Physics-Informed Neural Networks: Forward and Inverse Design Solutions for Hypersonic Blunt Cones

Physics-informed neural networks are a method for solving partial differential equations. Researchers have used this in a forward manner, by solving in the solution field, and inversely, where the equations are parameterized. Researchers have developed a method to simulate supersonic flows using physics-informed networks in backstep and conic flow geometries. In this study, we focus on hypersonic, blunt-cone flows, where we simulate the flow fields in a forward and inverse manner. Using data and partial differential equations, we can solve these fields derived from the Navier-Stokes Equations.