Multi- Swarm and Multi- Best particle swarm optimization algorithm
Junliang Li, Xinping Xiao · 2008
This paper proposes a novel particle swarm optimization algorithm: Multi-Swarm and Multi-Best particle swarm optimization algorithm. The novel algorithm divides initialized particles into several populations randomly. After calculating certain generations respectively, every population is combined into one population and continues to calculate until the stop condition is satisfied. At the same time, the novel algorithm updates particles’ velocities and positions by following multi-gbest and multi-pbest instead of single gbest and single pbest. The novel algorithm is not only a generalization of the basic particle swarm optimization, but can improve the searching efficiency, help the algorithm fly out of local optimum and increase the possibility of finding the real global best solution greatly. Finally one example is simulated to show the novel algorithm’s superiority.