统计运用与品管实务工具培训课程培训课件.ppt

统计运用与品管实务工具培训课程單元(二)統計運用及品管實務工具資料及數據離散型資料(通常)分組/分類是/否,合格/不合格不能計算離散型資料分級很少用很難加以計算連續型資料最常見的尺規計算時要很小心連續型資料比例關係可應用演算法的多數公式基礎統計運用概念變異(Variation)變異(Variation)變異(Variation)變異(Variation)變異(Variation)控制變異(Variation)因為用抽樣統計,其結果只是估計,和真實可能有差異.適當的抽樣可使統計分析更準確.抽樣方法一般準則均值:一組值的算術平均均值:-反映所有值的影響-受極值影響嚴重中位數:反應50%的序一組數排序後居中的數-在計算中不必包含所有值-相對於極值具有“可靠性”眾數值:-在一組資料中最常發生的值全距:在一組資料中，最高值和最低值間的數值距離變異(s2):每個資料點與均值的平均平方偏差標準偏差(s):變異數的平方根.量化變動最常用的量The"Rule"stateshowandcanbeusedtodescribetheentiredistribution:·Roughly60-75%ofthedataarewithin1of.·Roughly90-98%ofthedataarewithin2of.·Roughly99-100%ofthedataarewithin3of.Theshapeofadistributioncanbedescribedbyskewness歪斜(denotedby1)andbykurtosis凹凸平坦(denotedby2).母體均值ThemostimportantandusefuldistributionshapeiscalledtheNormaldistribution,whichissymmetric(對稱),uni-modal(單峰),andfreeofoutliers(沒有特異點):ANormalprobabilityplotisacumulativedistributionplotwheretheverticalscaleischangedinsuchawaythatdatafromaNormaldistributionwillformastraightline:第一個屬性:只要知道下面兩項就可以完全描述常態分配:均值標準差常態曲線和其概率常態概率圖資料收集時的重點ContinuousManufacturing(連續)occurswhenanoperationisperformedononeunitofproductatatime.Anassemblylineistypicalofacontinuousmanufacturingenvironment,whereeachunitofproductisworkedonindividuallyandacontinuousstreamoffinishedproductsrollofftheline.TheautomotiveindustryisoneexampleofContinuousManufacturing.Otherexamplesofcontinuouslymanufacturedproductare:·televisionsets,·fastfoodhamburgers,·computers.InContinuousManufacturingthemostimportantvariationisbetweenpartsInLot/BatchManufacturing,thevariationcanoccurbetweenthepartsinalotandbetweenthelots:·Productwithinthesamelotismanufacturedtogether.·Productfromdifferentlotsaremanufacturedseparately.Becauseofthis,eachlothasadifferentdistribution.ThisisimportantbecauseContinuousManufacturingisabasicassumptionformanyofthestandardstatisticalmethodsfoundinmosttextbooksorQChandbooks.ThesemethodsarenotappropriateforLot/BatchManufacturing.DifferentstatisticalmethodsneedtobeusedtotakeintoaccounttheseveralsourcesofvariationinLot/Ba