·408· 北京科技大学学报 1998年第4撕 当多元累积和图显示异常时，我们应用前述步骤找出异常因素x,图2(a),(b),(c,(是 其相应的控制图. 25 (a) 00 2(T (b) 0 20 74 1=13.5 15 10 UCL=12.85 UCL=10.6 0 20 10 (c) 15 13.5 13.5 10 LCL=10.6 UCL=10.6 ● 5 10 0 10 15 样本号 样本号 图2选控图及其累积和图：(a)T;(b)T;(c)T;(d)T 4 结论 本文论述了多元累积和控制图的有关内容，提出了两种质量多元累积和控制图的诊断理 论，并与多元T控制图进行了比较，结果表明多元累积和控制图在诊断过程均值小变动方面 比多元T控制图灵敏.本文的实例证实了本文理论与实践是一致的. 参考文献 I Woodall W H.Ncube MM.Multivariate CUSUM Quality Control Procedures.Technometnes.1985. 27:285 2 Healy J D.A Note on Multivariate CUSUM Procedures.Technometrics,1987.29:409 3 Pignatiello J J,Runger G C.Comparisons of Multivariate CUSUM Charts.Joumnal of Quality Technology,1990.22:173 4 Zhang Gongxu.Diagnosing Quality:Theory and Practices.Denmark:MAPP,1996 5张公绪，两种质量诊断理论与工序控制图设计.见：中国质量管理协会第二次年会论文，桂林.1然2 6张公绪选挖图理论与实践北京：人民邮电出版社，1984 Multivariate Cusum Control Chart and Its Diagnosis Theory Liu Yanyong Zhang Gongxu Management School.UST Beijing,Beijing 100083.China ABSTRACT Discussed the multivariate cusum,proposed the diagnosis theory with two kinds of cusum chart and compared with the multivariate T-chart.With a case study as a example is proposed. KEY WORDS quality controls/multivariate CUSUM;multivariate cause-selecting CUSUM; diagnosis theory with two kinds of quality;average run length