Linear Profile Monitoring Based on Nonparametric Spatial Rank Method

• Main Authors: Lin, Jian Chi, 林建錡
• Other Authors: Huwang, Longcheen
• Format: Others
• Language: zh-TW
• Published: 2016
• Online Access: http://ndltd.ncl.edu.tw/handle/06527663861270167163

碩士 === 國立清華大學 === 統計學研究所 === 104 === Nowadays, there are many production processes. In some condition, quality characteristics of a product in production can be described by the functional relationship between response and explanatory variable and then the relationship is called profile and the method of monitoring profile model is briefly called profile monitoring. The parameters in profile model can properly reveal the quality of a production process. Therefore, we can improve the capability of a process in the way of monitoring the shift of profile model parameters. When it comes to the traditional statistical control chart for profile monitoring, the error term in profile model usually is assumed to be normal distributed which is not satisfied with in most cases. Hence, we suggest a nonparametric control chart to monitor profile production. Also, we estimate the parameters in profile model in the nonparametric way. Meanwhile, we present the performance of our suggested control chart through statistical simulation. We also use out-of-control average run length to compare another nonparametric control chart which has proposed. Finally, we apply our proposed control chart to the Aluminum Electrolytic Capacitor production which is as a practical example.