統(tǒng)計(jì)過程控制-測量系統(tǒng)分析

SPCStatisticProcessControlStatistictechniqueProcessControlSPC由ProductSpec到ProcessParameter由ProductSpec到ProcessParameter1.BasicconceptofSPCVariationControl(Stability)-Statisticalstablestatus-ProcessperformancecanbepredictableCapabilityOver-adjustmentSPC-品管七大手法HistogramCheckSheetParetoCauseandeffectdiagramDefectconcentrationdiagramScatterdiagramControlchartBasicconceptofSPC檢定是否為常態(tài)分配管制圖的中心線X-bar管制圖的管制界限(信賴水平,信賴區(qū)間)ControlLimits=信賴區(qū)間

1.集中趨勢（通常以Xbar作代表）

2.離中趨勢（通常以σ

作代表）

3.信賴水平-被含蓋在特定范圍內(nèi)的機(jī)率

信賴水平-被含蓋在特定范圍內(nèi)的機(jī)率XbarXbar

信賴水平,信賴區(qū)間

信賴水平

信賴區(qū)間

90%

±1.645σ

95%

±1.96σ

99%

±2.575σ

99.73%

±3σ

99.94%±3.5σ99.99%±3.9σ

99.994%±3.99σ信賴水平,信賴區(qū)間,P,Ppk/Cpk信賴水平P信賴區(qū)間Ppk/Cpk90%10%±1.645σ0.54895%5%±1.96σ0.65399%1%±2.575σ

0.85899.73%0.27%

±3σ

199.94%0.06%±3.5σ1.16799.99%100ppm±3.9σ1.399.994%60ppm±3.99σ1.33

ZBasicconceptofSPC用管制圖判別specialcause的準(zhǔn)則超過管制界限連續(xù)七點(diǎn)向上連續(xù)七點(diǎn)向下連續(xù)七點(diǎn)向單邊不隨機(jī)分布BasicconceptofSPCStatisticalstable=nospecialcause=Undercontrol=processperformancecanbepredictable當(dāng)制程達(dá)到Statisticalstable時(shí)才可計(jì)算CpkStableProcessUnstableProcessBasicconceptofSPCOver-adjustment:當(dāng)對commoncause所造成的變異都采取措施調(diào)整制程時(shí),等于將一個(gè)specialcause加入己穩(wěn)定的制程。使原本穩(wěn)定的制程變成不穩(wěn)定。BasicconceptofSPCPpkCaCpCpkBasicconceptofSPCCa:SensitivetoSpecificationtargetCp:Sensitivetovariation(spread)Cpk:Sensitivetoboth,capabilityindexforstableprocessOnlycommoncausevariationincludedPpk:PerformanceindexBothcommoncauseandspecialcausevariationincludedCp≦CpkBasicconceptofSPCVariation:制程輸出不可避免的差異InherentVariation(固有變異):由commoncause所造成的變異σ=Rbar/d2TotalVariation:由commoncauseandspecialcause所造成的變異PpkCpkCpm

管制圖種類包括計(jì)量值管制圖計(jì)數(shù)值管制圖Data的種類計(jì)量值(location&spread)計(jì)數(shù)值計(jì)量值管制圖X-BarRChartX-medianRChartX-BarSChartX-RmChart計(jì)數(shù)值管制圖種類包括

PPnCU

管制圖的目的預(yù)防不良品產(chǎn)出判定制程是否有非機(jī)遇因子造成的變異去除非機(jī)遇因子造成的變異減小制程的變異改善制程判定制程是否穩(wěn)定制程能力分析100%檢驗(yàn)還需耍做管制圖嗎?如何做?特殊應(yīng)用管制圖CUSUMcumulative-SumControlChart)累積和管制圖EWMAexponentiallyWeightedMovingAverageControlChart)指數(shù)加權(quán)移勛平均管制圖EVOP(EvolutionaryOperation)-PartAverageTest漸進(jìn)操作法管制圖AcceptanceControlChart允收管制圖ModifiedControlLimitsControlChart修正界限管制圖ShortRun短制程管制圖

ShortRun短制程管制圖

GuardBandLimit-GRRGuardBandLimit-Cpk

MSAMeasurementSystemAnalysisStatistictechniqueMSControlMSA

MSA

GRRv.s.CPAccuracy&Precision

MSA計(jì)量值

Location-Bias,Linearity,StabilitySpread–Repeatability,Reproducibility(GRR)

RangeMethodAverageRangeMethodAverageRangeMethod(WithinPartVariation)ANOVADOE計(jì)數(shù)值-ShortMethod-LongMethodBiasStabilityLinearityRepeatabilityReproducibility

GRR-Average-RangeMethod%GRR=%R&R=100(R&R/TV)TV2=R&R2+PV2R&R2

=EV2+AV2

TV2EV2AV2PV2GRR-RangeMethodGRR-Average-RangeMethodANOVAMethodMSAtiming&applicationMSA/APQP/ProductRealizationMSA/FMEA

MSA的目的何種情況可不必做GRR?Cpisgood,notuseforprocessmeasurement,C/Inotrequired?只做GRR,不做Bias,Linearity,Stability可以嗎?MSAformeasurementcan’tberepeatedTheinabilitytotestthesameunitrepeatedlyisamajorobstacletoobtaininganestimateofpuremeasurementerror.DestructivetestingMeasuringatemperature,viscositySamplesobtainedfromtheproductstreamshouldbeasclosedtogetheraspossibleWithfilm,onemightpreferadjacentsamplescollectedinthemachinedirectionratherthanfromdifferentpositionsacrossthewebObtainpairsofsamplesthatareashomogeneousaspossiblewithineachpair,eachpairmaybetestedandthedifferenceinthepairedreadingstakentoobtainanestimateofthevariationofthemeasurementprocessDataCategories&ResolutionDataCategories&ResolutionDataCategories&ResolutionDataCategories&ResolutionDataCategories&ResolutionDataCategories&ResolutionDataCategories&ResolutionDataCategories&ResolutionPTRatioPrecisionToleranceRatioTheratioofparametertolerancetomeasurementsystemrepeatabilityPT=Tolerance/EV

MSAResultGageresolutiontospectoleranceGageresolutiontoProcessvariationBiasStabilityLinearityR&RWhenshouldMSAbeconductedinAPQP/ProductRealization/ProductLifecycleProductdesignreviewProcessdes