A Comparative Study of Techniques for Identifying Out-Of-Control Variable(S) In Multivariate Control Chart on Cable Production

Abstract

Multivariate statistical process control charts are used for process monitoring and control of two or more variables simultaneously for quality and quality improvement. A popular multivariate control chart is used to monitor the mean vector of the process. A usual problem in the use of the multivariate control chart is the identification and interpretation of variable(s) responsible for out-of-control signal that occurred in the chart. This has brought many developed techniques from many researchers to aid in finding the variable(s) responsible for the out-of-control signal in the multivariate control chart. Thus, the work is aimed at a comparative study of some developed techniques for identifying and interpreting out-of-control signal in the multivariate control chart when applied to the cable production process. The techniques are Mason-Tracy-Young, Donganaksoy-Faltin-Tucker, Univariate -chart using Bonferroni control limits by Alt and Principal component analysis by Jackson. A performance criterion, the power of a test was used to ascertain the most satisfactory technique that explained the out-of-control signal that occurred in the multivariate control chart. From the results and discussions, Mason Tracy-Young and Doganaksoy-Faltin-Tucker techniques are the most satisfactory for identifying and interpreting out-of-control signal in the multivariate control chart for the datasets considered.