DSMOPSO: A distance sorting based multiobjective particle swarm optimization algorithm

Abstract

Aiming at shortcomings in global searching capacity and diversity of Pareto set existing in the traditional MOPSO, a crowding distance sorting based multiobjective particle swarm optimization algorithm (DSMOPSO) is proposed. With the elitism strategy, the evolution of the external population is achieved based on individuals' crowding distance sorting by descending order, to delete the redundant individuals in the crowding area. The update of the global optimum is performed by selecting an individual with a relatively bigger crowding distance, to lead the particles evolving to the disperse region. A small ratio mutation is introduced to the inner swarm to enhance the global searching capacity. So the number of Pareto optimal solutions can be controlled, and the convergence and diversity of Pareto optimal set can be guaranteed as well. Effectiveness of the algorithm with two and three objectives is proved by the optimization of three standard test problems. Comparison results illustrate that it outperformed NSGA-II and SPEA2 in the convergence and diversity characteristics of Pareto optimal front. The sensitivity of control parameters is analyzed to illustrate the algorithm's robustness.