Efficient approach for fast synthesis of phased arrays with the aid of a hybrid genetic algorithm and a smart feed representation

Abstract

The efficiency of the approach described in the paper is provided thanks to the following three factors: (i) compact form representation of a primary feed with a given radiation pattern simulated by a combination of complex dipoles, (ii) hybridization of genetic and steepest descent gradient algorithms, and (iii) smart encoding schemes used for storing parameters of the array antennas implemented in order to reduce the number of unknowns needed to describe the system. These features enabled us to develop a reliable tool capable of fast synthesis of array antennas for various applications. The algorithm is approbated on synthesis of periodic linear arrays with narrow and flat-top beams. Comparing to a standard genetic algorithm, a twice faster convergence rate and ultimate stability in hitting the global extrema is demonstrated.