统计过程控制[898]---管理篇课件

单击此处编辑母版标题样式,单击此处编辑母版文本样式,第二级,第三级,第四级,第五级,*,#,单击此处编辑母版标题样式,单击此处编辑母版文本样式,第二级,第三级,第四级,第五级,单击此处编辑母版标题样式,单击此处编辑母版文本样式,第二级,第三级,第四级,第五级,单击此处编辑母版标题样式,单击此处编辑母版文本样式,第二级,第三级,第四级,第五级,单击此处编辑母版标题样式,单击此处编辑母版文本样式,第二级,第三级,第四级,第五级,单击此处编辑母版标题样式,单击此处编辑母版文本样式,第二级,第三级,第四级,第五级,本节内容结束,单击此处编辑母版标题样式,单击此处编辑母版文本样式,第二级,第三级,第四级,第五级,单击此处编辑母版标题样式,单击此处编辑母版文本样式,第二级,第三级,第四级,第五级,*,单击此处编辑母版标题样式,单击此处编辑母版文本样式,第二级,第三级,第四级,第五级,*,单击此处编辑母版标题样式,单击此处编辑母版文本样式,第二级,第三级,第四级,第五级,*,单击此处编辑母版标题样式,单击此处编辑母版文本样式,第二级,第三级,第四级,第五级,*,单击此处编辑母版标题样式,单击此处编辑母版文本样式,第二级,第三级,第四级,第五级,*,单击此处编辑母版标题样式,单击此处编辑母版文本样式,第二级,第三级,第四级,第五级,*,1,SPC,统计过程控制,SPC Variable Data,连续型数据的,SPC,SPC Attribute Data,离散型数据的,SPC,1SPC,Statistic Process Control,2,SPC,统计过程控制,Statistic Process Control2SPC,Introduction to SPC,介绍,Methodology which uses the basic graphical and statistical tools to analyze, control, and reduce variability within a process.,SPC,是一种使用基本图示与统计工具来分析、控制和减少过程变差的方法,.,SPC can be applied to nearly any area.,SPC,几乎可应用于任何领域。,SPC stands for Statistical,Process,Control. Unfortunately, most companies apply it to finished goods (Ys) rather than process characteristics (Xs). SPC,是统计,过程,控制的缩写。但目前大部分公司应用在做好的产品,(Y,输出结果,),而非过程特性上,(X,输入因子,),。,Until the process inputs which generate the outputs become the focus of our effort, the full power of SPC methods to improve quality, increase productivity, and reduce cost cannot be realized.,只有当产生输出的过程输入成为我们关注的焦点,SPC,在质量改善、生产率提升和成本降低方面的强大作用才能被完全实现。,3,Introduction to SPC 介绍Methodol,What is SPC?,什么是,SPC,SPC is one kind of Control charts, which are the means through which process and product parameters are tracked statistically over time.,SPC,是一种控制图， 它是按时间来追踪过程和产品参数的手段。,We use SPC to monitor and/or improve the process.,我们用,SPC,来监控与提高过程水平。,While control charts can indicate special causes through Out-of-Control signals, they,cannot,tell you why the process is out of control.,控制图可以通过失控信号显示出有特殊原因存在，但它并,不能,告诉我们为什么过程会失控。,4,What is SPC? 什么是SPCSPC is,What is SPC?,All processes have natural variability (due to common causes) and unnatural variability (due to assignable causes).,任何过程都包含自然变异（普通原因引起）与非自然变异（特殊原因引起）,5,Time,Reject,Rate,Special Variation,Historical Level (,0,),Optimum Level (,1,),Natural Variation under,0,Natural Variation under,1,(3,0,),(3,1,),What is SPC? All processes hav,Common Vs Special Causes,普通和特殊原因,The Control Limits separate common and special cause variation,控制限用于区分引起变差的普通原因或特殊原因,“Common Cause Variation” is random, expected variation,普通原因引起的波动是随机的，可以预估的。,“Special Cause Variation” is assignable or not inherent to the process,特殊原因引起的波动是特定的或不是过程所固有的。,Referred to as fleeting events,如：短时事件,Out of Control points,超出控制的点,Control limits should NOT be confused with specification limits,注意不要将控制限与规格限相混淆,6,Common Vs Special Causes 普通和特殊,Common Vs Special Causes,普通和特殊原因,7,If there is only common causes, the output will be stable and divinable:,如果仅存在变异的普通原因，则过程输出是稳定并可预测的。