Degree Type

Dissertation

Date of Award

1991

Degree Name

Doctor of Philosophy

Department

Statistics

First Advisor

Stephen B. Vardeman

Abstract

Four aspects of statistical monitoring and control of manufacturing processes are studied. First a machining process is modeled using a random walk observed with error and adjusted in discrete steps. An optimal adjustment policy is derived to minimize the expectation of variable off target costs plus fixed adjustment costs. Under some regularity conditions the optimal policy is shown to make nonzero adjustments only when the process is perceived to be substantially off target;A more common control objective is to minimize process variance. Monitoring techniques are studied for detecting abrupt changes in autoregressive moving average transfer function (ARMAX) systems under minimum variance feedback control. An example shows that a simple cumulative sum (CUSUM) monitoring scheme performs very favorably in comparison to several other schemes for detecting an underlying step shift in the process level;Properties of a likelihood ratio based monitoring scheme for ARMAX systems can be investigated using a Markov chain to approximate the scheme's stochastic behavior. A general approach is described for approximating signaling time distributions for such monitoring schemes possessing a certain recursive calculation structure;Finally, concepts and an application of algorithmic statistical process control (ASPC) are presented. ASPC refers to the use of feedforward and feedback techniques to reduce predictable quality variations in conjunction with statistical process monitoring to detect and remove root causes of unpredictable quality changes. The application describes the development of a minimum variance control algorithm and a CUSUM monitor for a polymerization process at the General Electric Company. The application resulted in a 35% reduction in off specification material as well as several fundamental process improvements attributable to signals from the CUSUM monitor.

DOI

https://doi.org/10.31274/rtd-180813-9250

Publisher

Digital Repository @ Iowa State University, http://lib.dr.iastate.edu/

Copyright Owner

Scott A. Vander Wiel

Language

en

Proquest ID

AAI9126263

File Format

application/pdf

File Size

245 pages

Share

COinS