SPC統(tǒng)計(jì)技術(shù)流程控制--stevexinli.ppt

1、,統(tǒng)計(jì)技術(shù)流程控制Statistical Process Control for Operators,惠亞集團(tuán)SPC課程 A Workshop from Our Worldwide SPC Curriculum,SPC for Operators - Version 2.1 Mar. 10, 2001,2,培訓(xùn)目的Training Goal,目的-助你獲得SPC的知識(shí)及技巧 Our Goal - To help you gain relevant SPC knowledge 多久收集一次? When do we collect it; how often? 流程中哪處收集數(shù)據(jù)? Where i

2、n the process do we get the data? 誰(shuí)收集數(shù)據(jù)? Who collects the data?,SPC for Operators - Version 2.1 Mar. 10, 2001,18,什么數(shù)據(jù)需要收集?What data do we collect?,SPC for Operators - Version 2.1 Mar. 10, 2001,19,為什么需要收集數(shù)據(jù)?Why should we collect the data?,監(jiān)控變異 To monitor variation. 盡早指出流程問(wèn)題 To identify process proble

3、ms as early as possible.,為什么? Why?,SPC for Operators - Version 2.1 Mar. 10, 2001,20,如何收集數(shù)據(jù)?How do we collect the data?,如何? How?,收集數(shù)據(jù) Collect data,我們需要一個(gè)測(cè)量系統(tǒng) We need a measurement system.,方法-技巧,工具,經(jīng)培訓(xùn)的員工 Method - technique, tool, trained people 記錄-數(shù)據(jù)收集表格,表達(dá)數(shù)據(jù)的格式(即控制圖) Record - data collection sheet,

4、format to present data (i.e., control chart) 計(jì)劃-負(fù)責(zé)人,收集數(shù)據(jù)地點(diǎn),頻率,時(shí)間 Plan - who, where, how often, when,SPC for Operators - Version 2.1 Mar. 10, 2001,21,何時(shí)及何處收集數(shù)據(jù)?When & where do we get the data?,那一個(gè)比較好? Which is better?,流程中作測(cè)量 Measure on-line,最后檢查 Inspect at end,SPC for Operators - Version 2.1 Mar. 10,

5、 2001,22,何時(shí)收集數(shù)據(jù)?When do we collect the data?,流程中作測(cè)量 Measure On-line, During the Process SPC SPC 預(yù)防性 Prevention focus 避免浪費(fèi) Avoids waste 提供質(zhì)量一致的產(chǎn)品 Delivers consistent products 減少變量 Reduces variation 主動(dòng)性 Proactive,于流程終點(diǎn)作檢查 Inspect Product at the End of the Process 檢查 Inspection 偵察性 Detection focus 容忍浪費(fèi)

6、 Tolerates waste 返工 Reworks products 容忍變異 Tolerates variation 被動(dòng)性 Reactive,SPC for Operators - Version 2.1 Mar. 10, 2001,23,數(shù)據(jù)的種類Types of Data,計(jì)量的 Variables 可用連續(xù)性的單位 Measure on a continuous scale 可以是流程或量度的特性 For a process or product 例子: Examples: 線寬 line width 板料厚度 lamination thickness,計(jì)數(shù)的 Attribute

7、s 數(shù)量,等級(jí),類別 Count, grade, categorize 不能量度的 Cant measure it 一般是產(chǎn)品特性 Usually for a product 例子: Examples: 壞品百分比 % defective parts 甩貨百分比 % orders late,SPC for Operators - Version 2.1 Mar. 10, 2001,24,控制圖種類Types of Control Charts,計(jì)量控制圖 Variables Charts 用于連續(xù)單位的量度 For data we get by measuring on a continuou

8、s scale. 例子 Examples 溫度 temperature 線寬 circuit width 厚度 pin height,分別監(jiān)察兩組統(tǒng)計(jì)數(shù)值的三對(duì)控制圖 Three pairs of charts that monitor two statistics.,平均及范圍(X-R),Average & R (average and range),平均及標(biāo)準(zhǔn)差(X-S),Average & S (average and standard deviation),個(gè)別值移動(dòng)范圍(X-MR),X & MR (individuals and moving range),SPC for Opera

9、tors - Version 2.1 Mar. 10, 2001,25,計(jì)數(shù)控制圖 Attribute Charts 用于數(shù)量等級(jí)或分類的數(shù)據(jù) For data we get by counting, grading or categorizing things. 例子 Examples 甩貨百分比 # of late deliveries 壞板百分比 # of defective circuit boards,計(jì)數(shù)控制圖,Attribute Charts,NP,P,U,C,控制圖種類Types of Control Charts,SPC for Operators - Version 2.1

