PINNup: Robust neural network wavefield solutions using frequency upscaling and neuron splitting

Abstract

Solving for the frequency-domain scattered wavefield via physics-informed neural network (PINN) has great potential in seismic modeling and inversion. However, when dealing with high-frequency wavefields, its accuracy and training cost limits its applications. Thus, we propose a novel implementation of PINN using frequency upscaling and neuron splitting, which allows the neural network model to grow in size as we increase the frequency while leveraging the information from the pre-trained model for lower-frequency wavefields, resulting in fast convergence to high-accuracy solutions. Numerical results show that, compared to the commonly used PINN with random initialization, the proposed PINN exhibits notable superiority in terms of convergence and accuracy and can achieve neuron based high-frequency wavefield solutions with a two-hidden-layer model.

7 pages, 9 figures

References in corpus (3)

• On the eigenvector bias of Fourier feature networks: From regression to solving multi-scale PDEs with physics-informed neural networks
• Solving the acoustic VTI wave equation using physics-informed neural networks
• Splitting Steepest Descent for Growing Neural Architectures

Cited by in corpus (8)

• Implicit Full Waveform Inversion with Deep Neural Representation
• A Critical Review of Physics-Informed Machine Learning Applications in Subsurface Energy Systems
• Efficient physics-informed neural networks using hash encoding
• GaborPINN: Efficient physics informed neural networks using multiplicative filtered networks
• Deep Neural Helmholtz Operators for 3D Elastic Wave Propagation and Inversion
• Physics-informed neural wavefields with Gabor basis functions
• Deep-pretrained-FWI: combining supervised learning with physics-informed neural network
• Microseismic source imaging using physics-informed neural networks with hard constraints