Loss-function-based control chart for AR(1) processes | |
| Author | Virojana Tantibadaro |
| Call Number | AIT Diss. no.ISE-04-01 |
| Subject(s) | Process control--Statistical methods Autocorrelation (Statistics) |
| Note | A disse1iation submitted in partial fulfillment of the requirements for the degree of Doctor of Engineering, School of Advanced Technologies |
| Publisher | Asian Institute of Technology |
| Series Statement | Dissertation ; no. ISE-04-01 |
| Abstract | The prime objective of this study is to answer two questions: what are the effects of autocorrelated data on the performance of control charts for detection of mean shifts and how such effects can be reduced or eliminated. To accomplish the main objective, the study is divided into three major parts. The first part of the study is to develop a fundamental understanding of the effects of autocorrelation on ARL performances of the standard control charts for both univariate and bivariate cases. In the bivariate case, the accuracy between two different methods for estimating covariance matrix of a process is evaluated. It is recommended that the first estimator should be used especially when the process observations are suspected to be serially dependent. The second pait of the study aims to investigate and evaluate ARL performances of control charts with adjusted control limits to account for AR(l) correlation as well as to attain the desire ARL0. The results for the univariate charts indicate that ARL performances of the charts deteriorates as a level of AR( 1) correlation increases. In bivai"iate case, ARL performances of the Shewhart charts for AR(l) data and for independent data are not different. Similar to the univariate case, as the level of AR(l) increases, ARL performances of the CUSUM and EWMA charts for bivariate case drastically deteriorate. The comparisons on ARL performances of the charts for positive AR(l) processes in both univariate and bivariate cases are also included in this study. In the last pait of this dissertation, the cornerstone of the study, an alternative control scheme for detection of mean shifts in AR(l) processes is introduced. Not only are the fundamental elements of the proposed chart based on Bayesian concept explored, but algorithm for implementation and technical guidelines to choosing control parameter of the chait are also presented. The experimental results on evaluating ARL performance and sensitivity of the chart indicate that if the suitable control parameter is chosen, the chait will yield significantly improved ARL performance compared to other control charts. |
| Year | 2004 |
| Corresponding Series Added Entry | Asian Institute of Technology. Dissertation ; no. ISE-04-01 |
| Type | Dissertation |
| School | School of Advanced Technologies (SAT) |
| Department | Department of Industrial Systems Engineering (DISE) |
| Academic Program/FoS | Industrial Systems Engineering (ISE) |
| Chairperson(s) | Voratas Kachitvichyanukul; |
| Examination Committee(s) | Huynh Ngoc Phien;Anulark Techanitisawad;Tsung, Fugee; |
| Scholarship Donor(s) | The Royal Thai Government (RTG); |
| Degree | Thesis (Ph.D.) - Asian Institute of Technology, 2004 |