Both manufacturing and service industries deal with quality characteristics, which include not only variables but attributes as well. In the area of Quality Control there has been substantial research in the area of correlated variables (i.e. multivariate control charts); however, little work has been done in the area of correlated attributes. To control product or service quality of a multi-attribute process, several issues arise. A high number of false alarms (Type I error) occur and the probability of not detecting defects increases when the process is monitored by a set of uni-attribute control charts. Furthermore, plotting and monitoring several uni-attribute control charts makes additional work for quality personnel.
To date, a standard method for constructing a multi-attribute control chart has not been fully evaluated. In this research, three different techniques for simultaneously monitoring correlated process attributes have been compared: the normal approximation, the multivariate np-chart (MNP chart), and a new proposed Neural Network technique. The normal approximation is a technique of approximating multivariate binomial and Poisson distributions as normal distributions. The multivariate np chart (MNP chart) is base on traditional Shewhart control charts designed for multiple attribute processes. Finally, a Backpropagation Neural Network technique has been developed for this research. Each technique should be capable of identifying an out-of-control process while considering all correlated attributes simultaneously.
To compare the three techniques an experiment was designed for two correlated attributes. The experiment consisted of three levels of proportion nonconforming p, three values of the correlation matrix, three sample sizes, and three magnitudes of shift of proportion nonconforming in either the positive or negative direction. Each technique was evaluated based on average run length and the number of replications of correctly identified given the direction of shifts (positive or negative). The resulting performances for all three techniques at their varied process conditions were presented and compared.
From this study, it has observed that no one technique outperforms the other two techniques for all process conditions. In order to select a suitable technique, a user must be knowledgeable about the nature of their process and understand the risks associated with committing Type I and II errors. Guidelines for how to best select and use multi-attribute process control techniques are provided.
| Identifer | oai:union.ndltd.org:PITT/oai:PITTETD:etd-09302003-180635 |
| Date | 02 February 2004 |
| Creators | Larpkiattaworn, Siripen |
| Contributors | Mary E. Besterfield-Sacre, Harvey Wolfe, Jerrold H. May, Mainak Mazumdar, Kim L. Needy |
| Publisher | University of Pittsburgh |
| Source Sets | University of Pittsburgh |
| Language | English |
| Detected Language | English |
| Type | text |
| Format | application/pdf |
| Source | http://etd.library.pitt.edu/ETD/available/etd-09302003-180635/ |
| Rights | unrestricted, I hereby certify that, if appropriate, I have obtained and attached hereto a written permission statement from the owner(s) of each third party copyrighted matter to be included in my thesis, dissertation, or project report, allowing distribution as specified below. I certify that the version I submitted is the same as that approved by my advisory committee. I hereby grant to University of Pittsburgh or its agents the non-exclusive license to archive and make accessible, under the conditions specified below, my thesis, dissertation, or project report in whole or in part in all forms of media, now or hereafter known. I retain all other ownership rights to the copyright of the thesis, dissertation or project report. I also retain the right to use in future works (such as articles or books) all or part of this thesis, dissertation, or project report. |
Page generated in 0.0117 seconds