An approach for diversity and convergence improvement of multi-objective particle swarm optimization

Abstract

To improve the diversity and convergence of multi-objective optimization, a modified Multi-Objective Particle Swarm Optimization (MOPSO) algorithm using Step-by-step Rejection (SR) strategy is presented in this paper. Instead of using crowding distance based sorting technique, the SR strategy allows only the solution with the least crowding distance to be rejected at one iteration and repeat until the predefined number of solutions selected. With introduction of SR to the selection of particles for next iteration, the modified algorithm MOPSO-SR has shown remarkable performance against a set of well-known benchmark functions (ZDT series). Comparison with the representative multi-objective algorithms, it is indicated that, with SR technique, the proposed algorithm performs well on both convergence and diversity of Pareto solutions. © Springer-Verlag Berlin Heidelberg 2013.