Comparing selected PSO modifications on CEC 15 benchmark set

Abstract

In this paper we choose to compare three popular modifications of Particle Swarm Optimization (PSO) using the recent CEC'15 benchmark set. Three different approaches for PSO modification are compared: the heterogeneous swarm, diversity-guided swarm and orthogonal learning based approach. The goal is to provide up-to date comparison of the performance of these methods on a state-of-art benchmark set and highlight the differences in performance on different types of fitness functions. The experiments were carried out according to CEC benchmark rules and statistically evaluated. © 2016 IEEE.