10、 Mar. 10, 2001,26,規(guī)格Specifications,客戶,供應(yīng)商或內(nèi)部工程要求 Customer, supplier or internal engineering requirements 包括: Include: 目標(biāo)或中間值(最理想的) target or nominal value (the ideal!) 在目標(biāo)值兩旁可接受的變異 allowable variation on either side of the target value 規(guī)格上限=目標(biāo)值+可接愛(ài)的變異 USL = target + allowable variation 規(guī)格下限=目標(biāo)值-可接受的

11、變異 LSL = target - allowable variation,SPC for Operators - Version 2.1 Mar. 10, 2001,27,例子-酸濃度Example - Acid Concentration,規(guī)格是130克/升+/-30克升 Specification is 130 grams/liter +/- 30 grams/liter 規(guī)格上限=目標(biāo)值+可接受的變量 USL = target + allowable variation 130 + 30 = 160 規(guī)格下限=目標(biāo)值-可接受的變異 LSL = target - allowable va

12、riation 130 - 30 = 100 可接受變異的總值=規(guī)格上限=規(guī)格下限 Total allowable variation = USL - LSL 160 - 100 = 60,SPC for Operators - Version 2.1 Mar. 10, 2001,28,控制限-控制上,下限Control Limits - UCL, LCL,若流程受到控制: If your process is in control: 99.7%的數(shù)據(jù)將會(huì)在控制上限及下限之間. 99.7 % of the data points will be between the UCL and LCL.

13、 數(shù)據(jù)沒(méi)有特定的圖樣=隨機(jī)的. Data points will show no pattern = random. 大部分?jǐn)?shù)據(jù)會(huì)趨向平均值. Most data points will be close to the average. 數(shù)據(jù)應(yīng)有一半在平均值以上及一半在以下. Data points be 50/50 above and below average.,SPC for Operators - Version 2.1 Mar. 10, 2001,29,控制限-好處Control Limits - Benefits,減少兩個(gè)可能的錯(cuò)誤 Minimizes two possible m

14、istakes: 忽略問(wèn)題 Ignoring a problem. 采取不必要的糾正行動(dòng) Taking unnecessary corrective action. 曲線下的出現(xiàn)可能性已知 Probabilities under the curve are known. 有利于預(yù)測(cè)和計(jì)劃 Good for prediction and planning. 有利于偵察不正常的數(shù)據(jù) Good for spotting weird patterns of points.,SPC for Operators - Version 2.1 Mar. 10, 2001,30,控制圖及標(biāo)準(zhǔn)差Control C

15、harts & Standard Deviation,使用標(biāo)準(zhǔn)差決定控制上限及下限,如鐘狀曲線. To determine UCL & LCL, use standard deviation (s) - just like the bell curve!,利用變異的數(shù)值(標(biāo)準(zhǔn)差)去建立變異的一般范圍 (控制上限及下限之間的范圍)是否合理? Doesnt it make sense to use a measure of variation (standard deviation) to establish a normal range of variation (area between UC

16、L and LCL) for a process?,SPC for Operators - Version 2.1 Mar. 10, 2001,31,設(shè)定控制限Setting Control Limits,將鐘狀曲線轉(zhuǎn)90度 Flip the bell curve on its side.,SPC for Operators - Version 2.1 Mar. 10, 2001,32,設(shè)定控制限Setting Control Limits,這是什么意思? What does it all mean?,SPC for Operators - Version 2.1 Mar. 10, 2001,

17、33,生產(chǎn)中使用SPCUsing SPC During Production,SPC for Operators - Version 2.1 Mar. 10, 2001,34,使用SPC-例子Using SPC - An Example,SPC for Operators - Version 2.1 Mar. 10, 2001,35,基本控制圖格式Basic Control Chart Format,一般原因?qū)е碌淖儺?Common cause variation,特別原因?qū)е碌淖儺?Special cause variation,8,8.1,8.2,8.3,8.4,8.5,8.6,8.7,8

18、.8,8.9,9,1,2,3,4,5,6,7,8,9,10,11,12,13,14,15,mils,UCL,LCL,Y軸-量度 y-axis - scale for measurements,X軸-時(shí)間 X-axis - time sequence,SPC for Operators - Version 2.1 Mar. 10, 2001,36,平均值(典形數(shù)值) Average (the typical value) 平均年齡=(27+ 27+33+46+57) 5 = 38 歲 Average age = (27+27+33+46+57) 5 = 38 years,Average age

