Simulation of Flow around a Body in a Two-Dimensional Channel Using Physics-Informed Neural Networks

This paper presents some aspects of the use of physics-informed neural networks to solve a two-dimensional stationary problem of flow around an obstacle using the Navier–Stokes equations. The influence of the activation function, quantitative parameters of the training set, adaptive regularization, and adaptive grids on the quality and accuracy of solutions is studied for a fixed neural network architecture. The relationship between these factors and modeling quality is analyzed to identify optimal conditions for increasing the accuracy and stability of solutions.