Loading [MathJax]/extensions/TeX/ietmacros.js
Cross-Entropy Method for Combinatorial Mixed-Parameter Optimization of Waveguide Polarizers for Ku-Band | IEEE Conference Publication | IEEE Xplore

Cross-Entropy Method for Combinatorial Mixed-Parameter Optimization of Waveguide Polarizers for Ku-Band


Abstract:

In this work, a cross-entropy (CE) method-based combinatorial mixed parameter optimization approach is successfully implemented to design a wideband waveguide polarizer f...Show More

Abstract:

In this work, a cross-entropy (CE) method-based combinatorial mixed parameter optimization approach is successfully implemented to design a wideband waveguide polarizer for Ku-band applications. Technically, the choice of permittivity and corresponding width of dielectrics required to load the walls of the waveguide polarizer to produce an optimum design remains a severe challenge. We classify this as an optimization problem and address it using the CE method. The dielectric loading on the walls of a square waveguide polarizer is optimized to achieve a broader bandwidth and an axial ratio (AR) below 0.2 dB. The CE algorithm coded in MATLAB in conjunction with the full-wave simulations in CST MWS is implemented to increase the AR bandwidth and reduce the overall length of the polarizer.
Date of Conference: 27-29 December 2022
Date Added to IEEE Xplore: 24 January 2023
ISBN Information:

ISSN Information:

Conference Location: Cairo, Egypt

Funding Agency:

School of Electrical and Data Engineering, University of Technology Sydney, NSW, Australia
School of Electrical and Data Engineering, University of Technology Sydney, NSW, Australia
School of Electrical and Data Engineering, University of Technology Sydney, NSW, Australia
School of Electrical and Data Engineering, University of Technology Sydney, NSW, Australia

I. Introduction

With the increasing popularity of evolutionary optimization methods and the easy availability of unprecedented computational resources, solving complex non-linear Electromagnetic (EM) problems has become more accessible. Evolutionary Algorithms (EAs) can achieve an optimal solution with a high probability for complex high-dimensional EM problems, unlike the manual design approach, which is time-consuming and often produces sub-optimal solutions [1]. Due to strong mutual coupling and other propagation effects, optimizing antenna and EM components involves highly non-linear objective functions that exhibit an epistatic behavior [2]. Such complex antenna and EM component design require a full-wave simulation-based derivative-free optimization approach. Evolutionary Algorithms are a class of derivative-free optimization with superior exploration skills, and they can handle multi-objective, multi-dimensional, and multi-extremal optimization problems. EAs are conducive to parallel computing, which significantly reduces the optimization run time by distributing the task among multiple computers. In conjunction with full-wave EM solvers, EAs automate the antenna design process, which brings a sharp reduction in the cost of design and production [1], [3].

School of Electrical and Data Engineering, University of Technology Sydney, NSW, Australia
School of Electrical and Data Engineering, University of Technology Sydney, NSW, Australia
School of Electrical and Data Engineering, University of Technology Sydney, NSW, Australia
School of Electrical and Data Engineering, University of Technology Sydney, NSW, Australia

References

References is not available for this document.