多元正态向量均值变点在线监测

Monitoring Change in the Mean Vector of Multivariate Normal Distribution

中文关键词:  变点  残差  递归残差  在线监测

英文关键词:change-point  residuals  recursive residuals  online monitoring

中文摘要:

      在这篇文章中, 我们提出了监测多元独立向量均值变点的两种方法. 第一种方法是构造基于残差的CUSUM统计量; 第二种方法是构造基于递归残差的CUSUM统计量. 并分别得到了对应统计量的渐进分布.模拟表明本文提出的方法监测效果良好, 生产实例也说明该方法具有一定的实用性和有效性.

英文摘要:

      In this paper, we propose two monitoring schemes to monitor change in the mean vector of independent multivariate process after a period of size $m$. The first procedure is based on the CUSUM of residuals, and the second procedure employs the CUSUM of recursive residuals. The corresponding asymptotic distributions of the statistics are derived. Simulations show that the proposed monitoring procedures perform well. The empirical application illustrates the practicability and effectiveness of the procedures.