,Common Causes:,普通原因,Predictable, Stable over time,随时间可预测、稳定,In Statistical control,统计受控,Due to: Inherent or natural,是由于内在或自然的原因,Natural or typical process variation,自然或典型的过程波动,Common Vs Special Causes 普通和特殊,Common Vs Special Causes,普通原因和特殊原因,8,If there is only special causes, the output will likes:,如果仅存在变异的特殊原因，则过程输出如下图,：,Assignable Causes:,特殊原因,Unpredictable (hour-to-hour, day-to-day, week-to-week),不可预测（每小时之间，每天之间，每星期之间都不一样）,Out of Control,不受控,Due to: improperly adjusted m.c, operator, defective raw material,由于不正确的调整、人员、原材料等,Target: Detect and eliminate assignable causes,目的：识别并消除特殊原因,Common Vs Special Causes 普通原因和,Common Vs Special Causes,普通原因和特殊原因,9,Individual Value,单值,Region of Special Cause Variation,特殊波动区,Region of Special Cause Variation,特殊波动区,TIME,时间,UCL,LCL,Region of Common Cause Variation,普通波动区,They can be separated by the control chart:,控制图可以区分这两种原因：,Common Vs Special Causes 普通原因和,The Theory of Control Charts,原理,The theory behind a control chart follows the principles of simple confidence intervals,控制图是基于简单的置信区间原理,10,When the true process is running at mean,m, the probability that a sample mean falls outside the control limits equals,a,如果一个过程数据服从均值为,的分布，那么样本平均值的分布落在控制限外的概率为,m,UCL,LCL,a,/2,a,/2,Distribution of Sample Means,The Theory of Control Charts 原,11,Normal Distribution,Properties,正,态分布的特征,Properties,特征,:,The mode occurs at x=,数据的模数是均值,The curve is symmetric about,分布曲线以均值对称,Point of inflexion at x=,在,处是曲线的弯曲点,Probability distribution,分布概率,:,1 68.27%,2 95.45%,3 99.73%,11Normal Distribution Properti,Establishing Process Control Limits,建立工艺控制限,控制限是,Are statistical limits set +/- 3 standard deviations from the mean,设立在平均值的,+/- 3,个标准差处,受控是,A statistical term for process variation,过程波动的统计性术语,Within three standard deviations of the mean,在均值的,3,个标准差内,That is random without cause,完全随机分布,That does not show run patterns,没有特别的运行规律,That does not show trend patterns,没有趋势规律,No assignable cause variation,没有特殊原因引起的波动,Control limits are not related to specification limits,控制限与规格限没有关系,12,Control Limits are,not,specification limits,控制限不是规格限！,Establishing Process Control L,Establishing Process Control Limits,建立工艺控制限,13,不合格,不合格,控制 下限,控制 上限,规格,下限,规格,上限,客户规格限,通常超出控制限之外,控制限,通常由过程控制人员根据,历史数据或实验数据计算得出,3,Sigma,+,产品规格容差,T,规格限,通常由设计给定,或由客户规定,Establishing Process Control L,The Time Base,以时间为基础,One advantage of a control chart involves its ability to track the process over time.,控制图的优点之一是其具有随时间追踪过程的能力。,14,This time-based dependence means that certain trends or patterns may indicate that special causes are occurring over time.,如果有一定的趋势或规律模式出现，就表明随时间推移有特殊原因发生。,Time,The Time Base 以时间为基础One advant,Components of a Control Chart,控制图组成,15,10,9,8,7,6,5,4,3,2,1,0,0 5 10 15 20,Upper Control Limit,控制上限,Lower Control Limit,控制下限,Mean,均值,Nonrandom Variation Region,非随机波动区域,Observation number,观察点号码,Observation value,观察值,Random Variation Region,随机波动区域,Observation 10,第十个观察点,Components of a Control Chart,Statistics of a Control Chart,控制图统计量,16,10,9,8,7,6,5,4,3,2,1,0,0 5 10 15 20,Nonrandom Variation Region,Observation number,Observation value,Random Variation Region,LCL,- 3,s,UCL,+ 3,s,Mean,99.73% area,Statistics of a Control Chart,One Control Chart Sample,控制图示例,17,One Control Chart Sample 控制图示例,What Does “Out of Control” Mean?,失控意味着什么,A process that shows a non-random pattern is considered to be,statistically,out of control,当一个过程呈现非随机形态时，我们称其为,统计,失控,Be careful with the terminology, since the process may still be producing good parts. The non-random pattern only means that there is strong statistical evidence that the process has changed.,在使用这个术语时须注意，因为这时过程生,产的产品可能仍是合格的。