Improved particle swarm algorithm with dynamic adjustment basing on velocity information

Particle swarm optimization algorithm is a simple and effective modern optimization algorithm, but it has the problem of being prone to premature and its convergence rate is slow. A new improved PSO algorithm is hence proposed. In the iteration of the proposed algorithm, the particles are distinguished to be active or stable according to their velocity information. For the active particles, to maintain the diversity of population, they are replaced by selection from its previous generation´s position and its reverse point, which combines the strategy of opposition-based learning. While for the stable particles, to enhance the convergence speed and increase the local search capability, they are improved by conjugate gradient method. The proposed improved PSO algorithm has come through numerical experiments of classic test functions. The results showed that, compared with other improved algorithms, this proposed improved PSO algorithm is feasible and effective.