A Dynamical Particle Swarm Algorithm with Dimension Mutation

In this paper, a dynamical particle swarm algorithm with dimension mutation is proposed. First, we design a dynamically changing inertia weight based on the degree of both the particle diversity and the improvement of the best solutions in the successive generations. By using this inertia weight the algorithm can more easily keep the diversity of the population, improve the convergent speed. Second, in order to escape from the local optimum easily, a dimension mutation operator is designed. This mutation operator can easily jump out the local optimum from the dimension with the minimal convergence degree (i.e., the dimension in which the particles focus to the center of the search region most). Finally, the simulation experiments are made and the results indicate the high efficiency of the proposed algorithm