Efficient optimization of wideband reflector feed antennas for optimal receiving sensitivity

Abstract Wideband reflector antenna feeds are a geometrically complex class of antennas, requiring a variety of design parameters to describe their form. This casts the optimization of such structures as a high‐dimensionality problem. The antenna responses are also difficult to predict and must be determined via computationally expensive full‐wave electromagnetic simulations. This paper proposes a surrogate‐based optimization approach for the design of wideband feed antennas. Beginning with high‐fidelity goal function data from a global search space, the framework nominates a number of local search regions where separate, formal, multifidelity optimizations are performed. The formulated goal function maximizes the receiving sensitivity, while ensuring that the reflection coefficient response satisfies a predefined criterion. An approximate, fast‐to‐evaluate coarse model of the goal function, based on a series of surrogate models estimating the feed and reflector antenna responses, is formulated. Most of the optimization burden is shifted to the coarse model, where high‐fidelity samples are adaptively added to the regions of the design space where the coarse model performs well or where the sampling is sparse. As a design example, the optimization of a quad‐ridge flared horn feed antenna, on a Square Kilometre Array reflector system, is presented to highlight the efficacy of the method.