Particle Swarm Optimization with Random Sampling in Variable Neighbourhoods for Solving Global Minimization Problems

Particle Swarm Optimization (PSO) is a bio-inspired evolutionary meta-heuristic that simulates the social behaviour observed in groups of biological individuals [4]. In standard PSO, the particle swarm is often attracted by sub-optimal solutions when solving complex multimodal problems, causing premature convergence of the algorithm and swarm stagnation [5]. Once particles have converged prematurely, they continue converging to within extremely close proximity of one another so that the global best and all personal bests are within one minuscule region of the search space, limiting the algorithm exploration. This paper presents a modified variant of constricted PSO [1] that uses random samples in variable neighbourhoods for dispersing the swarm whenever a premature convergence state is detected, offering an escaping alternative from local optima.