Data‐driven surrogate modeling of dielectric reflectarray antenna using 3D printing technology for X band applications

Abstract

Reflectarray (RA) antennas are gaining recognition as a high‐performance alternative for antenna design. They combine the functionalities of standard parabolic reflectors and phased array antennas without a feed network. However, the design of RA models requires a model that exhibits exceptional performance while meeting contradictory requirements such as size and operation band. An effective and computationally efficient approach is to employ a rapid and precise unit element model for the design of reflective antennas. Artificial intelligence (AI) regression techniques have been developed to predict electromagnetic behavior efficiently and accurately in microwave device design. A cross‐shaped 3D printable RA was examined as a case study, and a design‐ready surrogate model was developed with a mean absolute error of approximately 5° using only 800 data samples. The method can be considered a feasible substitute for current surrogate‐assisted RA design techniques due to its computational efficacy and dependability.