Contributions to multivariate control charting: studies of the Z chart and four nonparametric charts

Loading...
Thumbnail Image
Date
2010
Journal Title
Journal ISSN
Volume Title
Publisher
University of Alabama Libraries
Abstract

Autocorrelated data are common in today's process control applications. Many of these applications involve two or more related variables so that multivariate statistical process control (SPC) methods should be used in process monitoring since the relationship among the variables should be accounted for. Dealing with multivariate autocorrelated data poses many challenges. Even though no one chart is best for multivariate data, the Z chart proposed by Kalgonda and Kulkarni (2004) is fairly easy to implement and is particularly useful for its diagnostic ability, which is to pinpoint the variable(s) that is(are) out of control in case the chart signals. In this dissertation, the performance of the Z chart is compared to the chi-square chart and the multivariate EWMA (MEWMA) chart in a number of simulation studies. Simulations are also performed to study the effects of parameter estimation and non-normality (using the multivariate t and multivariate gamma distributions) on the performance of the Z chart. In addition to the problem of autocorrelation in multivariate quality control, in many quality control applications, the distribution assumption of the data is not met or there is not enough evidence showing that the assumption is met. In many situations, a control chart that does not require a strict distribution assumption, called a nonparametric or distribution-free chart, may be desirable. In this paper, four new multivariate nonparametric Shewhart control charts are proposed. They are relatively simple to use and are based on the multivariate forms of the sign and Wilcoxon signed-rank statistics and the maximum of multiple univariate sign and Wilcoxon signed-rank statistics. The performance of these charts is also studied. Illustrations and applications are also demonstrated.

Description
Electronic Thesis or Dissertation
Keywords
Statistics, Industrial engineering
Citation