Computationally Efficient Design Optimization of Multiband Antenna Using Deep Learning–Based Surrogate Models

Abstract

In this paper, deep learning–based data-driven surrogate modeling approach is proposed for enhancing cost-efficiency of multiband antenna design optimization. The proposed surrogate model–assisted design approach has achieved a computational cost reduction of almost 40% compared to the conventional direct electromagnetic solver–based design methodologies in case of single design example. As for the validation of the proposed method, the obtained optimal design parameters from the surrogate model are used to manufacture an antenna design. The obtained results from the experimental measurement are compared with counterpart results from the literature.