Multi-Objective Particle Swarm Optimization Algorithm Based on Comprehensive Optimization Strategies

Multi-objective particle swarm optimization algorithm based on comprehensive optimization strategies MOPSO-COS is proposed in this paper to deal with the problems of premature convergence and poor diversity. The velocity updating mode is modified by incorporating the information of the global second best particle to promote information flowing among particles. In order to improve the convergencei¾?accuracyi¾?and diversity, some effective strategies, such as chaotic mutation, external archiving with dynamic grid method, selection strategy based on a temporary population and so on, are introduced into MOPSO-COS. Theoretical analysis of MOPSO-COS is carried out including convergence and time complexity. Performance tests are conducted with ZDT test functions. Simulation results show that MOPSO-COS can improve the convergence accuracy and diversity of Pareto optimal solutions simultaneously, and particles can escape from local optimum point effectively.