EUVIMEDEUVIMED entdecken
Crossref

SARAS‐PINN: Enhancing Traveltime Modelling Through Structure‐Aware Initialization and Residual‐and‐Gradient‐Based Adaptive Sampling

Yanjin Xiang, Zhiliang Wang, Ziang Song, Rong Huang, Guojie Song, Fan Min, Shulin Pan, Lei Zhao

Geophysical Prospecting · 2026-07 · 74 (6)

DOI: 10.1111/1365-2478.70230 ↗

Kurz gefasst

Fachartikel in der Zeitschrift Geophysical Prospecting, online erschienen am 26.07.2026. Verfasst von Yanjin Xiang und 7 weiteren Autor:innen. Die Arbeit wurde bisher 1-mal zitiert (Crossref) und nennt 47 Literaturangaben.

AbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

Accurate seismic traveltime modelling in geologically complex media is fundamental to seismic imaging and inversion. While physics‐informed neural networks (PINNs) have emerged as a promising mesh‐free alternative to traditional eikonal solvers, they often suffer from low accuracy near sharp velocity contrasts and slow convergence in complex models. In this study, we develop SARAS‐PINN, a structure‐aware and residual‐gradient‐adaptive sampling framework designed to enhance PINN‐based traveltime modelling. Our method integrates two key innovations: (1) structure‐aware initialization using elliptical exclusion zones, where axes are dynamically scaled by local velocity and directional gradients to prioritize structural features; and (2) residual‐and‐gradient‐based adaptive sampling, which leverages both the local partial differential equation residual and its spatial gradient to refine the training set iteratively. Numerical experiments on layered, Marmousi and BP models demonstrate significant improvements. In the layered model, SARAS‐PINN reduces the mean absolute error (MAE) from 0.1274 to 0.0086 ms, a 15‐fold improvement in precision. For the Marmousi model, our approach achieves a 3.6‐fold computational speedup in reaching the target MAE ( ms) compared to standard PINN. Furthermore, the inclusion of the residual gradient allows for the precise capture of sharp interfaces and salt bodies that are typically undersampled by traditional residual‐only methods (e.g., residual‐based adaptive refinement with distribution). These results confirm that SARAS‐PINN provides a robust and efficient tool for high‐precision seismic simulations.

Zitieren

APA

Xiang, Y., Wang, Z., Song, Z., Huang, R., Song, G., Min, F., Pan, S., & Zhao, L. (2026). SARAS‐PINN: Enhancing Traveltime Modelling Through Structure‐Aware Initialization and Residual‐and‐Gradient‐Based Adaptive Sampling. Geophysical Prospecting, 74(6). https://doi.org/10.1111/1365-2478.70230

Vancouver

Xiang Y, Wang Z, Song Z, Huang R, Song G, Min F, et al. SARAS‐PINN: Enhancing Traveltime Modelling Through Structure‐Aware Initialization and Residual‐and‐Gradient‐Based Adaptive Sampling. Geophysical Prospecting. 2026;74(6). doi:10.1111/1365-2478.70230

BibTeX
@article{Xiang2026,
  author = {Xiang, Yanjin and Wang, Zhiliang and Song, Ziang and Huang, Rong and Song, Guojie and Min, Fan and Pan, Shulin and Zhao, Lei},
  title = {SARAS‐PINN: Enhancing Traveltime Modelling Through Structure‐Aware Initialization and Residual‐and‐Gradient‐Based Adaptive Sampling},
  journal = {Geophysical Prospecting},
  year = {2026},
  volume = {74},
  number = {6},
  doi = {10.1111/1365-2478.70230}
}
Kostenlos starten →