Broadband Shaped-Beam Reflectarray Optimization Using Support Vector Machines

Abstract

In this work, SVMs are employed to accelerate the optimization of a 1-meter contoured-beam reflectarray antenna for direct broadcast satellite application in a 15% bandwidth in dual-linear polarization. A method of moments based on local periodicity is used to obtain samples to train the SVMs for each frequency. The surrogate model is then used for a design at central frequency, that is later used as starting point for a broadband optimization procedure that is accelerated more than an order of magnitude without a significant loss of accuracy. The minimum copolar gain in the coverage zone is improved more than 10 dB at the upper frequency while maintaining a computationally efficient design procedure.