SPC-抽样教程

SPCPartII:SamplingPlansforX-chartFSG-SM00RevB2LearningObjectives•RevisionofSPCFundamentals/SPC原理校正MisconceptionsofSPC/SPC误解ProcessControlModel/制程控制模型TwoBasicCausesofVariability/2种基本变异原因ReviewofX-RChartandX-sChart/均值极差及单值控制图回顾ReviewofAttributesControlCharts属性控制图回顾•DesigningtheX-chart均值控制图的设计SamplingRisks(aandb)抽样风险ARL0andARL1DeterminingSamplingInterval决定抽样时间间隔DeterminingSampleSize决定样本大小FSG-SM00RevB3MisconceptionsofSPC•Let’scheckourunderstandingbylookingatthemostcommoncontrolchart,X-chart.通过回顾基本的均值控制图，检阅我们对SPC的了解252015105080706050SampleNumberSampleMeanX-BarChartforProcessBX=70.98UCL=77.27LCL=64.70LSLUSLPuttingUSL&LSLonX-charthelpsensurethatpartsaremeetingCpkrequirements.MythNo.1ItisalrighttoshrinkUSL&LSLdownto,say,70%or80%,toestablishUCLandLCL.MythNo.2FSG-SM00RevB4•MoremythsandmisconceptionsontheX-chart.对均值控制图的误解252015105080706050SampleNumberSampleMeanX-BarChartforProcessBX=70.98UCL=77.27LCL=64.70LSLUSLFortunately,partsstillmeetcustomerspec.althoughprocessisoutofcontrolMythNo.3Processincontrol:thereforepartsmeetcustomerspec.aswellasCpkrequirementsMythNo.4MisconceptionsofSPCFSG-SM00RevB5252015105080706050SampleNumberSampleMeanX-BarChartforProcessBX=70.98UCL=77.27LCL=64.70LSLUSL•MoremythsandmisconceptionsoftheX-chart.对均值控制图的误解Howdidweestablishthesamplingplan?-Gutfeel?-Passed-downfigures?-Statisticalcalculation?-Technicaljudgement?5pcs/2hrMythNo.6MisconceptionsofSPCMythNo.5A2RWeknowA2R=3s.Butwhichs?sx,sx,swithin,soverall??FSG-SM00RevB6ProcessControlModelforQualityControlRawMaterial,Components&Sub-Assemblies原材料，零件及半成品ProcessProductObservation观察:DataCollection数据收集Evaluation评估:DataAnalysis数据分析Diagnosis诊断:FaultDiscovery错误发现Decision决策:FormulateAction阐明行动Implementation推行:TakeAction采取行动UncontrollableInputs不可控的输入ControllableInputs可控的输入FSG-SM00RevB7StatisticalProcessControl•Theprocesscontrolmodelshiftsfocustothehomefront,i.e.themanufacturingprocess,takingapreventiveinsteadofreactivemode.制程控制模型具有前瞻性，如，制造制程，是采取预防措施而非直接解决问题•Italsohassomethingwhichtheoldconceptofproductcontrollacked-statistics.Thisallowsuseofsamplestounderstandtheentireprocess.较早时期产品控制缺乏统计这一让我们明白整个制程的概念•Thenewemphasishadtohaveaname-StatisticalProcessControl(SPC).新的产品控制观念强调了SPC•WeowetheapplicationofstatisticsasatoolformanufacturingtoDrWalterA.Shewhart.统计在制造中的应用应归功于修哈特博士FSG-SM00RevB8TwoBasicCausesofVariability•ChanceCausesofVariation变异的偶因Duetothecumulativeeffectofmanysmallunavoidablesourcesofvariation.许多不可避免的小的变异源的累积Alsoknownas:commonvariation普通变异randomvariation随机变异inherentvariation内在变异naturalvariation自然变异Aprocessoperatingwithonlychancecausesofvariationissaidtobe“instatisticalcontrol”.仅有偶因变异的制程处于统计控制FSG-SM00RevB9TwoBasicCausesofVariability•Assignable(orSpecial)CausesofVariation变异的异因Variationinaprocessthatisdifferentfromfromchancevariation;disturbsaprocesssothatwhatitproducesseemsunnatural.这种变异明显异常于偶因引起的变异，制程受到明显的影响，显得不自然。Examplesofsuchcausesofvariationare异因引起的变异的例子:improperlyadjustedmachine不正常的调机excessivetoolwear过多使用工具defectiverawmaterial使用有缺陷的原材料Aprocessoperatinginthepresenceofassignablecausesofvariationissaidtobe“out-of-control”异因引起的变异的制程处于不受控.FSG-SM00RevB10ObjectivesofSPCCharts•Allcontrolchartshaveoneprimarypurpose!所有控制图均有一个基本的目的Todetectassignablecausesofvariation探测严重引起制程变异的异因，因此：thatcausesignificantprocessshift,sothat:investigationandcorrectiveactionmaybeundertakentoridtheprocessoftheassignablecausesofvariationbeforetoomanynon-conformingunitsareproduced.大量不合格品产生前，必须调查研究并采取改善行动，以消除起制程变异的异因inotherwords,tokeeptheprocessinstatisticalcontrol.换句话说，使得制程处于统计控制FSG-SM00RevB11ObjectivesofSPCCharts•Thefollowingaresecondaryobjectivesordirectbenefitsoftheprimaryobjective:以下是采用控制图的好处Toreducevariabilityinaprocess.减少制程的变异Tohelpestimatetheparametersofaprocessandestablishitsprocesscapability.帮助估计制程参数，建立制程能力FSG-SM00RevB12ProcessControl制程控制•Meansthatchancecausesaretheonlysourceofvariationpresent.偶因是制程变异的唯一因数•Refersto“voiceoftheprocess”,i.e.weonlyneeddatafromtheprocesstodetermineifaprocessisincontrol控制图反映制程的声音，我们仅需要制程的数据来决定制程是否受控.•Qualitycharacteristicismonitoredtoverifyifitformsastabledistributionovertime,withcontrollimitscomputedfromtheprocessdataonly通过监控质量特性，由制程数据计算制程的控制范围，检验制程是否形成在一定控制范围内的稳定的分布.•Justbecauseaprocessisincontroldoesnotnecessarilymeanitisacapableprocess仅仅因为制程受控不能够确定制程是有能力.FSG-SM00RevB13•The“goodness”ofaprocessismeasuredbyitsprocesscapability制程的好会是由它的能力来量度的.•Compares“voiceoftheprocess”with“voiceofthecustomer”,whichisgivenintermsofcustomerspecs.orrequirements比较制程的声音和客户的声音，客户的声音即由客户给定规格或要求.•Measureshowwellastabledistribution(processincontrol)meetscustomerrequirementsbytheproportionofproductswithinoroutofcustomerspecs衡量一个稳定的制程如何满足客户的需要是看它的产品在客户规格范围内的分布情况来确定.ProcessCapability制程能力FSG-SM00RevB14ControlLimitsvsSpec.LimitsSpecificationLimits(USL,LSL)规格极限•determinedbydesignconsiderations由设计来决定•representthetolerablelimitsofindividualvaluesofaproduct产品的个别值代表偏差•usuallyexternaltovariabilityoftheprocess通常外在环境影响制程的变异ControlLimits(UCL,LCL)控制极限•derivedbasedonvariabilityoftheprocess来源于制程的变异•usuallyapplytosamplestatisticssuchassubgroupaverageorrange,rathe