ENBIS: European Network for Business and Industrial Statistics

Phase I Monitoring of Multivariate Linear Profiles

22 September 2009, 16:55 – 17:15

Abstract

Submitted by

Rassoul Noorossana

Authors

R. Noorossana, M. Eyvazian, and A. Vaghefi

Affiliation

Industrial Engineering Department , Iran University of Science and Technology, Tehran, Iran 16846-13114

Abstract

Abstract:
In certain quality control applications, quality of a product or process can be effectively represented by a relationship between two or more variables. This relationship which is referred to as profile can be either linear or nonlinear in nature. Several researchers have investigated the cases of single linear or nonlinear profiles. In this paper, we consider the case where quality of a product or process can be effectively monitored through the simultaneous use of two or more linear profiles. Three different approaches referred to as multivariate T2 statistics approach, multivariate T2 statistics approach with coded x values, and principal component approach are proposed for assessing the stability of the model parameters. The performances of the proposed approaches are investigated numerically in terms of signal probability associated with different shift magnitudes in the model parameters.

Key Words and Phrases:
Multivariate linear profiles, Principal component analysis, Statistical process control, Wilks’ lambda, T2 statistic, Likelihood ratio test

Return to programme