A Study of Diverse Modifications in Particle Swarm Optimization

Abstract

This paper presents a comprehensive study of the diverse modifications to Particle Swarm Optimization (PSO) for solving science and engineering-related optimization problems. PSO is a powerful, easy-to-implement, gradient-free optimization technique, and its versatility has led researchers to apply it to both simple and complex systems. In most optimization problems, we want to find the global optimal solution, so we need an optimizer which can successfully do this job, but PSO, in its basic form has the problem of premature convergence, due to which, sometimes, it traps in local optima and does not reach to the global solution. In this point of view, many researchers have made their efforts to eliminate this problem. If we go through the literature about improving the performance of PSO in diverse directions, then we will find four major directions of improving the PSO, which are new formulation for the basic parameters of PSO, Adding new parameters, alteration in the topology of basic PSO and hybridization with other algorithms. This study concludes that the PSO is a dynamic optimizer, which can efficiently solve the problems of numerous areas of science and engineering.