论文部分内容阅读
变点识别是统计过程控制第一阶段的主要任务,通过对大量历史数据进行分析,识别出运行状态发生变化的准确时间点,则有利于分析运行状态变化的原因,从而改善生产等各种过程的表现。针对时间顺序数据,本文提出将划分层次聚类法与凝聚层次聚类法相结合的变点识别方法,并在凝聚层次聚类中采用非参数检验法。通过仿真对方法性能进行分析和比较,结果显示本方法具有良好的变点识别性能。
Change-point identification is the main task of the first stage of statistical process control. By analyzing a large amount of historical data and identifying the exact time point when the operating status changes, it is helpful to analyze the reasons for the change of operating status so as to improve various processes such as production Performance. According to the chronological data, this paper proposes a change-point identification method combining hierarchical clustering method and cohesive hierarchical clustering method, and adopts nonparametric test method in cohesive hierarchical clustering. The performance of the method is analyzed and compared through simulation. The results show that this method has a good performance of changing point identification.