Study on Design of Median and IQR Control Charts for Monitoring Location and Dispersion

• Main Author: 姜亭安
• Other Authors: 楊素芬
• Format: Others
• Language: en_US
• Published: 2016
• Online Access: http://ndltd.ncl.edu.tw/handle/95428492037993908801

碩士 === 國立政治大學 === 統計學系 === 104 === Control charts are effective tools for monitoring the process parameters in manufacturing processes and other industries. The development of distribution-free charts is important for non-normal or unknown distributed quality variable in statistical process control. The distribution-free control charts based on nonparametric statistics are not easy for practitioners to apply because they are not statisticians and do not know the scheme. This paper proposes a simple EWMA median chart and IQR char with single sampling scheme and double sampling scheme to monitor the location and dispersion, respectively. Furthermore, a kernel control region is proposed for monitoring the location and dispersion simultaneously. The average run lengths (ARL) is used to measure the detection performance of the proposed control chart(s). We compare the location and dispersion detection performance of the proposed charts and those of some existing distribution-free charts. An example of service times is used to illustrate the application of the proposed EWMA median and EWMA IQR charts and kernel control region. The proposed charts show superior detection performance compared to the existing distribution-free location and dispersion charts when the shifts in process location and/or dispersion are small. The SS EWMA-Md and DS EWMA-D charts and SS kernel control region are thus recommended.