Physics-Inspired Hybrid Method for the Conducting Target's RCS Extrapolation in the Frequency Domain

Abstract

A physics-inspired hybrid method for extrapolating the conducting target's radar cross section (RCS) versus frequency is presented. Inspired by physical optics, we propose using the nonlinear least squares method to capture the global trend of the conducting target's RCS and using Gaussian process regression to automatically extrapolate the remaining local fluctuations. Experiments based on simulated and measured data are carried out to verify the proposed method. This method achieves a maximum root-mean-square error of only 0.444 dBsm on the simulated data of an electrically large aircraft model, and 0.065 dBsm on the measured data of a combinatorial model. These results fully demonstrate its high extrapolation accuracy.