Physics-Informed Neural Network for Rapid Prediction of Wide-Angle Electromagnetic Scattering from Three-Dimensional Objects

In this paper, we present a physics-informed neural network (PINN) for rapidly predicting wide-angle electromagnetic scattering from three-dimensional (3D) objects by embedding the formula of the method of moments (MoM) into the loss function of a fully connected neural network. The inputs of the proposed PINN-MoM are excitations at different incident angles, while the outputs are induced current coefficients of the corresponding angles. With the powerful nonlinear fitting ability of neural networks, the training process of the PINN-MoM can converge well without any dataset. Numerical results show that the trained PINN-MoM model can predict the electromagnetic scattering of 3D objects quickly and accurately.