國立雲林科技大學 工業工程與管理所 課堂名稱:高等品質管理 授課老師:童超塵 副教授 國立雲林科技大學 工業工程與管理所 課堂名稱:高等品質管理 授課老師:童超塵 副教授 報告學生: 蔣步海 (9421812) 王聖權 (9221801) 林榮慶 (9421748) 張咸明 (9321809) 蔡世賢.

Slides:



Advertisements
Similar presentations
Basic concepts of structural equation modeling
Advertisements

Dr. Baokun Li 经济实验教学中心 商务数据挖掘中心
第4章 VHDL设计初步.
B型肝炎帶原之肝細胞癌患者接受肝動脈栓塞治療後血液中DNA之定量分析
多元迴歸 Multiple Regression
-Artificial Neural Network- Hopfield Neural Network(HNN) 朝陽科技大學 資訊管理系 李麗華 教授.
第三章 隨機變數.
Chapter 8 Liner Regression and Correlation 第八章 直线回归和相关
沐阳老年社区.
XI. Hilbert Huang Transform (HHT)
何正斌 博士 國立屏東科技大學 工業管理研究所 副教授
-Artificial Neural Network- Adaline & Madaline
Rate and Distortion Optimization for Reversible Data Hiding Using Multiple Histogram Shifting Source: IEEE Transactions On Cybernetics, Vol. 47, No. 2,February.
Population proportion and sample proportion
模式识别 Pattern Recognition
SPC introduction.
Differential Equations (DE)
SAS  統計程序實作 CONTENTS By DR. Yang , Yi-Chiang /11/11.
Digital Terrain Modeling
項目分析與探索式因素分析 李茂能, 2007,成大 Fred Li, 2007.
次数依变量模型 (Models for Count Outcomes)
作者:游明新 指導教授:童超塵教授 報告學生:吳志權
非線性規劃 Nonlinear Programming
On Some Fuzzy Optimization Problems
第五讲 数据的分组、合并与转换.
Continuous Probability Distributions
Properties of Continuous probability distributions
Sampling Theory and Some Important Sampling Distributions
Digital Terrain Modeling
Decision Support System (靜宜資管楊子青)
第十一章. 簡單直線迴歸與簡單相關 Simple Linear Regression and Simple Correlation
十一、簡單相關與簡單直線回歸分析(Simple Correlations and Simple Linear Regression )
第二十九單元 方向導數與梯度.
组合逻辑3 Combinational Logic
製程能力分析 何正斌 教授 國立屏東科技大學工業管理學系.
二元隨機變數(Bivariate Random Variables)
Tel: 第11章 SPSS在时间序列预测中的应用 周早弘 旅游与城市管理学院
Interval Estimation區間估計
作者: DALE GOODHUE 來源: JOURNAL OF THE AMERICAN SOCIETY FOR INFORMATION
Formal Pivot to both Language and Intelligence in Science
磁共振原理的临床应用 福建医科大学附属第一医院影像科 方哲明.
The Nature and Scope of Econometrics
Decision Support System (靜宜資管楊子青)
CH6 Pairs Selection in Equity Markets
高性能计算与天文技术联合实验室 智能与计算学部 天津大学
Mechanics Exercise Class Ⅰ
Guide to a successful PowerPoint design – simple is best
相關統計觀念復習 Review II.
Chapter 04 流程能力與績效分析.
Safety science and engineering department
線性規劃模式 Linear Programming Models
计算机问题求解 – 论题 算法方法 2016年11月28日.
Simple Regression (簡單迴歸分析)
10-6 CONTROL CHARTS FOR MONITORING VARIABLITY
The Bernoulli Distribution
高考应试作文写作训练 5. 正反观点对比.
Q & A.
计算机问题求解 – 论题1-5 - 数据与数据结构 2018年10月16日.
Chapter 10 Mobile IP TCP/IP Protocol Suite
Review of Statistics.
磁共振原理的临床应用.
微观经济学(第三版) 高鸿业 LECTURE11 市场失灵和微观经济政策.
名词从句(2).
动词不定式(6).
何正斌 博士 國立屏東科技大學工業管理研究所 教授
國際會計準則(IFRS)推動現況及因應之道
簡單迴歸分析與相關分析 莊文忠 副教授 世新大學行政管理學系 計量分析一(莊文忠副教授) 2019/8/3.
Principle and application of optical information technology
Gaussian Process Ruohua Shi Meeting
SAS 統計程序實作 PROC MEANS (一個母體)
Presentation transcript:

國立雲林科技大學 工業工程與管理所 課堂名稱:高等品質管理 授課老師:童超塵 副教授 國立雲林科技大學 工業工程與管理所 課堂名稱:高等品質管理 授課老師:童超塵 副教授 報告學生: 蔣步海 ( ) 王聖權 ( ) 林榮慶 ( ) 張咸明 ( ) 蔡世賢 ( )

Many situations require the simultaneous monitoring or control of two or more related quality characteristics For example, consider a bearing with both an inner diameter (x 1 ) and an outer diameter (x 2 ) that together determine the usefulness of a part Monitoring the characteristics independently can be very misleading Refer to Figures 10-1 and -2 Use of multiple independent charts distorts the simultaneous monitoring of the averages Type I error and probability of a point correctly plotting in control are not equal to advertised levels for the individual control charts Distortion in process-monitoring procedures increases as the number of quality characteristics increases to Pr{all p means plot in control} = (1   ) p (10-2)

variate

10-3 The Hotelling T 2 Control Chart mean vectorMonitoring the mean vector of the process A direct analog of the univariate Shewhart chart Two versions of the Hotelling T 2 Control Chart  Subgrouped data  Individual observations

