Activity Number:
|
385
|
Type:
|
Contributed
|
Date/Time:
|
Thursday, August 15, 2002 : 8:30 AM to 10:20 AM
|
Sponsor:
|
Section on Quality & Productivity*
|
Abstract - #300861 |
Title:
|
Robust Estimation in Multivariate Control Charts for Individual Observations
|
Author(s):
|
Jose Vargas*+
|
Affiliation(s):
|
National University of Colombia
|
Address:
|
Calle 69 No 5-59 Ap. 704, Bogota, International, , Colombia - South America
|
Keywords:
|
T2 Control Chart ; MVE estimators ; Outliers
|
Abstract:
|
Multivariate control charts are used for monitoring several quality characteristics measured simultaneously on a product or process. When a set of individual observations is initially collected, construction of Hotelling's T2 control chart in the retrospective stage of Phase I involves the estimation of the in-control process parameters and the comparison of the T2 values with an appropriate upper control limit. If the chart shows one or more out-of-control points, a search for assignable causes is carried out. However, when there are multiple outliers and the mean and covariance matrix are estimated by the usual sample mean vector and sample variance-covariance matrix, the T2 control chart might not give signals of alarm. This paper proposes for this stage the use of T2 control charts based on robust estimates of parameters. Simulation studies show that the T2 control chart based on the minimum volume ellipsoid (MVE) estimators of location and scatter is very effective in signaling the presence of outliers.
|