6s的意义与计算s是一个希腊字母,用来描述过程的变差6s最初由Motorola公司发起,逐步被越来越多的世界级大公司所采用68.2%的数据落在±1s以内95.4%的数据落在±2s以内99.7%的数据落在±3s以内99.99999975%的数据落在±6s以内3s2s1s1s2s3sLSL下限USL上限平均值m或目标值Ts个数的计算(连续数据)Z=sMin{(USL-T),(T-LSL)}Z=3Cpk过程的漂移短期4短期3短期2短期1长期s的个数Zlt(长期)=Zst(短期)-1.5计算s个数的概念与公式Unit:单元,被研究对象的基本单位。如一件产品、一次服务、一个过程等。Defect:缺陷,任何未达到要求的特性DPU(Defectsperunit)=Defects/Unit=缺陷数/单元数PPM(DefectsperMillionUnits)=Defects/Unitx106=缺陷数/(单元数X106)TOP(TotalOpportunities)=Units*Opportunities(总机会)=单元数X每单元发生缺陷的机会DPO(DefectsperOpportunity)=Defects/TOP(每个机会中产生的缺陷数)=缺陷数/总机会DPMU(Defectspermillionunits)=DPUx106(每百万单元中的缺陷数)DPMO(DefectsperMillionOpportunities)=Defects/TOPx106=缺陷数/TOPx106FirstPassYield(一次合格率)=全过程中未产生任何缺陷的单元数/总单元数RolledYield(单元的缺陷数为零的几率)(Thelikelihoodthatanygivenunitofproductwillcontain0defects)YRT=e-DPUYRT=P(ND)*P(ND)*P(ND)*......P(ND)nP(ND)表示零缺陷的几率Ynorm(NormalizedYield正态合格率)=1-DPO=nYRT(连续过程的平均合格率)n=过程的个数Zlt=NormSinv(Ynorm)(在电子表格Excel的函数fx中的‘统计类’中,或查表)Zst=Zlt+1.5计算s个数的概念与公式(非连续数据)计算s个数实例M(每单元产生缺陷的机会)=50Units(单元数)=1000Defects(缺陷数)=500TOP=MxUnits=50000DPO=Defects/TOP=0.01DPMO=DPOx106=10000Ynorm=1-DPO=0.9900Zlt=NormSinv(Ynorm)=2.33Zst=Zlt+1.5=3.8393.32%Long-TermYieldLong-TermYield42308,537366,8076,210523363.4不同s个数的比较(DistributionShifted±1.5s)sPPM过程能力DPMO345671,000,000100,00010,0001,0001001012Sigma的个数PPMTheBasicObjective一般公司Best-in-Class世界级优秀公司----------------------------------ProcessEntitlement目标SweetFruit最甜的果实DesignforManufacturabilityBulkofFruit大量的果实ProcessCharacterizationandOptimization过程优化LowHangingFruit底处的果实SevenBasicTools统计七工具GroundFruit掉到地上的果实LogicandIntuition基本常识----------------------------------Wedon'tknowwhatwedon'tknowWecan'tactonwhatwedon'tknowWewon'tknowuntilwesearchWewon'tsearchforwhatwedon'tquestionWedon'tquestionwhatwedon'tmeasureHence,Wejustdon'tknowWHATLEVELAREYOU?6s的意义与计算s是一个希腊字母,用来描述过程的变差6s最初由Motorola公司发起,逐步被越来越多的世界级大公司所采用68.2%的数据落在±1s以内95.4%的数据落在±2s以内99.7%的数据落在±3s以内99.99999975%的数据落在±6s以内3s2s1s1s2s3sLSL下限USL上限平均值m或目标值Ts个数的计算(连续数据)Z=sMin{(USL-T),(T-LSL)}Z=3Cpk过程的漂移短期4短期3短期2短期1长期s的个数Zlt(长期)=Zst(短期)-1.5计算s个数的概念与公式Unit:单元,被研究对象的基本单位。如一件产品、一次服务、一个过程等。Defect:缺陷,任何未达到要求的特性DPU(Defectsperunit)=Defects/Unit=缺陷数/单元数PPM(DefectsperMillionUnits)=Defects/Unitx106=缺陷数/(单元数X106)TOP(TotalOpportunities)=Units*Opportunities(总机会)=单元数X每单元发生缺陷的机会DPO(DefectsperOpportunity)=Defects/TOP(每个机会中产生的缺陷数)=缺陷数/总机会DPMU(Defectspermillionunits)=DPUx106(每百万单元中的缺陷数)DPMO(DefectsperMillionOpportunities)=Defects/TOPx106=缺陷数/TOPx106FirstPassYield(一次合格率)=全过程中未产生任何缺陷的单元数/总单元数RolledYield(单元的缺陷数为零的几率)(Thelikelihoodthatanygivenunitofproductwillcontain0defects)YRT=e-DPUYRT=P(ND)*P(ND)*P(ND)*......P(ND)nP(ND)表示零缺陷的几率Ynorm(NormalizedYield正态合格率)=1-DPO=nYRT(连续过程的平均合格率)n=过程的个数Zlt=NormSinv(Ynorm)(在电子表格Excel的函数fx中的‘统计类’中,或查表)Zst=Zlt+1.5计算s个数的概念与公式(非连续数据)计算s个数实例M(每单元产生缺陷的机会)=50Units(单元数)=1000Defects(缺陷数)=500TOP=MxUnits=50000DPO=Defects/TOP=0.01DPMO=DPOx106=10000Ynorm=1-DPO=0.9900Zlt=NormSinv(Ynorm)=2.33Zst=Zlt+1.5=3.8393.32%Long-TermYieldLong-TermYield2308,537366,80746,210523363.4不同s个数的比较(DistributionShifted±1.5s)sPPM过程能力DPMO345671,000,000100,00010,0001,0001001012Sigma的个数PPMTheBasicObjective一般公司Best-in-Class世界级优秀公司--------------------------------------------------------------------Wedon'tknowwhatwedon'tknowWecan'tactonwhatwedon'tknowWewon'tknowuntilwesearchWewon'tsearchforwhatwedon'tquestionWedon'tquestionwhatwedon'tmeasureHence,Wejustdon'tknowLowHangingFruit底处的果实SevenBasicTools统计七工具ProcessEntitlement目标SweetFruit最甜的果实DesignforManufacturabilityBulkofFruit大量的果实ProcessCharacterizationandOptimization过程优化GroundFruit掉到地上的果实LogicandIntuition基本常识WHATLEVELAREYOU?