架構在權重相關法上之品管圖辨認系統

On Control Chart Pattern Recognition System with A Weighted Correlation Method

林文強(Wen-Qiang Lin)

指導教授 ： 楊敏生

中原大學/理學院/應用數學研究所/碩士(2012年)

https://doi.org/10.6840/cycu201200346

摘要

中文摘要
圖形辨識技術已被廣泛運用在識別不尋常的品項管制圖中，
一般而言，一套好的品管圖辨識系統都能夠對不同類型的異常做
出識別。然而，多數的管制圖應用中，正常點可能會出現在異常
點之前，因此從正常點到異常改變點都有可能發生在管制圖中，
所以如果我們沒有一個可以辨識這種變化的機制，這樣可能會導
致我們得到錯誤的分類結果。在這篇論文中，我們將使用加權的
方式來加強統計中的相關係數法，透過楊鎮槐與楊敏生[15]在
2005 年提出的管制圖識別系統，我們將使用加權的計算方法改善
並提高品管圖的識別精確度。我們將使用幾個比較的例子，並且
在最後的比較結果中，我們發現所提出的加權相關系數計算法確
實能提高品管圖的辨識精確度。

關鍵字

管制圖

並列摘要

Abstract
Pattern recognition techniques have been widely applied to identify
unnatural patterns in control charts where most of them are capable of
recognizing a single unnatural pattern for different abnormal types. In most
real control chart applications, normal points may appear before abnormal
points so that a change point from normal to abnormal may occur at any point
in control charts. If we do not have a mechanism for recognizing such change
patterns, we may incorrectly obtain the classification results. This paper
presents a better mode by using a weighted way to reinforce the statistical
correlation coefficient method. Then we consider the control chart pattern
recognition system by Yang and Yang [15] proposed in 2005. Using our new
method to calculate the statistical correlation coefficient during different time
points can improve the control chart pattern recognition system. Have
examples are considered. From comparison results, we can see that our
weighted correlation coefficient method actually presents better accuracy than
Yang and Yang [15] (2005).

並列關鍵字

control chart

參考文獻

[1] A. M. Al-Ghanim, L. C. Ludeman, Automated unnatural pattern recognition

on control charts using correlation analysis techniques, Computers and

[2] C. S. Cheng, A neural network approach for the analysis of control chart

[4] R. S. Guh, Y. C. Hsieh, A neural network based model for abnormal pattern

(1999) 97–108.