A Particle Swarm Optimization Algorithm Based on Adaptive Periodic Mutation

In this paper, in order to overcome the premature convergence and low stability of the particle swarm optimization (PSO) algorithm, a particle swarm optimization algorithm based on adaptive periodic mutation (APMPSO) is introduced. In the algorithm, the fitness uniformity distribution of the objective function, which can maintain the diversity of the population, is adaptively adjusted. The strategy is that the average fitness distance directly compares with the sort ascending fitness distance of the particles, in which the threshold constant can be avoided as much as possible, which may profoundly influence the stability of the algorithm. Moreover, the inertia weight with periodic mutation is proposed to update the velocity of the particles, which can improve the capability of local search and the stability of the algorithm. The improved algorithm is tested via a few benchmark functions in some simulations, the experiment results show that it not only has high global convergence precision and well stability, but also can prevent prematurity.