Exploration enhanced particle swarm optimization using guided re-initialization

Abstract

Particle Swarm Optimization (PSO) is a stochastic computation technique aimed at finding the optimal solution to a problem. It is a population based technique inspired by the behavior of a flock of birds or school of fish, developed by Dr. Eberhart and Dr. Kennedy in 1995. The original algorithm suffers from drawbacks like premature convergence at local optimum solution (optima), and high computational cost with little robustness in case of multi-modal problems (problems involving multiple optima). This paper introduces a concept aimed at increasing the diversity (exploration of the search space) portrayed by these particles. The algorithm implements a form of teleportation by which particles are randomly re-initialized in the search space once their behavior becomes predictable. Two approaches to the implementation of this idea shall be described and discussed here. The predictability is modeled using a hyper-sphere of variable radius, centered at the best known solution. © 2013 Springer.