改进型单边EWMA指数控制图设计  

Design of An Improved One-Sided Exponential EWMA Chart

作　　者：胡雪龙[1] 张素颖 孙金生[2] 唐安安 乔宇龙 HU Xue-long;ZHANG Su-yin;SUN Jin-sheng;TANG An-an;QIAO Yu-ong(Institute of High-Quality Development Evaluation,Nanjing University of Posts and Telecommunications,Nanjing 210003,China;School of Automation,Nanjing University of Science and Technology,Nanjing 210094,China)

机构地区：[1]南京邮电大学高质量发展评价研究院,江苏南京210003 [2]南京理工大学自动化学院,江苏南京210094

出　　处：《数理统计与管理》2023年第1期26-34,共9页Journal of Applied Statistics and Management

基　　金：国家自然科学基金(71802110,71872088,71671093);江苏省自然科学基金(BK20200750);江苏高校哲学社会科学研究项目(2020SJA0090);江苏高校哲学社会科学重点基地项目资助-信息产业融合创新与应急管理研究中心;江苏省哲学社会科学优秀创新团队(2017ZSTD022);江苏省研究生科研与实践创新计划项目(KYCX21_0840);南京邮电大学(NY222176)。

摘　　要：现代制造过程中,某些产品的不合格率非常低,通常将这类过程称为高质量过程。由于高质量过程中相邻不合格产品之间的时间间隔服从指数分布,可以通过指数控制图实现对过程状态的监控。因此,本文提出一种改进型指数加权移动平均(Improved Exponentially Weighted Moving Average,IEWMA)指数控制图,并采用蒙特卡洛仿真获得控制图的平均运行链长(Average Run Length,ARL),仿真结果表明该控制图的性能优于传统单边EWMA指数控制图,尤其针对过程中产生较小偏移的情形具有较好的检测性能。In modern manufacturing processes,some products have very low non-conforming rates and these processes are often referred as high quality processes.Since the time between successive nonconforming samples in high quality process follows an exponential distribution,the process state can be monitored through exponential control charts.Therefore,this paper proposes an improved one-sided exponential EWMA(Exponentially Weighted Moving Average)control chart and obtains the chart’s ARL(Average Run Length)using Monte Carlo simulations.The simulation results show that the performance of the proposed control chart is better than the traditional one-sided exponential EWMA control chart,especially in the monitoring of small shifts.

关 键 词：单边指数加权移动平均控制图 指数分布 平均运行链长 蒙特卡洛仿真 

分 类 号：O212[理学—概率论与数理统计] O213.1[理学—数学]