Author(s): Xiao Luo; Saiyu Yuan; Jiajian Qiu; Hongwu Tang
Linked Author(s): Saiyu Yuan, Hongwu Tang
Keywords: Physics-Informed Neural Networks; Unsteady flow; Saint-Venant; Channel networks
Abstract: Physics-Informed Neural Networks (PINNs) are adept at solving both forward and backward physics problems, predicting system outcomes, and inferring unknown parameters from system behavior. While PINNs have proven effective in hydraulic and groundwater modeling (He et al., 2020), their application to complex river hydrodynamics, especially in multi-channel networks with diverse geometries and tributary effects, is less developed. This paper delves into using PINNs for detailed hydrodynamic simulations in such complex river systems.
Year: 2024