Particle Swarm Optimization using Sobol mutation

Abstract

Particle Swarm Optimization (PSO) is an evolutionary computation technique based on Swarm Intelligence. The PSO algorithm is simple in concept, easy to implement and computationally efficient. Many researchers have worked on improving its performance in various ways and have developed several interesting variants. In this paper, we present a new mutation operator called the Systematic Mutation (SM) operator for enhancing the performance of Basic Particle Swarm Optimization (BPSO) algorithm. The SM operator unlike most of its contemporary mutation operators do not use the random probability distribution for perturbing the swarm population, but uses a quasi random Sobol sequence to find new solution vectors in the search domain. The presence of SM operator makes the mutated particles to move systematically in the search space. The comparison of SMPSO is made with BPSO and some other variants of PSO on a set of 15 benchmark global optimization problems and three real life engineering design problems. The empirical results show that SM operator significantly improves the performance of PSO in terms of fitness function value.