Non Symmetrical Multivariate Statistical Methods for Quality Control and Batch Processes Monitoring
A brief outline of the evolution of the statistical techniques for the measurement of quality is firstly sketched. A critical overview of T2-based methods that require strong assumptions on the variables distribution is given. The need for multivariate procedures is justified from both the methodological and application point of view and a proposal of non parametric control charts is shown. The attention is then focused on visualising and monitoring quality rather than building control charts both at unit and batch level. The interpretation properties and visualisation facilities of Non Symmetrical Data Analysis and PLS-based methods are shown to fit quite naturally the demands arising when measuring and monitoring quality of multivariate industrial processes. Some research perspectives are finally enhanced. Lien vers l'article
LAURO, C. and ESPOSITO VINZI, V. (2002). Non Symmetrical Multivariate Statistical Methods for Quality Control and Batch Processes Monitoring. In: Analisi Multivariata per la Qualità Totale. 1st ed. Milano: FrancoAngeli, pp. 45-65.
Mots clés : #Quality-Control, #Batch-MonitoringNon-symmetrical-data-analysis, #Partial-Least-Squares-methods, #Visualization-techniques