A modified PSO algorithm with exponential decay weight

Because of the convergence speed and the simple computation, the Particle Swarm Optimization (PSO) has been developing quickly, and many variants of the PSO have been proposed. By using some strategies, the major aim for improving the PSO is to obtain the global search ability at the early search stage and the better local search performance at the later stage. This paper proposed a modified PSO with exponential decay weight. By introducing a constraint factor to the velocity updating equation of the original PSO, and adopting the exponential decay mode for the inertia weight, a well global search ability at the early stage of the optimization procedure and a high local search performance at the later period can be obtained. We evaluate our algorithm with four benchmark functions and analyze the contributions of the exponential decay mode. For three common measures indices: convergence speed, convergence stability, and optimization success times, the experimental results show that the modified algorithm outperforms favorably against the famous improved PSO such as linear weight PSO and the constraint factor PSO.