Randomization in particle swarm optimization for global search ability

Abstract

This paper introduces a novel particle swarm optimization (PSO) with random position to improve the global search ability of particle swarm optimization with linearly decreasing inertia weight (IWPSO). Standard particle swarm optimization and most of its derivations are easy to fall into local optimum of the problem by lacking of mutation in those operations. Inspired by the acceptance probability in simulated annealing algorithm, the random factors could be put in particle swarm optimization appropriately. Consequently, the concept of the mutation is introduced to the algorithm, and the global search ability would be improved. A particle swarm optimization with random position (RPPSO) is tested using seven benchmark functions with different dimensions and compared with four well-known derivations of particle swarm optimization. Experimental results show that the proposed particle swarm optimization could keep the diversity of particles, and have better global search performance.