This paper proposes a new variant of particle swarm optimization, namely adaptive particle swarm optimization with population diversity control (APSO-PDC), to improve the performance of particle swarm optimization. APSO-PDC is formulated based on adaptive selection of particle roles, population diversity control, and adaptive control of parameters. The adaptive selection of particle roles which combines the evolutionary state and dynamic particle state estimation method will sort the particles into three roles to let different particles execute different search tasks during optimization process. The adaptive control of parameters which is created based on the evolutionary state and particle roles encourages the exploitation ability and enhances the algorithm’s convergence speed. The population diversity control which combines comprehensive learning strategy of the comprehensive learning particle swarm optimizer with evolutionary state to update the individual best position strengthens exploration ability and thus increases the algorithm’s robustness toward the premature convergence issue. The performance of APSO-PDC is comprehensively evaluated by 21 unimodal and multimodal functions with or without rotation. The results indicate APSO-PDC has more preferable searching accuracy, searching reliability, and convergence speed than the other well-established particle swarm optimization variants. Finally, compared with other six particle swarm optimization variants, APSO-PDC shows satisfactory performance in optimizing tandem blade. This excellent performance proves that APSO-PDC has a better control of swarm exploration and exploitation abilities.
Power load forecasting algorithm based on nonlinear inertial factor change pattern particle swarm optimization algorithm
