SPC0

0SPC培训教材1SPC培训教材统计过程控制讲义目录一、基礎統計二、前言（有关SPC的基本内容）三、数据的收集、整理与分析四、控制图的基本概念五、计量值控制图六、控制图分析过程控制七、过程能力分析“感觉上的东西不一定能理解，理解的东西必有深刻的感觉。”2MeasuresofLocation•Mean•Median•Mode•Quartiles基礎統計统计过程控制讲义一、基礎統計3Mean•Iftheobservationsinasampleofsizenarex1,x2,...,xn,•thenthesamplemeanis•Themeanisthemostcommonmeasureoflocationorcenter•ofthedata.nxnxxxxn1iin21统计过程控制讲义基礎統計4Mean•Thepullstrength(ingf)of10goldbondingwiresare16.8516.4017.2116.3516.5217.0416.9617.1516.5916.57•Thesamplemeanpullstrengthforthe10observationsisgf764.161064.1671057.1640.1685.16nxxn1ii统计过程控制讲义基礎統計5Mean•Thesamplemeanrepresentstheaveragevalueofallobservationsinthesample.•ForafinitenumberofobservationsN,thepopulationmean(denotedby)maybedeterminedbyxNxN1ii统计过程控制讲义基礎統計6Mean•DuringOperationDesertStormin1991,USAFF-117Apilotsflew1270combatsortiesforatotalof6905hours.•Hence,themeandurationofaF-117Amissionduringthisoperationwashours4.512706905NxN1ii统计过程控制讲义基礎統計7Median•Letx(1),x(2),...,x(n)denoteasamplearrangedinincreasingorderofmagnitude,thenthesamplemedianisdefinedas•Theadvantageofthemedianisthatitisnotinfluencedverymuchbyextremevalues.evenisnif2xxoddisnifxx~)1]2/n([)2/n()2/]1n([统计过程控制讲义基礎統計8Median•Ifthesampleobservationsare1342786•Thesamplemeanandmedianare4.4and4respectively.•Bothquantitiesgiveareasonablemeasureofthecentraltendencyofthedata.•Ifthelastobservationischangedsothatthedataare1342782450•Thesamplemeanis353.6whilethesamplemedianremainsunchanged.统计过程控制讲义基礎統計9Median•Justasthesamplemedianisthemiddlevalueinasample,•thereisamiddlevalueinthepopulation.•Thepopulationmedianisthatvalueatwhichhalfthepopulationliesbelowitandhalfliesabove.x~~统计过程控制讲义基礎統計10Mode•Themodeistheobservationthatoccursmostfrequentlyinthesample.•Themodemaybeunique,ortheremaybemorethan1mode.•Sometimes,themodemaynotexist.统计过程控制讲义基礎統計11Mode•Ifthesampleobservationsare3693583463110•Thesamplemodeis3,sinceitoccursfourtimes.•Ifthesampleobservationsare36935834631106256•Thesamplemodesareat3and6,sincetheybothoccurfourtimes.•Ifthesampleobservationsare1342768•Thesamplemodedoesnotexist.统计过程控制讲义基礎統計12Quartiles•Whenanorderedsetofdataisdividedintofourequalparts,•thedivisionpointsarecalledquartiles.•Thefirstorlowerquartileq1isavaluethathasapproximately25%oftheobservationsbelowinvalue.•Thesecondquartileq2isavaluethathasapproximately50%oftheobservationsbelowinvalue.Itisalsocalledthemedian.•Thethirdorupperquartileq3isavaluethathasapproximately75%oftheobservationsbelowinvalue.统计过程控制讲义基礎統計13Quartiles•Twentyorderedobservationsonthetimestofailure(inhours)•ofelectricalinsulationmaterialareshownbelow.20422825230032444462472081691211761296139214881512252028563192352837102016225201512q104421176912qx~3842444324q321统计过程控制讲义基礎統計14MeasuresofDispersion•Range•Variance•StandardDeviation•Inter-QuartileRange统计过程控制讲义基礎統計15Range•Thesamplerangeisdefinedasthedifferencebetweenthe•largestandthesmallestsampleobservations,i.e.r=x(max)–x(min)•Themeanisthesimplestmeasureofdispersionorvariationofthedata.•However,itignoresalltheinformationinthesamplebetween•thesmallestandthelargestobservations.统计过程控制讲义基礎統計16Range•Considerthetwosamples1,3,5,8,9and1,5,5,5,9.•Bothhavethesamerange(r=8).However,inthesecondsamplethereisvariabilityonlyinthetwoextremevalues,whileinthefirstsamplethemiddlevaluesvaryconsiderably.•Whenthesampleissmall(n10),theinformationlossassociatedwiththerangeisnottooserious.统计过程控制讲义基礎統計17MeasuresofShape•Skewness•Kurtosis统计过程控制讲义基礎統計18Skewness•Thedegreeofassymmetryofadistributionarounditsmeanis•referredtoasitsskewness.•Positiveskewnessimpliesadistributionwithanasymmetrictailextendingtowardshighervalues.•Sometimesreferredtoasright-handedskew.•Negativeskewnessimpliesadistributionwithanasymmetrictailextendingtowardslowervalues.•Sometimesreferredtoasleft-handedskew.统计过程控制讲义基礎統計19Skewness统计过程控制讲义基礎統計20Skewness•Ifthedataaresymmetric,themeanandmedianwillcoincide.•Ifthedataisunimodal,thenthemean,medianandmodewillallcoincide.•Ifthedataareskewed,themean,medianandmodewillnotcoincide.•Forright-handedskewness:modemedianmean•Forleft-handedskewness:modemedianmean统计过程控制讲义基礎統計21Kurtosis•Kurtosischaracterizestherelativepeakednessorflatnessofadistributioncomparedtoanormal(mesokurtic)distribution.•Positivekurtosisindicatesarelativelypeaked(leptokurtic)distributionomparedtothenormaldistribution.•Negativekurtosisindicatesarelativelyflat(platykurtic)distributioncomparedtothenormaldistribution.•Kurtosisisrelevantonlyforsymmetricaldistributions.统计过程控制讲义基礎統計22Kurtosis统计过程控制讲义基礎統計23前言统计过程控制讲义•检验、控制与预防---对实体的一个或多个特性进行诸如测量、检查、试验或度量，并将结果与规定要求进行比较，以确定每项特性合格状况所进行的活动。通常控制产品质量的方法是通过对产品的最终检验，并将检验中发现的不合格品进行剔除。这是检验。---对过程中的输出结果进行监控并分析反馈，以符合规格的要求。---生产前或第一步就可以避免生产过程中出现的不良品，从而避免浪费和给客户带来不便的有效方法是预防。第一次就把工作作好！运筹帷幄，决胜千里之外！二、前言24统计过程控制讲义传统的产品检验模式1．检验要有判定标准和检验方法2．检验要有记录3．检验员要经过相关训练4．检验信息要传递至相关部门输入输出检验判定输入生产过程返工、返修检验要素：转序交付纠正措施25统计过程控制讲义控制与检验检验----容忍浪费预防----避免浪费检验的缺陷(与控制相比)1．成本高--不从事生产，越依赖检验，做的越多，投入人力物力越多，成本越高。2．可靠性有限--据统计70%的不良品不被检出，而且每个检验员的检出结果亦不同。3.实效性不够--检验无法对产品做出贡献，只是在出货前或转入下一工序前做检验或测试。当发现问题时，此时产品已被判定为不良品。4．无法反应在规格标准的缺失-