Surrogate models have been increasingly used to reduce the computational cost of electromagnetic (EM) design in RF and microwave components. However, component types, surrogate model families, and design workflows vary substantially across the literature. This systematic review provides a structured synthesis of surrogate-assisted EM design and optimization for RF and microwave applications. A Scopus-based screening process was employed to identify 180 journal articles published between 2012 and February 2026. After eligibility assessment, 126 studies were included in the final review corpus, whereas 54 were excluded. Six previous review articles were used separately for contextual positioning. The studies included were classified according to component category, surrogate model family, surrogate usage mode, inverse-design approach, multifidelity integration, active-learning adoption, and workflow function. The results showed that antennas and filters dominate the literature, whereas the Gaussian process or Kriging models and neural networks are the most frequent surrogate families. Optimization-based inverse design is the most commonly used, whereas multifidelity and active learning are less common. Overall, the included literature indicates that surrogate-assisted design is widely represented in RF and microwave design studies. However, no study in the included literature corpus has implemented a unified workflow that combines surrogate modeling, inverse design, multifidelity interaction, and active learning.
Prousali et al. (Sat,) studied this question.
Synapse has enriched 5 closely related papers on similar clinical questions. Consider them for comparative context: