累加分數管制法:操作特性及設計策略之研究

A Cumulative Score Control Scheme for detecting process shift: Operating Characteristic and Design Strategy

黃一哲(I-Che Huang)

指導教授 ： 鄭春生

元智大學/工程學院/工業工程與管理學系/碩士(1993年)

摘要

統計製程管制為一連線之品質管制方法，其主要目的在儘快偵測出可歸屬
原因之發生，減少製程之變異使產品達到所預定之目標值，完成以經濟之
方法製造高品質的產品。管制圖為統計製程管制中之一重要工具，其原理
為根據抽樣產品品質之數據資料變化來決定製程是否在管制狀態內。蕭華
特管制圖 (Shewhart charts) 是工業上最廣泛使用之管制圖。其優點為
容易使用而且比其他管制法能更快偵測出製程平均值的大量變動。蕭華特
管制圖只利用到最後一個觀測值來判斷製程是否在管制狀態內，這種特性
使得蕭華特管制圖對於製程小量變動之偵測較不靈敏。累加分數管制法為
輔助傳統蕭華特管制圖之管制程序。本文介紹一個新的累加分數管制法，
並探討各個設計參數，對於累加分數管制法偵測製程異常之影響。 本文
所提出之管制程序是以平均連串長度 (ARL) 來評估其成效，並以馬可夫
鏈之方法求解。我們的比較結果發現累加分數管制法較蕭華特—累和管制
法更具效率。

關鍵字

統計製程管制 ； 累加分數管制圖

並列摘要

Control charting is an important technique in statistical
process control.It is often used to determine if a
manufacturing process has achieved a state of control and
to detect the out-of-control situations. A process is in
control if no point falls outside the control limits or no
systematic variations are present in the process data. Over
the years, Shewhart control charts have been the most widely
used control scheme in industry. They are simple to use and can
detect large shifts faster than other cont ol schemes. A
criticism of Shewhart control schemes is that they only
use the information about the process contained in the
last observation. This feature makes Shewhart control charts
relatively insensitive to small shifts in the process mean.
In this research, a cumulative score control procedure based on
a single control statistic has been developed. The proposed
control procedure is designed to supplement the
traditional Shewhart charts in detecting small process
shifts. The ARL properties of the proposed schemes were
determined by a finite state Markov chain. An extensive
comparison shows that the ARL properties of the improved
cumulative score procedure are comparable to those of
Shewhart-CUSUM control schemes. The proposed scheme
required only simple algorithmetic and is very easy to
implement.

並列關鍵字

Statistical Process Contorl ； Cumulative Score Control Chart