Abstract:

Particle Swarm Optimization (PSO) is a meta-heuristic algorithm to determine the social behavior of different insects and to implement their optimization techniques in various real-life optimization problems. Swarms use these optimization techniques to lead their daily life for communicating with other swarms, foraging and mating activities and travelling to their destination. PSO algorithms used these methods to solve optimization problems in several fields of computer science. However, real-life problems have different parameters and problem space is not the same for all. That’s why the original version of PSO cannot be applied in every problem. PSO algorithm changed its needful features to solve new problems based on problem specification. Different versions of PSO come up with different optimization methods focusing on specific problem space. In each version, many methods or approaches have been proposed by many researchers and their optimization strategy and performance of finding an optimal solution can be different. A study on extensive analysis of different variants of PSO algorithms should have a great value for prospective researchers focusing in this field. In this research, we will explain optimization techniques of assorted versions of PSO algorithm (BPSO, DPSO, PSO-DE, PSO-NE and HPSO) and their variants. Then analyze their performance and present comparative results. We have compared the performance of each variant in terms of benchmark functions and other related algorithms. From this research, we will have acquainted which version is better for which type of optimization problems..

Keywords: