Marie-Christine
    Wagner, Claus
    COMPUTERS & FLUIDS, 2024, 283
  • [4] Physics-informed neural networks for periodic flows
    Shah, Smruti
    Anand, N. K.
    PHYSICS OF FLUIDS, 2024, 36 (07)
  • [5] Approximating Rayleigh scattering in exoplanetary atmospheres using physics-informed neural networks
    Dahlbuedding, David
    Molaverdikhani, Karan
    Ercolano, Barbara
    Grassi, Tommaso
    MONTHLY NOTICES OF THE ROYAL ASTRONOMICAL SOCIETY, 2024, 533 (03) : 3475 - 3483
  • [6] Assimilation of statistical data into turbulent flows using physics-informed neural networks
    Sofía Angriman
    Pablo Cobelli
    Pablo D. Mininni
    Martín Obligado
    Patricio Clark Di Leoni
    The European Physical Journal E, 2023, 46
  • [7] Assimilation of statistical data into turbulent flows using physics-informed neural networks
    Angriman, Sofia
    Cobelli, Pablo
    Mininni, Pablo D.
    Obligado, Martin
    Di Leoni, Patricio Clark
    EUROPEAN PHYSICAL JOURNAL E, 2023, 46 (03):
  • [8] Mean flow reconstruction of unsteady flows using physics-informed neural networks
    Sliwinski, Lukasz
    Rigas, Georgios
    DATA-CENTRIC ENGINEERING, 2023, 4 (01):
  • [9] Physics-informed neural networks for inverse problems in supersonic flows
    Jagtap, Ameya D.
    Mao, Zhiping
    Adams, Nikolaus
    Karniadakis, George Em
    JOURNAL OF COMPUTATIONAL PHYSICS, 2022, 466
  • [10] Physics-informed neural networks for learning fluid flows with symmetry
    Kim, Younghyeon
    Kwak, Hyungyeol
    Nam, Jaewook
    KOREAN JOURNAL OF CHEMICAL ENGINEERING, 2023, 40 (09) : 2119 - 2127