The Distance-Guided Particle Swarm Optimizer with Dynamic Mutation

Abstract

This paper presents a distance-guided particle swarm optimizer with dynamic mutation (PSODM). Two characteristics are proposed in the PSODM: distance-guided and dynamic mutation. The goal of these characteristics is to overcome premature convergence of the swarm and accelerating the convergence velocity. With distance-guided operation, some particles too intense can be separated, while with dynamic mutation operation, the particles mutated can find a better position more easily by effectively using the foregone experience. The PSODM is compared to other four types of improved PSO, and the experimental results show that the PSODM performances better on a six-function test suite with different dimensions.