An Enhanced Quality Evaluation System for Continuous Manufacturing Processes, Part 1: Theory

Abstract

A quality evaluation system is presented which gives enhanced information about the nature of quality changes in continuous manufacturing processes. The system monitors a quality characteristic of the process via stochastic time series models and detects shifts in the process mean, variance, and dynamic parameters. A rule base classifies the type of change that occurred, and change magnitude and time of occurrence are estimated. Part 1 of the work addresses the theoretical issues concerning the statistics used by the rule base and their behavior under various process changes. The characteristics of the cusum chart, autocorrelation chart, and Chi-Square test are derived with respect to each of the possible change mechanisms.