Forward electromagnetic modeling and inverse scattering of cylinder with various cross-section using physics informed neural network

Physical informed Neural Networks (PINNs) represent an innovative approach that integrates the precision of physical modeling with the computational power of deep learning to effectively solve partial differential equations (PDEs) and other complex physical problems. The primary objective of this study is to thoroughly investigate the electromagnetic scattering and inverse scattering characteristics of an object using PINNs, by utilizing the Maxwell equations expressed in sparse matrix form as the physical constraints for PINNs, and accurately incorporating the results from finite-difference frequency-domain method (FDFD) as input field data for the neural network, in order to achieve a comprehensive and precise analysis of electromagnetic phenomena.