聚类分析是描述数据群体特征的有效方法.然而,随着挖掘数据库规模的增大,评分函数极值搜索是计算复杂度很高的问题.文中提出了一种新的算法,并将其运用到束流状态描述建模中.试验结果表明,该算法快速有效.同时,重复性是电子储存环的重要指标,聚类模型可作为机器研究和决策的依据,具有指导意义. Clustering analysis is an effective way to describe the characteristics of data population.However, as the size of mining database increases, the extremum search of scoring function is a computationally complex problem.In this paper, a new algorithm is proposed, and its application To the modeling of beam state description.Experimental results show that the algorithm is fast and effective.Repeatability is an important index of electronic storage ring, and clustering model can be used as the basis for machine research and decision-making, which has guiding significance.