基于小生境的极性PSO多目标优化研究及应用

摘要

This paper puts forward an improved particle swarm optimization algorithm, in which a niche technique with subgroup was applied in polarity particle swarm algorithm. In the new algorithm each subgroup individual evolutions separately with polarity acceleration added to the internal particle swarm evolution process. The transfer of social information happens only in the boundary. As a result, not only the convergence of the subgroup was ensured, but also the global diversity was increased. The improved particle swarm algorithm combined with multi-objective optimization was then applied in the setting of parameters in the steam generator level controller. Simulation results show that the system using the new algorithm proves more stable, accurate and faster at the same time has strong robustness compared with that using traditional parameter setting method.%提出一种改进粒子群算法,即将子群优化的小生境技术应用于极性粒子群算法,每个子群单独进化,内部粒子群进化增加了极性加速度,仅在边界处进行社会信息的传递.这样既保证了子群内部的有效收敛,又增加了全局多样性.将改进的粒子群算法与多目标优化相结合应用于蒸汽发生器的液位控制器参数整定,仿真结果表明,应用该算法的系统特性与使用传统参数整定法相比,更加稳定、准确和快速,具有较强的鲁棒性.