这个非随机形态,只是表明有很强的统计证据显示该过程已发,生变化。,18,What Does “Out of Control” Me,Tests for Special Causes,特殊原因判别法则,19,Any point outside a control limit,任一点超出控制限,Possible Causes,可能原因,:,new workers, methods, raw materials or machine,新的工人，方法，原材料或机器,change in inspection methods or standards,检验方法或标准的改变,change in skill and/or motivation of operators,操作者技能或积极性方面的改变,UCL,LCL,A,A,B,C,C,B,x,x,Test 1. One Point Beyond Zone A,Tests for Special Causes 特殊原因判,Tests for Special Causes,特殊原因判别法则,20,2. 9 points in a row on one side of the center line,9,点在中心线的同一侧,x,Test 2. Nine Points in a Row,on One Side of the Center Line,UCL,LCL,A,A,B,C,C,B,Possible Causes,可能原因,:,new workers, methods, raw materials or machine,新的工人，方法，原材料或机器,change in inspection methods or standards,检验方法或标准的改变,change in skill and/or motivation of operators,操作者技能或积极性方面的改变,Mean probably is changed,平均值已发生变化,Shift or Run,改变,Tests for Special Causes 特殊原因判,Tests for Special Causes,特殊原因判别法则,21,3. 6 points in a row steadily increasing or decreasing,连续,6,点稳步上升或下降,UCL,LCL,A,A,B,C,C,B,x,x,Upward/Downward Trend,Excursion of the mean,向上或向下的趋势，均值有偏离,Possible Causes,可能原因,:,new workers, methods, raw materials or machine,新的工人，方法，原材料或机器,change in inspection methods or standards,检验方法或标准的改变,change in skill and/or motivation of operators,操作者技能或积极性方面的改变,component wear,部件磨损,Trend,趋势,Tests for Special Causes 特殊原因判,Tests for Special Causes,特殊原因判别法则,22,Mixture 2 Stream Symptom,有两簇数据混在一起的征兆,Possible Causes,可能原因,:,over-control by operators,操作者控制过严,controlling 2 processes on the same chart,用同一张控制图控制两个过程,2 different shifts, 2 different raw materials, 2 different operators,两个不同的班次、原材料或操作者等,4. 14 points in a row alternating up and down,连续,14,点交替升降,UCL,LCL,A,A,B,C,C,B,x,Mixture,混合,Tests for Special Causes 特殊原因判,Tests for Special Causes,特殊原因判别法则,23,UCL,LCL,A,A,B,C,C,B,x,x,x,5. 2 out of 3 points in a row in zone A or beyond,3,点中有,2,点在,A,区或超出（同侧）,Possible Causes,可能原因,:,new workers, methods, raw materials or machine,新的工人，方法，原材料或机器,change in inspection methods or standards,检验方法或标准的改变,change in skill and/or motivation of operators,操作者技能或积极性方面的改变,Shift or Run,改变,Tests for Special Causes 特殊原因判,Tests for Special Causes,特殊原因判别法则,24,UCL,LCL,A,A,B,C,C,B,x,x,6. 4 put of 5 points in a row in zone B and beyond,5,点中有,4,点在,B,区或超出（同侧）,Possible Causes,可能原因,:,new workers, methods, raw materials or machine,新的工人，方法，原材料或机器,change in inspection methods or standards,检验方法或标准的改变,change in skill and/or motivation of operators,操作者技能或积极性方面的改变,Shift or Run,改变,Tests for Special Causes 特殊原因判,Tests for Special Causes,特殊原因判别法则,25,UCL,LCL,A,A,B,C,C,B,x,7. 15 points in a row in zone C (Above and Below CL),15,点在,C,区（在中线上方或下方）,Possible Causes,可能原因,：,incorrect calculation of control limits (over-estimating process variability),控制线计算错误（对过程变动性估计过大）,continuous improvements paying off,持续改进，并获得成功,Stratification,层叠,Tests for Special Causes 特殊原因判,Tests for Special Causes,特殊原因判别法则,26,Possible Causes,可能原因,：,systematic environmental changes (temperature, operator fatigue, rotation of operators, fluctuations in machine settings),系统环境改变（温度、操作者疲劳或轮换、机器设定波动等）,maintenance schedules,设备维护计划安排,tool wear,刀具磨损,UCL,LCL,A,A,B,C,C,B,x,8. 