Enhanced exponentially weighted moving average (EWMA) control chart performance with autocorrelation
This research introduces an enhanced exponentially weighted moving average (EWMA) control chart which is effective in detecting small and unnoticed shifts in monitoring process mean for autocorrelated data. The control chart is based on extension or modification of EWMA control chart statistic. The...
Saved in:
| Main Author: | |
|---|---|
| Format: | Thesis |
| Language: | English |
| Published: |
2015
|
| Subjects: | |
| Online Access: | http://eprints.utm.my/id/eprint/54897/1/AbbasUmarFaroukPFS2015.pdf http://eprints.utm.my/id/eprint/54897/ |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| Summary: | This research introduces an enhanced exponentially weighted moving average (EWMA) control chart which is effective in detecting small and unnoticed shifts in monitoring process mean for autocorrelated data. The control chart is based on extension or modification of EWMA control chart statistic. The proposed control chart is named the new EWMA (NEWMA) and is applied to simulated autocorrelated data for different autocorrelation levels (low, moderate and large) to study the performance of the chart. The run rules schemes were introduced to enhance the performance of the NEWMA chart namely; three out of three and three out of four schemes and three out of four is the best among the schemes. The NEWMA chart performs for observations that are autocorrelated. The NEWMA control chart has been tested on 100,000 simulations and it is found that it is quick in detecting process shift and able to identify the out of control points as it occurs. The performance of the technique has been evaluated using the average run length (ARL) and compared with modified exponentially weighted moving average (MEWMA) and classical exponentially weighted moving average (CEWMA) control charts and found that the NEWMA chart is faster in detecting shift. The NEWMA chart was applied to the KLSE Share index data, water quality data and Malaysian labour force data which are autocorrelated in nature and found to be effective in detecting the shifts. |
|---|