A novel multi-objective particle swarm optimization based on dynamic crowding distance

Abstract

In this article, a multi-objective particle swarm optimization algorithm based on dynamic crowding distance (DCD-MOPSO) was proposed, in which the definition of individual's DCD was based on the degree of difference between the crowding distances on different objectives. The proposed approach computed individual's DCD dynamically during the process of population maintenance to ensure sufficient diversity amongst the solutions of the non-dominated fronts. Introducing the improved quick sorting to reduce the time for computation, both the dynamic inertia weight and acceleration coefficients are used in the algorithm to explore the search space more efficiently. Experiments on well known and widely used test problems are performed, aiming at investigating the convergence and solution diversity of DCD-MOPSO. The obtained results are compared with MOPSO and NSGA-II, yielding the superiority of DCD-MOPSO.