Faster Antenna Optimization Using a Hybrid EP-PSO Algorithm.

Application of the evolutionary and multi-agent stochastic optimization techniques such as Evolutionary Programming (EP), Evolution Strategies (ES), Genetic Algorithms (GAs) and Particle Swarm Optimization in electromagnetic and antenna design have demonstrated that these probabilistic methods can yield robust globally optimized solutions to problems that otherwise are not amenable to traditional gradient-based localsearch optimization methods [1-2]. A comparative study ofEP, GAs and PSO to various antenna problems were reported in [3] where it was shown that EP has the best performance among these algorithms, whereas PSO provides a few very high quality solutions, which are comparable to or better than those of EP, majority of its solutions over a large number of trials are rather poor as compared to other algorithms. The questions naturally arises as to whether a judicious hybridization of these algorithms may result in a new algorithm with a faster convergence in typical antenna optimization problems. In this work we investigate the efficiency of an evolutionary optimization approach that incorporates swarm directions in standard EP algorithm. Our hybrid EP-PSO approach is inspired by the work reported in [4] for test-function optimization. Unlike [4] however, we have also investigated the use of various mutation operators in the hybrid EP-PSO technique. The technique is applied to two antenna problems: the side-lobe-level reduction of a non-uniform spaced (aperiodic) linear array and the beam shaping of a patch antenna loaded with a partially reflective surface.

Main Author: Hoorfar, Ahmad.
Other Authors: Lakhani Shamsha.
Language: English
Published: 2009
Online Access: