|
" Fault Identification in Multivariate Processes Using the Analysis of Multiple Univariate Statistical Quality Control Charts Along With T2 Multivariate Quality Control Charts Based on the Effect of the Underlying Variances and Covariances "
Alothman, Hussam Ali
Nagarur, Nagendra N.
| Document Type
|
:
|
Latin Dissertation
|
| Language of Document
|
:
|
English
|
| Record Number
|
:
|
1052393
|
| Doc. No
|
:
|
TL51510
|
| Main Entry
|
:
|
Alothman, Hussam Ali
|
| Title & Author
|
:
|
Fault Identification in Multivariate Processes Using the Analysis of Multiple Univariate Statistical Quality Control Charts Along With T2 Multivariate Quality Control Charts Based on the Effect of the Underlying Variances and Covariances\ Alothman, Hussam AliNagarur, Nagendra N.
|
| College
|
:
|
State University of New York at Binghamton
|
| Date
|
:
|
2019
|
| Degree
|
:
|
Ph.D.
|
| student score
|
:
|
2019
|
| Note
|
:
|
242 p.
|
| Abstract
|
:
|
Quality represents a very important attribute of any product, manufacturing system or in general, any other system in industry. Whether quality is interpreted as the conformance to specifications, the inverse of variability, or any other interpretation the quality might have, it is always important to control quality. Statistical process control is one of the main areas of quality control in which control charts in their different types are widely used. Most of the modern industry applications use advanced technologies like the internet of things (IoT) which involve the use of large number of connected devices and sensors that result in large amounts of data in terms of size and number of quality characteristics. To deal with such multivariate processes, multivariate statistical process control techniques including the multivariate control charts are highly needed. In this work, an overview of statistical quality control charts is presented for both univariate and multivariate control cases. In multivariate processes, using only univariate control charts like the xbar/R, xbar/S or I/MR charts can be misleading. However, on the other hand, there are cases where multivariate control like the T2 charts can show that processes are in control while in fact, one or more variates might be out of control on the individual level. So, in this work, the different scenarios where this can occur are discussed in details for the case with two variates or bivariate case and the case with three variates along with corresponding decision tress or decision roles. In addition, the general case with any number of variates is discussed. The effect of the values and signs of covariance(s) between the different variates and variances of these variates is explained in details and is supported by several case studies. Thus, this work shows the need to use multiple univariate control charts along with T2 multivariate control charts when dealing with multivariate processes in such cases where these covariance(s) and variances make it possible that T2 control chart is not able to detect out of control situations. This becomes more important when dealing with big amounts of data as missing out of control situations might be costly in terms of money, time, resources and efforts and thus it is very important to capture out of control situations quickly and accurately.
|
| Descriptor
|
:
|
Industrial engineering
|
|
|
:
|
Systems science
|
| Added Entry
|
:
|
Nagarur, Nagendra N.
|
| Added Entry
|
:
|
State University of New York at Binghamton
|
| |