SeismicTransformer: An attention-based deep learning method for the simulation of seismic wavefields

Improving the accuracy and efficiency of seismic wavefield simulation aids geophysical problem-solving. The finite difference (FD) is widely used, but efficiency drops with increasing grids and higher order of difference formats. We propose an attention mechanism-based deep learning method called Se...

Full description

Saved in:
Bibliographic Details
Published in:Computers & geosciences 2024-08, Vol.190, p.105629, Article 105629
Main Authors: Xiang, Yanjin, Wang, Zhiliang, Song, Ziang, Huang, Rong, Song, Guojie, Min, Fan
Format: Article
Language:English
Subjects:
Attention mechanism
Deep learning
Seismic wavefield simulation
Transformer
Citations: Items that this one cites
Online Access:Get full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:Improving the accuracy and efficiency of seismic wavefield simulation aids geophysical problem-solving. The finite difference (FD) is widely used, but efficiency drops with increasing grids and higher order of difference formats. We propose an attention mechanism-based deep learning method called SeismicTransformer. Compared with theory-driven methods, such as the second-order central difference method, SeismicTransformer offers at least a tenfold improvement in speed. Compared with the networks without the attention mechanism, the SeismicTransformer achieves better results by utilizing global information. The proposed SeismicTransformer offers a promising solution for seismic wavefield simulation. •Improving the computational efficiency for seismic wavefield simulation.•Attention mechanism utilizes the global information to obtain seismic wavefield information.•Comparison of different deep learning methods for seismic wavefields simulation.•The hyperparameter tuning experiments highlighted the importance of hyperparameters and loss functions.
ISSN:0098-3004
1873-7803
DOI:10.1016/j.cageo.2024.105629