Surrogate Model‐Assisted Grey Wolf Algorithm for Multi‐Objective Antenna Optimization Design

Abstract

Antenna structures usually involve many strongly coupled geometric parameters. Directly embedding full‐wave simulations into swarm‐intelligence optimization leads to high computational cost and slow convergence, while making it difficult to balance conflicting objectives such as impedance bandwidth, circular‐polarization axial ratio, and gain. To address this issue, this paper proposes a surrogate‐model‐assisted improved multi‐objective grey wolf optimization framework. The framework takes “structural parameters + frequency” as inputs to rapidly predict the swept‐frequency responses of |S11|, axial ratio (AR), and realized gain (Gain). By adaptively updating the convergence factor and introducing probabilistic mutation perturbations after position updates to enhance global exploration and avoid premature convergence, an improved grey wolf optimization strategy (IGWO) is constructed. On this basis, the method is extended to a multi‐objective framework by introducing an external archive to store non‐dominated solutions and combining grid crowding with a leader‐selection mechanism to maintain diversity, thereby obtaining a more evenly distributed Pareto‐optimal solution set. Using a millimeter‐wave circularly polarized magnetoelectric dipole antenna element as an example, the proposed method significantly reduces the number of full‐wave simulations while improving impedance bandwidth and axial ratio bandwidth, achieving comprehensive optimization of high gain, demonstrating the effectiveness and practical applicability of the framework.