ANN Characterization of Multi-Layer Reflectarray Elements for Contoured-Beam Space Antennas in the Ku-Band

The analysis of a 1.2-meter, contour-shaped reflectarray antenna through the use of Artificial Neuronal Networks (ANNs) is carried out in this paper. The analysis is a two-step procedure: reflectarray element modeling and pattern synthesis. In the first step, artificial neural networks are found to reproduce both the amplitude and the phase of the complex reflection coefficient of the three-layered reflectarray element. For this task, up to 9 free input parameters are considered: six geometrical parameters, the incident angle in terms of azimuth, theta, and elevation, phi, and the frequency. Because of this large number of free parameters, a new artificial neural network training methodology has been developed regarding both the training set and the training process itself. In the second step, extensive full wave electromagnetic computation is replaced by trained artificial neural networks to calculate the electric field on the planar structure and the radiation pattern. A good agreement is obtained compared to an analogous analysis carried out by Method of Moments. Thanks to this methodology, the speed up factor in terms of time is in the order of 7x10(2), which represents a significant improvement in Computer Aided Design (CAD) of reflectarray antennas.