Charts for Monitoring the Auto correlated Processes
In this paper,we present two control charts combined with an exponentially weighted moving average (EWMA) procedure for monitoring the auto correlated data.One is a conventional parametric chart and the other is a nonparametric sign test.A number of autoregressive and moving average (ARMA) models is applied to generate the auto correlated data.For the case of ARMA with a random white noise from normal distribution,that is to say,the Gaussian process,2 and 4 have studied the performance of mean shifts detecting for the auto correlated data by using the famous HotellingsT2 chart.However,for the Non-Gaussian processes,suppose the random white noise is from some heavy-tailed or very skewed distributions,the performance may be highly affected.Simulated results show that our proposed charts have very good performance over HotellingsT2 chart for the unknown distributions,especially for the small and moderate shifts.
Auto correlated-Processes ARMA EWMA Procedure MEWMA MSEWMA
WANG Zhengang ZI Xuemin ZOU Changliang
School of Mathematical Sciences,Tianjin University of Technology and Education,P.R.China,300222 Department of Statistics School of Mathematical Sciences,Nankai University,P.R.China,300071
国际会议
青岛
英文
15-20
2012-07-20（万方平台首次上网日期，不代表论文的发表时间）