A Study of Global Optimization Using Particle Swarms

  • Authors:

  • Jaco F. Schutte;Albert A. Groenwold

  • Affiliations:

  • Structural Optimization Research Group, Department of Mechanical and Aeronautical Engineering, University of Pretoria, Pretoria, South Africa 0002;Structural Optimization Research Group, Department of Mechanical and Aeronautical Engineering, University of Pretoria, Pretoria, South Africa 0002

  • Venue:
  • Journal of Global Optimization
  • Year:
  • 2005

Quantified Score

Hi-index 0.01

Visualization

Abstract

A number of recently proposed variants of the particle swarm optimization algorithm (PSOA) are applied to an extended Dixon-Szeg und constrained test set in global optimization. Of the variants considered, it is shown that constriction as proposed by Clerc, and dynamic inertia and maximum velocity reduction as proposed by Fourie and Groenwold, represent the main contenders from a cost efficiency point of view. A parameter sensitivity analysis is then performed for these two variants in the interests of finding a reliable general purpose off-the-shelf PSOA for global optimization. In doing so, it is shown that inclusion of dynamic inertia renders the PSOA relatively insensitive to the values of the cognitive and social scaling factors.