Particle Swarm Optimization (PSO) has been widely used to solve real world optimization problems. Since the original PSO fails to maintain the diversity along the search process, some PSO variations have been proposed to overcome this limitation. The recently introduced Adaptive Particle Swarm Optimization (APSO) presents faster convergence, while maintaing the population diversity, by including into the PSO algorithm the auto-adaptation of the parameters according to the current spatial

more »

... ution of the swarm. Besides, there are some PSO variations which incorporate a cooperative behavior, such as the Clan Particle Swarm Optimization (ClanPSO). In this paper, we propose to include the auto-adaptation ability solely in the Clans of the ClanPSO algorithm and compare its performance to previous approaches. We evaluated our proposal in six benchmark functions recently proposed in IEEE Congress on Evolutionary Computation 2010. Our proposal obtained similar or better results in terms of fitness when compared to the original ClanAPSO, but presents a lower computational cost. We obtained better results when compared to other cooperative approaches and the classical PSO approaches.