19、= 38,統(tǒng)計(jì)技術(shù)-中間值傾向Statistics - Central Tendency,SPC for Operators - Version 2.1 Mar. 10, 2001,37,統(tǒng)計(jì)技術(shù)-分布Statistics - Spread,范圍(R)-最大數(shù)字減最小數(shù)字 Range (R) - highest number minus lowest number R= (57 - 27) = 30 歲 R = (57 - 27) = 30 years,幅度=30歲 Range = 30 years,27,27,33,46,57,SPC for Operators - Version 2.1

20、Mar. 10, 2001,38,統(tǒng)計(jì)技術(shù)-分布Statistics - Spread,SPC for Operators - Version 2.1 Mar. 10, 2001,39,例子-計(jì)數(shù)控制圖Example - Attribute Chart,P Chart - VAAS Fastech Final Inspection,SPC for Operators - Version 2.1 Mar. 10, 2001,40,流程穩(wěn)定性Process Stability,穩(wěn)定的流程 Stable Process 一致的 Consistent 可預(yù)測(cè)的 Predictable 特別原因 No

21、special causes 受控的 In control 穩(wěn)定性可用控制圖量度 Stability is measured using control charts.,流程可以是受控,但不符合規(guī)格的. A process could be in control, but not meet specifications.,SPC for Operators - Version 2.1 Mar. 10, 2001,41,什么是流程?What is a process?,完成一件工作的一連串步驟 A series of steps we follow to accomplish a task.,S

22、PC for Operators - Version 2.1 Mar. 10, 2001,42,流程的內(nèi)容The parts of a Process,SPC for Operators - Version 2.1 Mar. 10, 2001,43,穩(wěn)定,能力及質(zhì)量Stability, Capability & Quality,為能做到優(yōu)越的表現(xiàn),工作流程必須: To achieve performance excellence, work processes must be:,穩(wěn)定,Stable,一致的,可預(yù)計(jì)的,consistent, predictable,有能力,Capable,持續(xù)符

23、合規(guī)格,continually meets specifications,持續(xù)改善,Continually improved,SPC for Operators - Version 2.1 Mar. 10, 2001,44,流程穩(wěn)定及能力Process Stability & Capability,有能力的流程 Capable Process 流程有能力持續(xù)的滿足規(guī)格要求 Process has the ability to continually meet specifications.,流程必須處于受控狀態(tài)(統(tǒng)計(jì)技術(shù)角度),才可被認(rèn)為有能力 Process must be in stati

24、stical control before it can be considered capable.,SPC for Operators - Version 2.1 Mar. 10, 2001,45,穩(wěn)定Stability,SPC for Operators - Version 2.1 Mar. 10, 2001,46,變異Variation,個(gè)別成品之間不可避免的分別 The inevitable differences among individual outputs of a process.,一般原因?qū)е碌淖儺?Common Cause Variation,正常,常見(jiàn),可預(yù)計(jì)的-自然

25、的,normal, usual, expected things - natural,特別原因?qū)е碌淖儺?Special Cause Variation,偶發(fā)性,不可預(yù)計(jì)的-特定的,sporadic, unexpected things - assignable,SPC for Operators - Version 2.1 Mar. 10, 2001,47,變異何時(shí)何地都會(huì)出現(xiàn)Variation is Everywhere!,SPC for Operators - Version 2.1 Mar. 10, 2001,48,變異結(jié)果Consequences of Variation,SPC f

26、or Operators - Version 2.1 Mar. 10, 2001,49,Cp及Cpk值Value of Cp and Cpk,根據(jù)規(guī)格評(píng)估流程表現(xiàn).,Evaluate process performance against specs.,決定流程是否能滿足新規(guī)格要求.,Determine if process can meet new specs.,根據(jù)Cpk及Cp值的距離制訂改善目標(biāo).,Set improvement goal based on distance between Cpk and Cp.,評(píng)估供應(yīng)商.,Evaluate suppliers.,用于銷售過(guò)程的一部分

27、.,Use as part of sales process.,SPC for Operators - Version 2.1 Mar. 10, 2001,50,Cpk -量度流程設(shè)定Cpk - Measures Process Setting,比較流程中間值設(shè)定及目標(biāo). Compares setting of process center (average) to target. 基本定律: Cpk要求最少1.33 General rule - aim for Cpk of 1.33 minimum,SPC for Operators - Version 2.1 Mar. 10, 2001,

