Bare-Bones Multiobjective Particle Swarm Optimization Based on Parallel Cell Balanceable Fitness Estimation

Abstract

The convergence and diversity of the Pareto optimal solutions is of great importance for multiobjective evolutionary algorithms. Based on parallel cell balanceable fitness estimation (PCBFE), a novel bare-bones multiobjective particle swarm optimization (NBBMOPSO) algorithm is proposed in this paper. First, the PCBFE strategy, which is based on the parallel cell mapping approach, is developed to retain the balance between the proximity and the diversity. After that, the PCBFE strategy is adopted to maintain external archive and update leaders. Second, an adaptive update strategy for crossover probability is designed to repair the weakness of particle search. Finally, an elitism learning strategy is performed to exchange useful information among solutions in the external archive, which can enhance the capability of dropping out of the local Pareto front. To demonstrate the merits of NBBMOPSO for multiobjective optimization, Zitzler-Deb-Thiele (ZDT) and Deb-Thiele-Laumanns-Zitzler (DTLZ) test suits are examined with comparisons against the other seven state-of-the-art competitors. Experimental results show that the proposed NBBMOPSO outperforms all the other methods in terms of the chosen performance metrics.

Keywords

• Multiobjective optimization problems
• bare-bones particle swarm optimization
• parallel cell balanceable fitness estimation
• adaptive crossover probability
• elitism learning strategy