Subgrouped Data We have

Control Ellipse

Disadvantage:  The time sequence of the plotted points is lost  Hard to deal with more than two quality characteristics Solution: chi-square control chart

Extend to p related quality characteristics => p-variate normal distribution

Lowry and Montgomery (1995): m are always greater than 20, and often more than 50

Quality Characteristics n=10 m=20

Interpretation of Out-of-Control Signals

Individual Observations Recall control chart The phase II control limits: Subgroup size The phase II control limits:

Individual Observations(cont.) (covariance is known) Note: Lowry and Montgomery (1995) show that the chi-square limit should be applied when Tracy, Young and Mason (1992):

Individual Observations(cont.) Sullivan and Woodall (1995), Vargas (2003) Holmes and Mergen (1993)

Covariance matrices:

10.4 The Multivariate EWMA control chart 多變量統計製程管制圖 ~ 多變量指數加權移動平均管制圖  1931 年以來 ,有學者針對多變量品質管制發表相關研究 例如: Hotelling’s 管制圖、 MCUSUM 、及 MEWMA 等 …. 例如: Hotelling’s 管制圖、 MCUSUM 、及 MEWMA 等 ….  比較管制圖的方法是利用平均連長度( ARL ) Wieringa(1998) 對 所做的推導,由時間序列 AR ( 1 ) Wieringa(1998) 對 所做的推導,由時間序列 AR ( 1 ) Model : 令 =0 ~N ( 0 , )符合標準常態下, EWMA 管制圖定義得 課本

多變量統計製程管制圖 ~ 多變量指數加權移動平均管制圖 推導管制界限為 : 假設製程在時間 t=T 時發生均值的偏移

多變量統計製程管制圖 ~ 多變量指數加權移動平均管制圖 之值 :

多變量統計製程管制圖 ~ 多變量指數加權移動平均管制圖

The quantity plotted on the control chart is Where the covariance matrix is

多變量統計製程管制圖 ~ 多變量指數加權移動平均管制圖 推導過程 :

多變量統計製程管制圖 ~ 多變量指數加權移動平均管制圖 Given an out-of control signal is chosen to achieve a specified in-control ARL and is the covariance matrix of is the asymptotically stationary

多變量統計製程管制圖 ~ 多變量指數加權移動平均管制圖 當t當t where is analogous to the variance of the univariate EWMA

多變量統計製程管制圖 ~ 多變量指數加權移動平均管制圖 noncentrality paramater Usually called the noncentrality paramater ( 非中心參數 ) ( 非中心參數 ) The value is in-control state (because the control chart can be constructed using “standardized” data) Note that for a given shift size , ARLs generally tend to increase as increases. Since the MEWMA with =1 is equivalent to the control chart, the MEWMA is more sensitive to smaller shifts.

多變量統計製程管制圖 ~ 多變量指數加權移動平均管制圖

10-5 REGRESSION ADJUSTMENT Hotelling T 2 control chart - It is not necessarily an optimal control- charting procedure for detecting mean shifts - Hotelling T 2 is not optimal for more structured shifts in the mean, such as shifts in only a few of the process variables.

NICE FEATURE FOR REGRESSION ADJUSTMENT If the proper set of variables is included in the regression model, the residuals from the model will typically be unrelated, even though the original variable of interest y 1 exhibited correlation.

There is no evidence of autocorrelation in the residuals

Application for regression adjustment The regression adjustment has many possible applications in chemical and process plants where there are often cascade processes with several inputs but only a few out puts, and where many of the variables are highly autocorrelated. Example: 半導體製程中,其中一個重要的步驟為化學氣相沉 積 (Chemical Vapor Deposition) ,其目的在於晶圓 上沉積一層薄膜,應用了熱能、電能及光學反應, 可知薄膜厚度受到許多變因的影響,彼此之間也有 關聯性,如電壓改變會影響溫度,所以這些製程及 品質變數不完全獨立。

10-6 Control chart for monitoring variability

Analogy with photographic film on which a hidden or latent image is stored as a result of light interacting with the surface of the film.

Original Variables: X 1,X 2 Principal Components: Z 1,Z 2 Original Variables: X 1,X 2, X 3 Principal Components: Z 1,Z 2 in a plane Find a new set of orthogonal directions that defines the maximum variability in the original data. Use fewer than p principal components to obtain a satisfactory description.

Matrix of scatter-plots: The first two variables are highly correlated. The other two variables exhibit moderate correlation.

Cumulative percent of first two principal components is over 83% of the variability in the original for variables.

The ellipse is an approximate control limit. All 20 scores are inside the ellipse.

Components Trajectory Chart If scores plot outside the ellipse, then the process is out of control. Control charts and trajectory plots based on PCA will be the most effective in detecting shifts in the directions defined by the principal components.

Refer to the textbook, pp. 520 – 521 for more information

On-line monitoring when the process yields a linear profile Lan Kang; Susan L Albin Journal of Quality Technology; Oct 2000; 32, 4; ProQuest Science Journals, pg. 418

On the monitoring of linear profiles Keunpyo Kim; Mahmoud A Mahmoud; William H Woodall Journal of Quality Technology; Jul 2003; 35, 3; ProQuest Science Journals, pg. 317

聽累了ㄇ ………… 忍很久了吧 忍很久了吧 恭喜您終於聽完了 恭喜您終於聽完了 好好回家補眠吧 !! 好好回家補眠吧 !!