An improved particle swarm optimization algorithm with neighborhoods topologies

The impacts of neighborhoods topologies on particle swarm optimization for complex functions are discussed. The suitability of three topologies to the kind of functions is experimented. The convergence features affected by constant parameters on particle swarm optimization are illustrated. The velocity and position disturbances are introduced to prevent premature phenomenon of the original algorithm. A valve is introduced and the selection criteria are discussed. Simulations have been carried and the results show this improved algorithm has better performance compared with original particle swarm optimization.