【正文】 facturing to Dr Walter A. 應(yīng)用應(yīng)歸功于修哈特博士FSGS M00 Rev B 8Two Basic Causes of Variability? Chance Causes of Variation變異的偶因216。Review of Attributes Control Charts屬性控制圖回顧? Designing the Xchart均值控制圖的設(shè)計(jì)216。Misconceptions of SPC/SPC 誤解216。Process Control Model/制程控制模型216。Sampling Risks (? and b)抽樣風(fēng)險(xiǎn)216。 Due to the cumulative effect of many small unavoidable sources of 變異源的累積216。inherent variation內(nèi)在變異232。 disturbs a process so that what it produces seems ，制程受到明顯的影響，顯得不自然。 excessive tool wear過(guò)多使用工具232。 in other words, to keep the process in statistical 說(shuō)，使得制程處于統(tǒng)計(jì)控制FSGS M00 Rev B 11Objectives of SPC Charts? The following are secondary objectives or direct benefits of the primary objective:以下是采用控制圖的好處216。 Spec. limits are imposed by external demands, thus do not help identify assignable causes規(guī)格極限被應(yīng)用于外部，不能識(shí)別異因 .? Points plotted within the control limits do not indicate directly that parts are meeting customer specs. or Cpk requirements點(diǎn)在控制線內(nèi)不代表產(chǎn)品滿足客戶規(guī)格或 Cpk要求 . Spec. Limits has No Business in Control Charts!FSGS M00 Rev B 16Shewhart Control Charts Overview變量 屬性缺陷有缺陷的FSGS M00 Rev B 17Review of XR Charts均值極差圖回顧均值極差圖回顧FSGS M00 Rev B 18Central Limit Theorem and Normal Distribution? Shewhart variables control charts for subgroups work because of two important principles修哈特控制圖應(yīng)用于子組，因?yàn)橐韵?2個(gè)重要原理 :216。 It maximizes the chance of variability due to an assignable cause to appear between samples可以最大化異因在不同樣本內(nèi)出現(xiàn)的機(jī)會(huì) .216。3? limits, ? = % 222。FSGS M00 Rev B 58Sampling Interval Sample Size Procedure抽樣間隔和樣本大小選擇程序9. Based on engineering judgment, define D = max. allowable process drift in m that must be detected根據(jù)工程判斷，決定 允許制程在 m內(nèi)漂移的最大值 D10. Establish process ?x from historical data, or if not available, collect some parts to get a rough estimate根據(jù)歷史數(shù)據(jù)建立制程，如果不可能，建立一個(gè)大概11. Alternatively, D/?x ratio (k) can be derived from process capability considerations using following equation:比例 k可以通過(guò)下列公式而得到k = D/?x = 3[ Ppkcurrent – Ppkmin ]12. With that, all parameters for puting n are found這樣，所有計(jì)算 n的參數(shù)已經(jīng)建立， n可以確定了 :n = f [? , b , D , ? ]Designing an XR Chart 252。FSGS M00 Rev B 59Designing an XR Chart This chart is for ? = % only.k,(1 b)FSGS M00 Rev B 60Sampling Interval Sample Size Procedure抽樣間隔和樣本大小選擇程序13. Instead of using above graph, following equation can be used to calculate n下列等式也可以計(jì)算 n值 :14. Alternatively, above formula isalso available in Minitab. Use:Stat Power and Sample Size 1sample Z這個(gè)公式也可以在 Minitab 中找到Designing an XR Chart FSGS M00 Rev B 61Example 1: Designing an XR ChartYour management has the following concerns with regards to implementing control charts at the shopfloor在推行控制圖時(shí)，管理層很關(guān)心下列問(wèn)題 :1. False Alarm:錯(cuò)誤報(bào)警Not more than 1 false alarm for every 500 hrs of production每500小時(shí)的生產(chǎn)不超過(guò) 1個(gè)錯(cuò)誤報(bào)警 .2. Exposure:暴露Not more than 3 hrs exposure if process has shifted如果制程偏移，不超過(guò) 3小時(shí)暴露 .3. Allowable Shift允許偏移 :From historical data, Ppk ? .Min customer requirement, Ppk ? Design an appropriate XbarR chart to meet the above FSGS M00 Rev B 62After some time, the factory decides to switch to 14 hrs of production per ，工廠決定每天 14小時(shí)生產(chǎn)Tolerable false alarm:極限錯(cuò)誤報(bào)警Average 1 false alarm per month (30 days)每月平均 1個(gè)錯(cuò)誤報(bào)警What changes do you need to make to the charts you have designed in Example 1?照此條件，對(duì)所設(shè)計(jì)的例子 1的控制圖有何改變？Example 2: Designing an XR ChartFSGS M00 Rev B 63Where to Use SPC Charts? When a mistakeproofing device is not feasible當(dāng)過(guò)失防治設(shè)施不可行時(shí)? Identify processes with high RPNs from FMEA從 FMEA中識(shí)別高風(fēng)險(xiǎn)度的制程216。 If the process mean can be easily adjusted by manipulating some process inputs, it may be helpful to shift the process mean to the desired value.216。 If we discard the point, no analytical justification other than assuming that the process has been out of control but will not now. Trial control limits after revision may cause more false alarms.216。 252。 Exp. rate 7. From puted ARL1, establish the required b, or (1 – b), using ARL1 = 1/(1 – b)計(jì)算 b, or (1 – b), 8. At this juncture, sampling interval is already established some parameters already established for sample size calculation到現(xiàn)在已經(jīng)有了計(jì)算樣本大小的一些參數(shù) : n = f [? , b , D , ? ]Designing an XR Chart 252。 p ? 1/2 and n 10 implying np 5216。 Normal Distribution正態(tài)分布? Shewhart found that when the averages of subgroups from a constantcause system are plotted in the form of a histogram, the normal distribution appears修哈特發(fā)現(xiàn)當(dāng)把常量系統(tǒng)的子組平均值作成一個(gè)直方圖時(shí)，就出現(xiàn)了常態(tài)分布 . FSGS M00 Rev B 19Construction of XR Charts? The XR chart is the most versatile of control charts, and is used in most 用控制圖? Charting of averages and charting of ranges are used to check if a constantcause system exists均值圖和極差圖用于檢驗(yàn)常量系統(tǒng)是否存在 . Xchart measures variability between samples均值圖測(cè)量 2個(gè)樣本間的差異Rchart measures variability within samples極差圖測(cè)量樣本內(nèi)的差異FSGS M00 Rev B 20The Center Line and Control Limits of a Xchart:The Center Line and Control Limits of a Rchart:Construction of XR ChartsFSGS M00 Rev B 21For sample size n 10, R loses its efficiency in estimating process sigma and Rchart may not be appropriate.樣本數(shù)大于 10時(shí)，極差不能用于估算制程標(biāo)準(zhǔn)差，極差圖不適用Construction of XR ChartsShewhart Constants修哈特常數(shù)FSGS M00 Rev B 22How do we begin to set up an XR chart from scratch?如何建立均值極差控制圖？Implementing the Control Chart控制圖的推行控制圖的推行FSGS M00 Rev B 23Implementing the Control Chart1) Preparation of Sampling抽樣準(zhǔn)備2) Data Collection數(shù)據(jù)收集3) Construct the Control Chart組建控制圖4) Analysis Interpretation分析5) Use the Control