Particle swarm optimization algorithms with novel learning strategies

This paper proposes three versions of particle swarm optimizers (PSO) with novel learning strategies where each dimension of a particle learns from just one particle's historical best information, while each particle learns from different particles' historical best information for different dimensions for a few generations. These strategies ensure that the diversity of the swarm is preserved to discourage premature convergence. In addition, these novel PSO variants do not introduce any complex computations to the original PSO algorithm. We obtain outstanding performance on solving multimodal problems in comparison to several other variants of PSO.