Knowledge-Based Methodology of CPW-Fed Open Stub Loaded C-Shaped Microstrip Antenna by Surrogate-Based Modeling

Abstract

Antenna systems with more complicated geometries have been created as a result of evolving technology and rising performance standards in the industry. The geometric complexity of modern antennas makes it difficult for circuit theory tools or parametric studies to produce adequate findings, making it difficult for designers to create the designs they want. Even though full-wave electromagnetic (EM) modeling tools are widely utilized today, they are computationally expensive for local optimization alone. In order to speed up the stages of the simulation-based design of high-performance systems, many strategies have been devised to solve and/or minimize this challenge. Thanks to their adaptability, affordable computing costs, and widespread usage, surrogate-based models have grown to be a well-known branch. Herein, an innovative knowledge-based methodology for building a coplanar waveguide (CPW)-fed antenna using surrogate models is presented. In this work, a knowledge-based methodology using surrogate modeling is applied as an advanced approach that combines domain-specific knowledge with surrogate models to optimize the performance of a microwave antenna in a computationally efficient manner. For this aim, firstly, a 3D-EM simulator is deployed to generate a dataset for a deep learning-based surrogate model. When compared to the conventional optimization approach using direct deployment of EM simulators which is around 47.2 h, the proposed surrogate model approach has an average cost reduction of over 50% which corresponds to a total computational time of 24.3 h. The collected results are compared to performance metrics for prototype antenna designs as well as simulated outcomes from EM simulators and counterpart works from the literature.