Review of physics-informed machine-learning inversion of geophysical data

被引:0
|
作者
Schuster, Gerard T. [1 ]
Chen, Yuqing [2 ]
Feng, Shihang [3 ,4 ]
机构
[1] Univ Utah, Dept Geol & Geophys, Salt Lake City, UT USA
[2] China Univ Geosci, Key Lab Intraplate Volcanoes & Earthquakes, Beijing, Peoples R China
[3] Los Alamos Natl Lab, Div Theoret, Los Alamos, NM 87545 USA
[4] Viridien, Houston, TX 77072 USA
关键词
WAVE-FORM INVERSION; JOINT INVERSION; NEURAL-NETWORKS; MODELS;
D O I
10.1190/GEO2023-0615.1
中图分类号
P3 [地球物理学]; P59 [地球化学];
学科分类号
0708 ; 070902 ;
摘要
We review five types of physics-informed machine-learning (PIML) algorithms for inversion and modeling of geophysical data. Such algorithms use the combination of a data-driven machine-learning (ML) method and the equations of physics to model or invert geophysical data (or both). By incorporating the constraints of physics, PIML algorithms can effectively reduce the size of the solution space for ML models, enabling them to be trained on smaller data sets. This is especially advantageous in scenarios in which data availability may be limited or expensive to obtain. In this review, we restrict the physics to be that from the governing wave equation, either as a constraint that must be satisfied or by using numerical solutions of the wave equation for modeling and inversion. This approach ensures that the resulting models adhere to physical principles while leveraging the power of ML to analyze and interpret complex geophysical data. There are several potential benefits of PIML compared to standard numerical modeling or inversion of seismic data computed by, for example, finite-difference solutions to the wave equation. 1) Empirical tests suggest that PIML algorithms constrained by the physics of wave propagation can sometimes resist getting stuck in a local minima compared with standard full-waveform inversion (FWI). 2) After the weights of the neural network are found by training, then the forward and inverse operations by PIML can be more than several orders of magnitude more efficient than FWI. However, the computational cost for general training can be enormous. 3) If the ML inversion operator H-w is locally trained on a small portion of the recorded data d(obs), then there is sometimes no need for millions of training examples that aim for global generalization of H-w. The benefit is that the locally trained H-w can be used to economically invert the remaining test data d(test) for the true velocity m approximate to H(w)d(test), where d(test) can comprise more than 90% of the recorded data.
引用
收藏
页码:T337 / T356
页数:20
相关论文
共 50 条
  • [41] Physics-informed machine learning models for ship speed prediction
    Lang, Xiao
    Wu, Da
    Mao, Wengang
    EXPERT SYSTEMS WITH APPLICATIONS, 2024, 238
  • [42] Physics-informed machine learning model for bias temperature instability
    Lee, Jonghwan
    AIP ADVANCES, 2021, 11 (02)
  • [43] Physics-Informed Machine Learning Improves Detection of Head Impacts
    Raymond, Samuel J.
    Cecchi, Nicholas J.
    Alizadeh, Hossein Vahid
    Callan, Ashlyn A.
    Rice, Eli
    Liu, Yuzhe
    Zhou, Zhou
    Zeineh, Michael
    Camarillo, David B.
    ANNALS OF BIOMEDICAL ENGINEERING, 2022, 50 (11) : 1534 - 1545
  • [44] Physics-Informed Machine Learning Improves Detection of Head Impacts
    Samuel J. Raymond
    Nicholas J. Cecchi
    Hossein Vahid Alizadeh
    Ashlyn A. Callan
    Eli Rice
    Yuzhe Liu
    Zhou Zhou
    Michael Zeineh
    David B. Camarillo
    Annals of Biomedical Engineering, 2022, 50 : 1534 - 1545
  • [45] Special Issue: Physics-Informed Machine Learning for Advanced Manufacturing
    Guo, Yuebin
    JOURNAL OF MANUFACTURING SCIENCE AND ENGINEERING-TRANSACTIONS OF THE ASME, 2024, 146 (08):
  • [46] Physics-Informed Machine Learning for Modeling and Control of Dynamical Systems
    Nghiem, Truong X.
    Drgona, Jan
    Jones, Colin
    Nagy, Zoltan
    Schwan, Roland
    Dey, Biswadip
    Chakrabarty, Ankush
    Di Cairano, Stefano
    Paulson, Joel A.
    Carron, Andrea
    Zeilinger, Melanie N.
    Cortez, Wenceslao Shaw
    Vrabie, Draguna L.
    2023 AMERICAN CONTROL CONFERENCE, ACC, 2023, : 3735 - 3750
  • [47] Physics-Informed Machine Learning-An Emerging Trend in Tribology
    Marian, Max
    Tremmel, Stephan
    LUBRICANTS, 2023, 11 (11)
  • [48] Wind Farm Modeling with Interpretable Physics-Informed Machine Learning
    Howland, Michael F.
    Dabiri, John O.
    ENERGIES, 2019, 12 (14)
  • [49] Physics-Informed Machine Learning for Inverse Design of Optical Metamaterials
    Sarkar, Sulagna
    Ji, Anqi
    Jermain, Zachary
    Lipton, Robert
    Brongersma, Mark
    Dayal, Kaushik
    Noh, Hae Young
    ADVANCED PHOTONICS RESEARCH, 2023, 4 (12):
  • [50] Unit Operation and Process Modeling with Physics-Informed Machine Learning
    Li, Haochen
    Spelman, David
    Sansalone, John
    JOURNAL OF ENVIRONMENTAL ENGINEERING, 2024, 150 (04)