Abstract: In order to improve the weaknesses of the particle swarm's easily premature and slow convergence in late stage, the H-MOPSO, based on the integration of improved MOPSO and local search, was proposed. First of all, the non-uniform mutation operator and self-adaptive inertia weight were adopted to enhance its ability of global search. I}hen, the model of MOPSO hybrid with local search was established. According to the model, the local search algorithm based on hill climbing strategy with sidesteps was periodically used to optimize the swarm, making particles search a- long descent direction when they were away from Pareto front, and search along non-dominated direction while they were near Pareto front Simulation results of benchmark functions show that H-MOPSO has better performance com- pared with MOPSO,NSGA-II and MOEA/D. The solving of supplier selection problem further validates its effective- ness.

Key words: Multi-objcctivc optimization, Particle swarm optimization, Local search, Hybrid algorithm