Particle Swarm Optimization for Global Optimization Problems

Abstract

In this paper, the particle swarm optimization (PSO), originated as a simulation of a simplified social system with swarm intelligence and having exploring and exploiting characteristics of the particle swarm, is adopted to deal with the global optimization problems. To begin with, four non-linear two-dimensional functions are adopted as benchmark examples for determining the suitable ranges of parameters in the PSO. It is found that uses of the PSO can make the search quickly converge to the global optimum. Then, several benchmark functions below ten-dimensions are adopted especially for comparisons of search efficiency on the PSO and other algorithms. Results show that the proposed approach is not only superior to other algorithms but also has the higher success rate. Finally, the PSO is also applied to deal with the optimization problem of a grillage structure. Optimization solutions of the success rate and the used average generation show that the PSO also has the better performance for the optimal design of structures with constraints.