8 points in a row on both sides of CL with none in zone C,连续,8,点在中心线两侧，没有一点在,C,区,Cycle or Periodicity,周期性,Tests for Special Causes 特殊原因判,法则记忆口诀,有时一点正好落在边界线上,这点不必计入.,记忆口诀,：,一,点出界,(法则1),二 /三,A,(法则5)，,四 /五,B,区,(法则6),十五,C,(法则7).,九点排同侧,(法则2)，,六点单调追,(法则3).,十四交替升与降,(法则4),八点两侧无一C,(法则8),27,CL,中心线,A,B,C,C,B,A,UCL,控制上限,LCL,控制下限,法则记忆口诀有时一点正好落在边界线上,这点不必计入.27CL,Tests for Instability,判断异常,CAUTION :,Do not apply,“,tests,”,blindly,注意：不要盲目使用,“,法则,”,Not every,“,test,”,is relevant for all charts,不是每个法则可用于所有控制图,Excessive number of,“,tests,”, Increased -error,过多地使用,“,法则,”,会增加生产者风险,Nature of application,应用特点,28,Tests for Instability 判断异常CAUT,Reaction,对策,It is very important to investigate every signal of special-cause variation,调查每个特殊原因的波动信号是十分重要的。,Without attention to the charts,those responsible will begin ignoring the signals.,如果不去留意控制图，负责人将从此忽略这些信号，从而也就失去许多改善的机会。,29,Reaction 对策It is very importan,Reaction,对策,Necessary to have a planned approach,一些可采取的方法：,Verify that data is correct and has been entered correctly,验证数据是正确的并被正确地输入,Determine if a simple explanation of special cause is available (e.g. parts were dropped),对特殊原因是否有简单的解释（如零件跌落）,Verify process change by taking a bigger sample (20). Use proper sample size to detect critical difference.,选取更大的样本量（,20,）来验证过程是否明显改变。使用合适的样本量来侦测明显差异。,If process change is real, assess process impact to determine next courses of action (best to have in SOP):,如果过程确实发生改变，评估对过程的影响以确定下一步的行动路线（最好写进文件中）,Line shut down,停线,Root cause investigation,根本原因调查,30,Reaction 对策 Necessary to hav,SPC : Objectives,Focuses attention on detecting and monitoring process variation over time,将注意力放在探测和监控随时间产生的过程波动上,Distinguishes special from common causes of variation,辨别特殊原因与普通原因,Serves as a tool for ongoing control of a process,将,SPC,做为一种长期的过程控制工具,Reduces variability in a process,减少过程波动,Reduces the repeat of the problem.,减少问题的重复发生,31,Goal : Monitor and control process inputs over time, eliminate the need for SPC charts.,目标：随时间监测和控制过程输入，最终使过程不再需要使用,SPC,。,SPC : ObjectivesFocuses attent,The Goal of SPC,We want to monitor and improve the Xs,监控和改善过程输入,X,32,Input,输入,Measurement,System,测量系统,Process,过程,Output,输出,2.,Identify Root Cause,确定根本原因,3.,Implement Corrective Action,采取纠正措施,4.,Verify and Monitor,验证与监控,Detect Assignable Cause,侦测特殊原因,The Goal of SPCWe want to moni,Where to Use SPC Charts,何时用,SPC,When a mistake proofing device is not feasible,当防错装置不可行时（无法采取防错保护）,Critical variables based on DOEs,基于实验设计的关键变量,Customer requirements,顾客的要求,Management commitments,管理者的承诺,Initial process outputs that requires monitoring,要监控的原始过程输出,33,Where to Use SPC Charts 何时用SPC,Kinds of Control Charts,控制图种类,34,There are variable data and attribute data because of the data characteristic, so there are two kinds of control charts.,根据数据特点，有计量型（连续）数据与计数型（离散）数据，因此控制图也分为计量值控制图与计数值控制图。,Kinds of Control Charts 控制图种类3,控制图选择流程,35,连续型数据,离散型数据,选择参数,连续型数据？,数据是,分组的么？,比率,计数,单值移动极差图,(I-MR),子组样本数,大于,9?,能计算每个,子组数据么,?,平均值,极差图,(Xbar-R),平均值标准差图,(Xbar-s),No,Yes,No,Yes,Yes,No,样本量,恒定,?,np,p,图,p,图,No,Yes,No,Yes,c or u,图,Yes,u,图,No,Yes,No,样本量,恒定,控制图选择流程35连续型数据离散型数据选择参数连续型数据？