Application of neural network and its extension of derivative to scattering from a nonlinearly loaded antenna

The neural network and its extension of derivative are applied to the scattering of a nonlinearly loaded antenna. Initially, the radar cross section (RCS) of a nonlinearly loaded antenna is modeled or predicted by a neural network. By using some extension of the neural network, the derivative, i.e., slope information, about the output of the original neural network can be obtained easily. This slope information about the RCS characteristics will help one design the nonlinearly loaded antenna efficiently. It should be emphasized that the training work of the neural network is performed only once, and can be finished in advance. Numerical examples show that the neural network can predict the RCS as well as the derivatives of RCS for a nonlinearly loaded antenna with only once of training work. Therefore, the proposed method will be helpful in the design of a nonlinearly loaded antenna.