טכניון מכון טכנולוגי לישראל
הטכניון מכון טכנולוגי לישראל - בית הספר ללימודי מוסמכים  
M.Sc Thesis
M.Sc StudentShourky Keren
SubjectProduction Process Control by Multivariate Statistical
Process Control (MSPC)
DepartmentDepartment of Quality Assurance and Reliability
Supervisor Dr. Pnina Sason


Abstract

Statistical Process Control (SPC) is one of the most important and common methods in which a product or a service process control is conducted. The industry field is characterized with multivariable processes where the variables are usually correlated among themselves. In these cases using SPC is not sufficient enough.

This paper introduces a statistical tool which answers the problem described above: Multivariate Statistical Process Control (MSPC). The method chosen to monitor the process is controlling Hotelling's T2 statistic.  The basic idea of this method is to transform the number of the quality variables, which are being checked, into one single variable T2. This variable stands for the statistical distance between the observation vector in each checkpoint and the mean vector of those quality variables.

A quality control process is performed on the production of mixed nuts roasting, by measuring three quality variables of the product during the production process: percentage of moisture, percentage of salt, and a qualitative variable - product organoleptic  (looks, taste and smell).

The implementation of MSPC for this production process includes the formulation of a control chart and limits for the Hotteling’s T2, based on a set of historical data (HDS), applying new observations to the chart and examination of the outliers. Analysis of the distribution of the process variables, as well as that of T2 statistic, and the assumptions which have been taken, are explained in the paper.