Comparison of archiving methods in multi-objectiveparticle swarm optimization (MOPSO): empirical study

Over past few years, several successful proposals for handling multi-objective optimization tasks using particle swarm optimization (PSO) have been made, such methods are popularly known as Multi-objective Particle Swarm Optimization (MOPSO). Many of these methods have focused on improving characteristics like convergence, diversity and computational times by proposing effective 'archiving' and 'guide selection' techniques. What has still been lacking is an empirical study of these proposals in a common frame-work. In this paper, an attempt to analyze these methods has been made; discussing their strengths and weaknesses. Combined effect of 'guide selection' and 'archiving' is also understood, and it turns out that there exist certain combinations which perform better in terms of convergence, or diversity, or computational times. Finally a new hybrid proposal, by coupling-dominance with Sequential Quadratic Programming (SQP) search, has been made to achieve faster and accurate convergence.