基于协变量信息的自适应MEWMA控制图  

Adaptive MEWMA Control Chart Based on Covariate Information

作　　者：张岩 朱永忠[1] ZHANG Yan;ZHU Yong-zhong(College of Scinence,Hohai University,Nanjing,Jiangsu 211100,China)

机构地区：[1]河海大学理学院,江苏南京211100

出　　处：《计算技术与自动化》2023年第1期39-45,共7页Computing Technology and Automation

基　　金：国家自然科学基金资助项目(11771120)。

摘　　要：在统计过程控制中,质量变量通常会受到许多协变量因素的影响,充分利用协变量的有用信息可以进一步提高控制图的灵敏度,因此提出一种新的自适应多元EWMA控制图,并计算ARL进行比较。在MEWMA控制图的基础上引入一个权重函数,根据协变量的有用信息自适应的调节统计量的加权参数:当收集到的协变量信息发生偏移时,选择较大的加权参数,更关注当前和附近时间点观测值的偏移程度;反之则选择较小的参数。大量数值模拟分析表明,充分利用协变量中的有用信息之后,监控效果明显优于MEWMA、MCUSUM控制图。In statistical process control,quality variables are usually affected by many covariates,and the sensitivity of control charts can be further improved by making full use of the useful information of covariates.Therefore,a new adaptive multivariate EWMA control chart is proposed and ARL is calculated for comparison.Based on the MEWMA control chart,a weight function was introduced to adjust the weighted parameters of the statistics adaptively according to the useful information of the covariable.When the collected covariable information was offset,a larger weighting parameter was selected to pay more attention to the offset degree of the observed values at the current and nearby time points.Otherwise,select a smaller parameter.A large number of numerical simulation analysis shows that the effect is obviously better than MEWMA and MCUSUM control charts after fully utilizing the useful information of covariates.

关 键 词：协变量 辅助变量 权重函数 MEWMA控制图 平均运行长度 

分 类 号：O212[理学—概率论与数理统计]