Low-Cost Feature-Based Tolerance Optimization of Multi-Band Antennas
Anna Pietrenko‐Dabrowska, Slawomir Marcin Koziel · 2022
Manufacturing inaccuracies are detrimental to performance of engineering designs. Antenna structures are no exception here. The effects of parameter deviations are particularly noticeable for narrow- and multi-band antennas and manifest themselves through relocations of the operating frequencies or bandwidth degradation. The ability to reduce the system sensitivity to tolerances is therefore an important design consideration. Unfortunately, robust design is a challenging task, primarily because it has to be carried out at the level of full-wave electromagnetic (EM) simulation models, which generates considerable computational expenses. In this paper, we propose a novel approach to tolerance optimization of multi-band antennas. The objective is to enlarge the maximum levels of geometry parameter deviations that still ensure perfect (100-percent) fabrication yield. The presented approach involves feature-based regression surrogates as well as the trust-region framework to ensure low cost and convergence of the optimization process. Numerical validation of the method, conducted for a dual- and a triple-band microstrip antenna, corroborates its efficacy with the average robust design cost of only a few dozens of EM analyses of the respective structure.