MONITORING AND IDENTIFICATION OF INFLUENTIAL PROCESS CHARACTERISTICS IN THE PRESENCE OF AUTOCORRELATION

No Thumbnail Available
Date
2015-12
Authors
ADEPOJU, AKEEM AJIBOLA
Journal Title
Journal ISSN
Volume Title
Publisher
Abstract
The traditional methods of multivariate statistical process control (MSPC) are primarily based on the assumptions that the successive observation vectors are independent and normally distributed.However, some process observations are found to be dependent (known as autocorrelation or serial correlation) and if the autocorrelation is left untreated, this can consequently lead to wrong monitoring decision as well as wrong variable identification in the case of out-of-control, which consequently affect the performance of the control charts. This thesis looked into the problem of monitoring the mean vector of a process embedded with autocorrelation and failure of normality assumption. In order to remove the autocorrelation effect and normalized the original data, we proposed vector autoregressive model,VAR model whose residual is assumed to be independent and Johnson transformation(JT) to transform the original data to normality. We were able to show and compare the effect of applying traditional Hotelling’s 2 T control chart on autocorrelated and non-normal data as against the residualsobtained from VAR (1) model and normally transformed data. However, since our intention is to achieve a better decision in industrial settings, we thereby complement this work by further adopted MYT model to decompose the overall contribution of the five process variables into individual contribution such that the influential variable(s) is/are identified
Description
A THESIS SUBMITTED TO THE SCHOOL OF POSTGRADUATE STUDIES, AHMADU BELLO UNIVERSITY, ZARIA NIGERIA IN PARTIAL FULFILLMENT FOR THE AWARD OF MASTER OF SCIENCE (M.Sc) DEGREE IN STATISTICS, DEPARTMENT OF MATHEMATICS AHMADU BELLO UNIVERSITY, ZARIA NIGERIA
Keywords
MONITORING,, IDENTIFICATION,, INFLUENTIAL PROCESS CHARACTERISTICS,, PRESENCE,, AUTOCORRELATION,
Citation
Collections