We have presented a computational framework for physics-informed neural networks (PINNs) for solving the elastodynamic wave equation with a rate-and-state frictional fault boundary in both 1D and 2D...This suggests that PINNs may be a highly effective tool in inferring subsurface friction parameters.
We found that a PINN defined by hard enforcement of initial conditions produces reasonable approximations for displacements and the desired friction parameter distribution...the desired friction parameter is learned within the first couple of iterations.
#physics-informed-deep-learning #elastodynamic-wave-equations #rate-and-state-frictional-faults #subsurface-friction-parameters #neural-networks
Collection
[
|
...
]