有限相依随机序列的非贝叶斯变点最优监测  

The Optimality in Non-Bayesian Change-Point Detection for Finite Observation Sequence

作　　者：韩东 宗福季[2,3] HAN Dong;TSUNG Fugee(Department of Statistics,School of Mathematical Sciences,Shanghai Jiao Tong University,Shanghai,200240,China;Department of Industrial Engineering and Decision Analytics,The Hong Kong University of Science and Technology,Hongkong;Data Science and Analytics Thrust,The Hong Kong University of Science and Technology(Guangzhou),Guangzhou,520700,China)

机构地区：[1]上海交通大学数学科学学院统计系,上海200240 [2]香港科技大学工业工程及决策分析学系,中国香港 [3]香港科技大学(广州)数据科学与分析系,广州510700

出　　处：《应用概率统计》2024年第2期277-286,共10页Chinese Journal of Applied Probability and Statistics

基　　金：supported by the National Natural Science Foundation of China(Grant Nos.72371271,11531001).

摘　　要：本文研究了有限相依样本序列的非贝叶斯变点检测问题.通过引入非负动态随机控制线,我们不仅构造并证明了两个最优控制图,而且还得到了比原定义更容易计算的Lorden测度和Pollak测度的最小值的表达式.This paper studies the non-Bayesian change-point detection of finite dependent samples sequence.By presenting the nonnegative dynamic random control limits,we not only constructed and proved two optimal control charts,but also obtained the expressions for the minimum values of Lorden’s measure and Pollak’s measure that are easier to calculate than the original definition.

关 键 词：最优控制图 非贝叶斯变点监测 相依样本序列 

分 类 号：O213.1[理学—概率论与数理统计]