过程能力分析研究

能力分析CapabilityStudies过程能力判断的第一步：SPCS=统计技术过去经常检查偏差StatisticaltechniquesusedtoexamineprocessvariationC=控制过程通过积极管理ControllingtheprocessthroughactivemanagementP=过程,任何过程Process,ANYProcess现在我们管理数据的方法-SPCUCLLCL1Sigma(ZoneC)2Sigma(ZoneB)3Sigma(ZoneA)1Sigma(ZoneC)2Sigma(ZoneB)3Sigma(ZoneA)时间TIME我们测量的项目TheItemWeAreMeasuring应用测试ApplyingtheTests1Sigma2Sigma3Sigma1Sigma2Sigma3Sigma60-75%90-98%99-99.9%%ofDataPointsUCLLCL时间TIME我们测量的项目TheItemWeAreMeasuring标准偏差的规则:数据应该在哪?控制极限对规格限制ControlLimitsvs.SpecificationLimits•过程控制极限是在过程中数据自己计算出来的ProcessControlLimitsarecalculatedbasedondatafromtheprocessitself•他们根据+/-3s(99.73%我们期望过程偏差落在这些极限之间)Theyarebasedon+/-3s(99.73%oftheprocessvariationisexpectedtofallbetweentheselimits)确定过程执行如何满足顾客期望,需要进行过程能力研究。TodeterminehowtheprocessperformstoCustomerExpectations,aProcessCapabilityStudyisrequired.两种一般数据TwoGeneralKindsofData属性ATTRIBUTE-离散,可计的数据Discrete,CountedDataEx:1,2,3,4etc…Good/BadMachine1,2,3...变量VARIABLES-连续,可测量的数据Continuous,MeasuredDataEx:Weight=10.2LbsThickness=11.211inches控制图的主要类型MajorTypesofControlCharts•变量图(连续的数据）VariablesCharts–I-MR(个体individuals)–X-Bar(组（平均）average)•特性图（离散的数据）AttributeCharts–NP(有缺陷的数字Numberdefective)–P(有缺陷（不合格）的比率Proportiondefective)–C(过失数量Numberofdefects)–U(每个单位的过失数量Numberofdefects/unit)过程受控判别的八个准则•准则七：过程改善后变异变小，控制限要求太低，应该把控制限变小。准则八则与其相反。•过程能力判断的第二步：正态性（1）正态性检测的路径：统计→基本统计量→正态性检验（2）一般情况下当P值﹥0.05时表明数据符合正态分布。数据是否正态?IsdataNormal?正态能力分析NormalCapabilityAnalysis计量数据能力分析路线图VARIABLESCAPABILITYANALYSISROADMAP计量型数据Variablesdata检查稳定性CheckStabilityYN转换数据TransformData能力分析,只看图表CapabilityAnalysisLookonlyatgraphs西格码计算器SigmaCalculator黑带大师SeeMBB•由上图可以看出只有数据过程受控和符合正态分布的情况下才能进行下一步，也就是我们的关注重点：过程流程能力。能力指数CapabilityIndexSpecWidth(road)USL-LSLCp=MfgCapability(car)=±3sTLSLUSL-3s+3s可见的流程能力43210-1-2-3-40.40.30.20.10.0LowerSpec.LimitUpperSpec.LimitCust.Tolerance86420-2-4-6-80.40.30.20.10.0LowerSpecLimitUpperSpec.LimitCust.ToleranceProcessCapabilityProcessCapabilityCp=1Cp=2哪个好些?为什么?Whichisbetter?Why?流程能力比率ProcessCapabilityRatiosCp&Cpk是2个测量流程能力的主要指标Cp&Cpkaretwokeymeasuresofprocesscapability.提示,S在这里是样本总体标准差的估计Note:Sinthiscaseistheestimateofthepopulationstandarddeviationbasedonthesample.USLC-LSL6psCMin(X-LSL3USL-X3pkss,)27mm=LSLUSL=33mmxbar=30mms=133Cp=__________ssCUSL-LSL6ps33ssxbar=30mms=0.333mmCp=__________CUSL-LSL6ps28mm=LSLUSL=32mm33ssxbar=33mms=0.333mmCp=__________CUSL-LSL6ps28mm=LSLUSL=32mm是否流程偏离了目标更大的设计余量能带来更好的累计产出率Cp=1.33Cpk=1.335.334.02.671.33-1.33-2.67-4.0-5.3300.40.30.20.10.0LowerSpec.LimitUpperSpec.LimitCust.Tolerance00.40.30.20.10.05.334.02.671.33-1.33-2.67-4.0-5.33LowerSpec.LimitUpperSpec.LimitCust.ToleranceCp=1.33Cpk=0.83流程能力比率ProcessCapabilityRatiosCMin(X-LSL3USL-X3pkss,)CX-LSL3pLsUSL-X3sCpUCpk说明了流程居中和分散的程度Cpkaccountsforprocesscenteringandspread.CpU=___________CpL=___________Cpk=____________33ssxbar=33mms=0.333mm28mm=LSLUSL=32mm能力分析中的统计假设1.