Dynamic Spatial Guided Multi-Guide Particle Swarm Optimization Algorithm for Many-Objective Optimization

Abstract

The multi-guide particle swarm optimization (MGPSO) algorithm utilizes random tournament selection in determining the archive guide for the velocity update of a particle, choosing the least crowded solution of a static number of solutions in the external archive. This report aims to determine the feasibility of utilizing a linearly decreasing tournament size with the aim of improving initial exploration and final exploitation of the search space by the particle swarms. The archive guide for a given particle is determined from the nearest archive solutions with the aim of increasing swarm exploration efficiency. The proposed dynamic spatial MGPSO algorithm is compared with the original MGPSO algorithm and state-of-the-art algorithms specifically designed to solve many-objective optimization problems. The results show that the dynamic soatial guided MGPSO (DSG-MGPSO) scales well to many-objective problems, with performance very competitive to that of other many-objective optimization algorithms.