Modified PSO Algorithm for Planar Arrays Pattern Optimization

Qiu Yongbin,Huang Xianlin,Zhang Shuchun,Jiang Wei

DOI: https://doi.org/10.11999/JEIT170167

2017-01-01

Abstract:Particle Swarm Optimization (PSO) is an efficient technology to synthesize desired pattern of antenna arrays. But in some complicated cases, the optimizer will fail because the optimum particles fall into local best solutions and the convergence is so bad, especially for nonlinear optimization of large planar arrays pattern. To solve this problem, a modified optimizer is presented to improve the convergence of traditional PSO by means of initializing the particles with analytical values rather than random values. For any given desired pattern, the corresponding aperture weights can be derived by matrix operations and these weights are then used as a particle's initial values while other particles are still initialized randomly. By this initialization, an efficient estimation of the optimum particle's initial values can be achieved before the beginning of all particles searching process. After that the standard PSO iterations work as usual. The simulation results prove that this modified optimizer converges more rapidly and deeply than the traditional PSO and more satisfying global solutions and desired pattern could be obtained.