A Hybrid Improved Particle Swarm Optimization Based on Dynamic Parameters Control and Metropolis Accept Rule Strategy

Abstract

Particle Swarm Optimization (PSO), a population-based intelligent modern heuristic algorithm, is inspired from the simulation of flock prayer behavior. It is vastly employed in various industrial applications due to its fast convergence and easy to carry out. Based on the analysis of current existing PSO algorithms, a Hybrid Improved PSO (HIPSO) is proposed in this paper, in which chaos initialization is introduced to improve the population diversity, and adaptive parameters' control strategy is employed to make it independent from specific problem, besides, novel acceptance policy based on Metropolis rule, which comes from Simulated Annealing, is taken to guarantee the convergence of the algorithm. In order to verify the effectiveness of the HIPSO, two typical numerical benchmarks are employed for comparison study with the other 3 well-known PSOs. Statistical optimization results show that, the new proposed HIPSO has outperformed the other PSOs, either on solution optimality, or on convergence speed.