数据来自稳定的流程Datacomesfromastableprocess如果不是,先尽量控制流程Ifnot,worktowardgettingtheprocessincontrol2.正态分布HasaNormalDistribution如果没有,找个可替代的能力测量方法Ifnot,usealternativecapabilitymeasure(wewillcoverthislaterinthecourse)如果假设1和2不能满足,结果将被误导.Minitab的能力6项会帮你检查这些假设.Ifitems#1and#2aren’tmet,resultswillbemisleading.Minitab’scapabilitySixpackhelpsyouchecktheseassumptions能力实例CapabilityExample让我们从020CapabilityStudiesData03JAN05.MPJ文件中的工作表“WaitTimeTest”来计算等候时间的数据能力.Let’scalculatecapability(CpandCpk)withthewaittimedatafromworksheetWaitTimeTestinthe020CapabilityStudiesData03JAN05.MPJfile.客户等候时间的规格界限：SpecificationLimitsforCustomerWaitTime:•USL(规格上限)=16(UpperSpecificationLimit)•LSL(规格下限)=10(LowerSpecificationLimit)MINITAB能力分析:统计等候时间Minitab:StatQualityToolsCapabilityAnalysisNormal…Minitab能力分析:统计等候时间16151413121110LSLUSLProcessDataSampleN30StDev(Within)0.428271StDev(Overall)0.469257LSL10Target*USL16SampleMean13.5106Potential(Within)CapabilityOverallCapabilityPp2.13PPL2.49PPU1.77Ppk1.77CpmCp*2.33CPL2.73CPU1.94Cpk1.94ObservedPerformancePPMLSL0.00PPMUSL0.00PPMTotal0.00Exp.WithinPerformancePPMLSL0.00PPMUSL0.00PPMTotal0.00Exp.OverallPerformancePPMLSL0.00PPMUSL0.06PPMTotal0.06WithinOverallProcessCapabilityofSTWaitTimeProject:020CAPABILITYSTUDIES03JAN05.MPJ;Worksheet:WaitTimeTest;1/14/20051:19:48PM这些数据显示期望的合理能力(Cpk)是1.94,但是这个流程能达到多好?Thesedatashowtheexpectedcapability(Cpk)is1.94,whichisreasonablecapability.Buthowgoodcanthisprocessget?能力分析结论CapabilityAnalysisConclusions•能力分析(正态)Fromthe“CapabilityAnalysis(Normal)–流程基本是居中的Theprocessisalmostcentered–流程形态是正态的Theprocessshapeisnormal–流程分散程度比规格要窄Theprocessspreadisnarrowerthanthespecs•改进策略Improvementstrategy–居中–均值轻微移动Center–shiftmeanslightly–形状–保持流程控制Shape–maintainprocesscontrol–分散程度–减少偏差Spread–decreasevariation正态的能力6项能力6项-统计等候时间在控制中吗?IsItInControl?是正态吗?IsitNormal?最后25个点看起来怎样?Whatdothelast25ptslooklike?流程偏差与规格的比较如何?Howdoestheprocessvariationcomparetothespeclimits?IndividualValue28252219161310741151413_X=13.511UCL=14.795LCL=12.226MovingRange282522191613107411.60.80.0__MR=0.483UCL=1.578LCL=0ObservationValues302520151014.413.612.816151413121110LSLUSLSpecificationsLSL10USL1615141312WithinOverallSpecsWithinStDev0.428271Cp2.33Cpk1.94Ov