Data-Free Finite-Volume Convolutional Neural Networks for Steady Compressible Euler Flow around a Circular Cylinder
The study proposes a data-free Physics-Informed Neural Network (PINN) framework to solve steady compressible Euler flow around a circular cylinder across supersonic and hypersonic speeds (Ma=2 to 15). The innovation lies in combining hybrid convolutions with Mach-number-guided dynamic residual scaling and specialized loss functions. This method achieves stable, global convergence for complex flows without needing any reference data.