1887

Abstract

Summary

Solving the wave equation numerically constitutes the majority of the computational cost for applications like seismic imaging and full-waveform inversion (FWI). One approach is to solve the frequency-domain Helmholtz equation which allows a reduction in dimensionality as it can be solved per frequency. However, computational challenges with the classical Helmholtz solvers such as the need to invert a large stiffness matrix can make these approaches infeasible for large 3D models or for modeling high frequencies. Moreover, these methods do not have a mechanism to transfer information gained from solving one problem to the next. This becomes a bottleneck for applications like FWI. Therefore, recently an approach based on the emerging paradigm of physics informed neural networks (PINNs) has been proposed to solve the Helmholtz equation. The method has shown promise in addressing several challenges associated with the conventional algorithms. However, the approach still needs further developments to be fully practicable. Foremost amongst the challenges is the slow convergence speed, especially in the presence of sharp heterogeneities in the velocity model. Therefore, we study different activation functions routinely used in the PINN literature, in addition to the swish activation function, which we find to yield superior performance compared to the rest.

Loading

Article metrics loading...

/content/papers/10.3997/2214-4609.202113254
2021-10-18
2022-06-29
Loading full text...

Full text loading...

References

  1. Alkhalifah, T., Song, C., Waheed, U. and Hao, Q.
    [2020] Wavefield solutions from machine learned functions that approximately satisfy the wave equation. In: 82nd EAGE Annual Conference & Exhibition, 2020. European Association of Geoscientists & Engineers, 1–5.
    [Google Scholar]
  2. Cao, Y., Fang, Z., Wu, Y., Zhou, D.X. and Gu, Q.
    [2019] Towards understanding the spectral bias of deep learning. arXiv preprint arXiv:1912.01198.
    [Google Scholar]
  3. Haghighat, E. and Juanes, R.
    [2021] SciANN: A Keras/TensorFlow wrapper for scientific computations and physics-informed deep learning using artificial neural networks. Computer Methods in Applied Mechanics and Engineering, 373, 113552.
    [Google Scholar]
  4. Pratt, R.G.
    [1999] Seismic waveform inversion in the frequency domain, Part 1: Theory and verification in a physical scale model. Geophysics, 64(3), 888–901.
    [Google Scholar]
  5. Rahaman, N., Baratin, A., Arpit, D., Draxler, F., Lin, M., Hamprecht, F., Bengio, Y. and Courville, A.
    [2019] On the spectral bias of neural networks. In: International Conference on Machine Learning. PMLR, 5301–5310.
    [Google Scholar]
  6. Ramachandran, P., Zoph, B. and Le, Q.V.
    [2017] Searching for activation functions. arXiv preprint arXiv:1710.05941.
    [Google Scholar]
  7. Song, C., Alkhalifah, T. and Waheed, U.B.
    [2021] Solving the frequency-domain acoustic VTI wave equation using physics-informed neural networks. Geophysical Journal International, 225, 846–859.
    [Google Scholar]
  8. Waheed, U., Haghighat, E., Alkhalifah, T., Song, C. and Hao, Q.
    [2020] Eikonal solution using physics-informed neural networks. In: 82nd EAGE Annual Conference & Exhibition, 2020. European Association of Geoscientists & Engineers, 1–5.
    [Google Scholar]
http://instance.metastore.ingenta.com/content/papers/10.3997/2214-4609.202113254
Loading
/content/papers/10.3997/2214-4609.202113254
Loading

Data & Media loading...

This is a required field
Please enter a valid email address
Approval was a Success
Invalid data
An Error Occurred
Approval was partially successful, following selected items could not be processed due to error