Abstract: The standard particle swarm optimization in the evolution formula only considers both the colony's best fitness value and the individual's fitness values. Therefore, it leads to the low accuracy of the convergence on account of being lack of the diversity in later evolution period. In order to improve the accuracy of the algorithm, the paper proposed the core master group particle swarm, and combined a core master group particle swarm with the improved formula. The improved algorithm is proved through experiments to be more accurate.

Key words: Particle Swarm Optimization (PSO), diversity, core master group, fitness value, convergence

摘要： 标准的粒子群算法在其进化的公式中,只是考虑了群体最佳的适应度值和个体最佳适应度值,这导致了标准的粒子群算法在算法的进化后期由于缺乏多样性收敛精度不高。为了提高算法的精度,提出了核心主子群粒子群算法,并将提出的核心主子群算法与改进的公式相结合。通过实验证明,改进算法使得所求结果的精度有进一步的提高。

关键词: 粒子群优化, 多样性, 核心主子群, 适应度值, 收敛