On enhancing efficiency and accuracy of particle swarm optimization algorithms

TL;DR

This paper proposes modifications to the basic particle swarm optimization algorithm, incorporating fuzzy logic and Bayesian theory, to improve its efficiency and accuracy in solving non-linear programming problems.

Contribution

It introduces novel PSO variants based on fuzzy logic and Bayesian theory that outperform or match existing algorithms in efficiency and accuracy.

Findings

Fuzzy logic-based PSO variants show improved convergence.

Bayesian theory-based PSO variants enhance solution accuracy.

Proposed variants outperform basic PSO in benchmark tests.

Abstract

The particle swarm optimization (PSO) algorithm has been recently introduced in the non--linear programming, becoming widely studied and used in a variety of applications. Starting from its original formulation, many variants for improvement and specialization of the PSO have been already proposed, but without any definitive result, thus research in this area is nowadays still rather active. This paper goes in this direction, by proposing some modifications to the basic PSO algorithm, aiming at enhancements in aspects that impact on the efficiency and accuracy of the optimization algorithm. In particular, variants of PSO based on fuzzy logics and Bayesian theory have been developed, which show better, or competitive, performances when compared to both the basic PSO formulation and a few other optimization algorithms taken from the literature.