Cavity Model Surrogate-Based Optimization for Electrically Thick Circularly Polarized Rectangular Microstrip Antennas

Abstract This paper presents a new cavity model surrogate-based optimization for electrically thick probe-fed circularly polarized rectangular microstrip antennas. The proposed methodology feeds back input impedance and far-field information from full-wave electromagnetic solvers in order to calibrate known sources of inaccuracy within the cavity model. Said calibration is achieved by solving a constrained non-linear least squares problem. The calibrated cavity model is used to generate a new antenna geometry that will be subjected to the same calibration process until convergence is achieved. The amount of iterations needed to achieve convergence for several configurations is less than or equal to three, this indicates that the cavity model surrogate-based optimization can be used as an efficient design technique. In addition, the optimization process correctly anticipated that unequal fringing field lengths for the resonant modes are required to calibrate the cavity model, demonstrating that great physical insight is gained throughout the optimization process. For the validation of the proposed design procedure, eight electrically thick antennas were designed and optimized, of which three were manufactured and tested. Good agreement between simulated and experimental results were observed, validating the applicability of the proposed strategy.