Study of artificial neural network capabilities for the analysis and design of shaped-beam reflectarrays

The performance of contoured-beam and pencil-beam reflectarrays, the analysis and design aided by the use of artificial neural networks is evaluated in this work. The global antenna radiation pattern as well as the field distribution on the reflector surface obtained from full-wave electromagnetic computation are compared with the pattern and the field distribution obtained by the aid of artificial neural network characterization. The latter case presents a good agreement with the former at much less time-consumption cost which may be taken as a positive advantage in future reflectarray design.