A New Model Based Hybrid Particle Swarm Algorithm for Multi-objective Optimization

In this paper, a hybrid PSO algorithm is proposed. The new algorithm uses a simulated annealing based weighted-sum method to perform local search. The local search mechanism prevents premature convergence, hence enhances the convergence ability to the true Pareto front. Meanwhile the multi-objective optimization problem is converted into the constrained optimization problem. For the converted problem, a new selection strategy based on the constraint dominance principle is used to select the next swarm. This attempt integrates particle swarm and evolutionary algorithm together in order to take advantage of both algorithms and improve the quality of solutions. The computer simulations for four difficulty benchmark functions show that the new algorithm is able to find uniformly distributed Pareto optimal solutions and is able to converge to the true Pareto-optimal front.