数,Sub-Grouping And Sampling Guidelines,分组和取样指导,Sample Size,样本大小,Variables data: 5 if possible,连续型数据：,5,Attributes data: 30,离散型数据：,30,Sample Frequency,取样频率,Not too many (costly),不能太多（成本浪费）,Not too few (unreliable information),不能太少（信息不可靠,）,More frequent, the better,频率越高，效果越好,On-line measurement systems offer real-time process control,在线测试系统提供了实时的过程监控,Between 20-25 subgroups,子组一般取在,20-25,之间,36,Sub-Grouping And Sampling Guid,Control Charting Steps,控制图的制作步骤,1. Select characteristic to monitor/control,选择需要监控的特性,2. Select the data collection point,确定数据采集点,3. Select type of control chart (s),选择相应控制图,4. Establish basis for rational sub-grouping,合理地分子组,5. Determine appropriate sample size and frequency,确定合适的样本容量与抽样频率,6. Determine measurement method/criteria,确定数据测量方法和标准,37,Control Charting Steps 控制图的制作步,Control Charting Steps,控制图的制作步骤,7. Determine gage capability,确定量具能力,8. Perform initial capability study to establish trial control limits,进行初始工序能力研究以确定控制限,9. Set up forms for collecting and charting data,设计表格收集数据与制图,10. Develop procedures for collection, charting,analyzing and acting on information,收集数据、制图、分析并根据所得信息采取行动,11. Train personnel,培训人员,12. Institutionalize the charting process,将过程控制制度化,Control Charting Steps 控制图的制作,Statistic Process Control,39,SPC Variable data,计量型数据统计过程控制,Statistic Process Control39SPC,Control Chart Roadmap,40,Variable,Xbar-R Chart,I-MR Chart,Xbar-s chart,NStackColumn,Xbar-R Charts in MINITAB48Data,Xbar-R Charts in MINITAB,49,Xbar-R Charts in MINITAB49,Xbar-R Charts in MINITAB,50,Xbar-R Charts in MINITAB50,The Xbar, s chart is useful when the sample size more than 10,当样本数超过,10,时，,Xbar-s,更为精确,This chart plots averages and standard deviation,控制图以平均值与标准差为基础,Xbar, s charts enable process owners to evaluate changes in central tendency and stability using sub-grouping,Xbar-s,控制图以子组来评估过程的趋势与稳定性,51,Xbar-s Chart,平均值标准差图,X, s,X, s,_,51 Xbar-s Chart 平均值标准差图X, s,Range vs Standard Deviation,极差和标准差,For subgroups of size 2, 3 or 4, there is little difference in accuracy,对于样本数,2,、,3,或,4,， 极差与标准偏差在精度上几乎没有区别,As subgroups exceed 4, the standard deviation becomes increasingly more accurate than the range, to the point that the range should not be used for subgroups greater than 10,当样本数超过,4,后，标准偏差的精度比极差将随样本数的增加而增加，当样本数超过,10,个时，一般不再用极差,。,Strategy: use the std dev approach except when.,策略：使用标准差法，以下情况除外：,Manual calculation is required,需手工计算,Operators (or others), without under-standing,of the standard deviation, will be interpreting the charts,员工对标准偏差概念不了解,52,Range vs Standard Deviation 极差,Xbar & s Control Chart Limits,均值标准差图控制限,The standard deviation can be used instead of the range in variables control charts. Then the Xbar & s Control Chart limits are:,标准差可以取代极差用在连续型数据控制图中。其控制限为：,StDev,X bar,Where the Xbar & s Control Chart constants (A,3, B,3,and B,4,) have been tabulated for various sample sizes (see Appendix),以上常量均有表可查（见附录）,53,Xbar & s Control Chart Limits,平均值图,-,示例,平衡测试数据图,54,Web SPC,平衡测试合格率图,有异常么,?,平均值图-示例平衡测试数据图54Web SPC平衡测试合格率,I-MR Charts,单值移动极差图,I, mR,(also called X,mR),Individual, Moving Range,The Individual Moving Range chart is used when variation within sample subgroups is essentially zero.,单值,-,移动极差图用于子组内变化量为零时,。,This chart plots,individual,measures,over time. 3,control limits are often used.,控制图直接应用随时间测量的单值，,3,为控制限。,Moving ranges, typicall