28、51,Cpk -例子Cpk Examples,SPC for Operators - Version 2.1 Mar. 10, 2001,52,Cp-量度流程分布Cp - Measures Process Spread,比較可接受變異(A)及實(shí)際變異(B) Compare allowable variation (B) to actual variation (A). B = USL - LSL A=6標(biāo)準(zhǔn)差 A = 6 standard deviations Cp = B/A,SPC for Operators - Version 2.1 Mar. 10, 2001,53,Cp 必須大于1C

29、p Should Be Greater Than 1,SPC for Operators - Version 2.1 Mar. 10, 2001,54,Cp -總結(jié)Cp - Conclusions,SPC for Operators - Version 2.1 Mar. 10, 2001,55,流程能力不足的兩種情況Two Ways a Process Is NOT Capable,1.實(shí)際流程變異超出可接受變異的總值 Actual process variation (A - six sigma) exceeds total allowable variation (B - spec),Ta

30、rget,A,LSL,USL,SPC for Operators - Version 2.1 Mar. 10, 2001,56,流程能力不足的兩種情況Two Ways a Process Is NOT Capable,2.如流程表現(xiàn)不足處于規(guī)格的中間位置,并其中一邊的 控制限超越規(guī)格限. If process is not centered properly and one of the control limits exceeds the spec limit.,SPC for Operators - Version 2.1 Mar. 10, 2001,57,若流程能力不足如何辦?What

31、if a process isnt capable?,流程需要改善,The process needs improvement!,首先,采取改善行動(dòng),減少變異,然后,按目標(biāo)值將流程表現(xiàn)調(diào)至中間值.,First, take action to reduce the variation. Then, center the process over the target.,SPC for Operators - Version 2.1 Mar. 10, 2001,58,你需要做什么?What should you do?,SPC for Operators - Version 2.1 Mar. 10

32、, 2001,59,你需要做什么?What should you do?,SPC for Operators - Version 2.1 Mar. 10, 2001,60,你需要做什么?What should you do?,SPC for Operators - Version 2.1 Mar. 10, 2001,61,失控圖樣-第一定律Out-of Control Patterns - Rule 1,任何點(diǎn)超越控制限 Any point beyond the control limits,控制上限以外 Outside the Upper Control Limit,8.0,8.1,8.2,

33、8.3,8.4,8.5,8.6,8.7,1,2,3,4,5,6,7,8,9,10,11,12,13,14,15,LCL,Average,UCL,控制下限以外 Outside the Lower Control Limit,8.0,8.1,8.2,8.3,8.4,8.5,8.6,8.7,1,2,3,4,5,6,7,8,9,10,11,12,13,14,15,LCL,Average,UCL,SPC for Operators - Version 2.1 Mar. 10, 2001,62,失控圖樣-第二定律Out-of Control Patterns - Rule 2,連續(xù)七點(diǎn)向上或向下-走勢(shì) U

34、pward or downward run of 7 consecutive points -trend,連續(xù)七點(diǎn)向下 Downward run of seven consecutive points,連續(xù)七點(diǎn)向上 Upward run of seven consecutive points,8.0,8.1,8.2,8.3,8.4,8.5,8.6,8.7,1,2,3,4,5,6,7,8,9,10,11,12,13,14,15,LCL,Average,UCL,8.0,8.1,8.2,8.3,8.4,8.5,8.6,8.7,1,2,3,4,5,6,7,8,9,10,11,12,13,14,15,L

35、CL,Average,UCL,SPC for Operators - Version 2.1 Mar. 10, 2001,63,失控圖樣-第三定律Out-of Control Patterns - Rule 3,連續(xù)七點(diǎn)在平均值以上或以下-偏移 A run of seven consecutive points above the average or below the average - shift,連續(xù)七點(diǎn)在平均值以上 Run of seven consecutive points above the average,8.0,8.1,8.2,8.3,8.4,8.5,8.6,8.7,1,2

36、,3,4,5,6,7,8,9,10,11,12,13,14,15,LCL,UCL,連續(xù)七點(diǎn)在平均值以下 Run of seven consecutive points below the average,8.0,8.1,8.2,8.3,8.4,8.5,8.6,8.7,1,2,3,4,5,6,7,8,9,10,11,12,13,14,15,LCL,Average,UCL,Average,SPC for Operators - Version 2.1 Mar. 10, 2001,64,失控圖樣其它定律Out-of Control Patterns - Rule others,One point more than 3 sigmas from center line Nine points in a row on same side of center line Six points in a row , increasing or all decreasing Fourteen points in a row , alternating up and down Two out of three points more than 2 sigmas fro