RCS CALCULATION USING HYBRID FDTD-NARX TECHNIQUE

This paper amalgamates two uncorrelated techniques namely finite difference time domain technique (FDTD) and nonlinear autoregressive with exogenous input (NARX) neural network to achieve a faster computation of radar cross section (RCS). It generates only a limited number of FDTD data and uses them to train a NARX neural network. The data beyond this limited number for the FDTD come from the NARX prediction. Comparison of the performance of FDTD-NARX hybrid with other methods indicates good matching with better timing for RCS of electrically larger objects.