Intelligent Electromagnetic Scattering Analysis Based on Inherent Feature Extraction

In this paper, intelligent prediction methods for electromagnetic scattering problems based on inherent feature extraction are presented. The electromagnetic characteristics of a target are expressed by a set of inherent feature parameters (IFPs), which are distinctive to the target itself, and independent of incident and scattering directions. Using IFPs as the output of neural networks is conducive to reducing the network size and training difficulty. The scattered far field and radar cross-section (RCS) can be readily computed using the IFPs. Numerical results show that the proposed approaches are viable for both 2D and 3D metal targets.