Coupled Electromagnetic-Thermal Analysis for Temperature-Dependent Materials with Physics-Informed Neural Networks

We present a Physics-Informed Neural Network (PINN) based method for the coupled electromagnetic-thermal analysis of microwave structures with temperature-dependent materials. Combined with the Finite-Difference-Time domain technique, the proposed approach efficiently handles the dynamic change of material parameters with temperature, without compromising accuracy. We demonstrate this method with the electromagnetic-thermal modeling of a micro-electro-mechanical switch on a coplanar waveguide. This study demonstrates the potential of employing PINNs in real-world multiphysics applications for the first time.