工具用途1.工程能力分析前做正态性验证。P-value0.05,是正态分布P-value0.05,不是正态分布2.工程能力分析得出工程能力指数。CP:短期工程能力指数CPK:短期工程能力指数(考虑平均值偏移)PP:长期工程能力指数PPK:长期工程能力指数(考虑平均值偏移)3.算出短期工程能力和长期工程能力及工程能力变动。ZST=3CP;ZLT=3PPK;Zshift=ZSTZLT4.做出4-Block图,决定改善方向。A:技术管理都需要改善B:技术优秀,管理需要改善C:管理优秀,技术需要改善D:世界一流水平工程能力分析&正态性验证GageR&R1.测量系统信赖性的三要素。测量者选择水平接近的两个人测量仪器的精度10%ToleranceSample选择在整个Tolerance内2.分析工具的选择。选择可重复测量的方法(Crossed)选择可以区分交互作用的ANOVA方法3.分析结果得出。EV仪器偏差再现性AV人的偏差反复性%GageR&R2=Repeat2+Reproduction24.结果分析。%Studyvariation不考虑公差的变动%Tolerance考虑公差的变动%GageR&R20%,测量系统信赖性可以接受Average:0.785StDev:0.198334N:30Anderson-DarlingNormalityTestA-Squared:0.851P-Value:0.0250.380.480.580.680.780.880.981.08.001.01.05.20.50.80.95.99.999ProbabilityC6NormalProbabilityPlot1)试用下列DATA分析工程能力(考虑subgroupsize)考虑是否正态分布,要说明理由?0.650.601.001.000.850.800.850.950.550.451.001.000.950.950.850.801.001.000.600.700.550.551.050.950.800.750.800.750.400.40正态分布结果P值=0.0250.05,不是正态分布。1)输入数据2)选择命令3)OKPartOperatorResponse110.65211.00310.85410.85510.55611.00710.95810.85911.001010.60110.60211.00310.80410.95510.45611.00710.95810.80911.001010.70120.55221.05320.80420.80520.40621.00720.95820.75921.001020.55120.55220.95320.75420.75520.40621.05720.90820.70920.951020.503)选择ANOVA方法4)输入公差范围12)选择命令1)输入数据5)OK,OKGagename:Dateofstudy:Reportedby:Tolerance:Misc:00.30.40.50.60.70.80.91.01.112XbarChartbyC2SampleMeanMean=0.7975UCL=0.8821LCL=0.712900.000.050.100.1512RChartbyC2SampleRangeR=0.045UCL=0.1470LCL=0123456789100.40.50.60.70.80.91.0C1C2C2*C1InteractionAverage12120.40.50.60.70.80.91.01.1C2ByC2123456789100.40.50.60.70.80.91.01.1C1ByC1%Contribution%StudyVar%ToleranceGageR&RRepeatReprodPart-to-Part050100ComponentsofVariationPercentGageR&R(ANOVA)forC3GageR&R%ContributionSourceVarComp(ofVarComp)TotalGageR&R0.0038759.12Repeatability0.0016253.82Reproducibility0.0022505.29C20.0016603.90C2*C10.0005901.39Part-To-Part0.03863290.88TotalVariation0.042507100.00StdDevStudyVar%StudyVar%ToleranceSource(SD)(5.15*SD)(%SV)(SV/Toler)TotalGageR&R0.0622490.3205830.1932.06Repeatability0.0403110.2076019.5520.76Reproducibility0.0474340.2442923.0124.43C20.0407400.2098119.7620.98C2*C10.0242960.1251211.7812.51Part-To-Part0.1965501.0122395.33101.22TotalVariation0.2061721.06179100.00106.18NumberofDistinctCategories=41)考虑公差范围的GageR&R的值为(32.06)%,测量者引起的误差与测量设备引起的误差哪一个较大,为什麽(把分析结果拷贝,并解释)?2)分析/解释考虑公差范围的GageR&R的结果.TotalGageR&R=32.06%20%,测量系统需要改善。部品间,测量者间以及部品与测量者间的影响哪个较大?(用图形表示)测量者引起的误差是再现性,24.43%。测量设备引起的误差是反复性,20.76%。测量者引起的误差比较大。部品间影响较大。(右侧较高为部品间变动)2)解释工程能力分析的结果,并且说明CP和PPK的含义什麽(从散布的角度说明)?4)选择命令5)Subgroupsize输入50.20.40.60.81.01.21.4LSLUSLProcessCapabilityAnalysisforC6USLTargetLSLMeanSampleNStDev(Within)StDev(Overall)CpCPUCPLCpkCpmPpPPUPPLPpkPPMLSLPPMUSLPPMTotalPPMLSLPPMUSLPPMTotalPPMLSLPPMUSLPPMTotal1.150*0.8500.785300.1884780.2000500.270.65-0.11-0.11*0.250.61-0.11-0.11533333.330.00533333.33634903.2726399.11661302.38627378.4834035.22661413.70ProcessDataPotential(Within)CapabilityOverallCapabilityObservedPerformanceExp.WithinPerformanceExp.OverallPerformanceWithinOverallCP是不考虑平均值偏移的短期工程能力。CP=0.27,ZST=0.81,短期工程能力很低。技术水平低。散布大。Cpk=-0.11,平均值偏移出规格界限。平均值偏移很大。PPK是考虑平均值偏移的长期工程能力。PPK=-0.11,ZLT=-0.33,长期工程能力很低。ZSHIFT=ZST-ZLT=1.24,管理很好。6)输入规格界限值0.85和1.157)OK工具用途1.分散同质性验证(散布的比较)。平均值比较之前进行分散同质性验证P-value0.05,分散同质P-value0.05,分散不同质2.T-Test(两个集团以下平均值的比较)。1-sampleT-Test:一个集团平均值与目标值比较2-sampleT-Test:两个集团比较平均值差异P-value0.05,平均值无差异P-value0.05,平均值有差异3.ANOVA方法(两个集团以上平均值的比较)。根据同质性验证的结果选择是否分散同质图表选择BoxplotP-value0.05,平均值无差异P-value0.05,平均值有差异假设验证回归分析1.回归方程式。得出Y与所有X的关系式R-Sq可说明变动的比例(64%可说明)R-Sq(adj)方程式的信赖性(64%可信赖)2.重要因子筛选。P-value0.05,因子是重要因子P-value0.05,因子不是重要因子3.预测Y值。根据方程式给出X值预测Y值4.Fittingline。Y与X关系的散点图CI:信赖区间,平均值出现概率为95%。PI:预测区间,个别值出现概率为95%。AB38413736323733333638343441453529444829391)原始输入的Data2)选择合并命令3)选择A、B两列4)选择新列5)选择区分列6)OK7)得到合并后的数据列38A37A32A33A36A34A41A35A44A29A41B36B37B33B38B34B45B29B48B39B38A37A32A33A36A34A41A35A44A29A41B36B37B33B38B34B45B29B48B39B1)使用的Data2)选择命令3)选择数据列4)选择区分列5)OK2.53.54.55.56.57.58.59.510.511.595%ConfidenceIntervalsforSigmasBA304050BoxplotsofRawDataC10F-TestTestStatistic:0.605P-Value:0.465Levene'sTestTestStatistic:0.410P-Value:0.530FactorLevelsABTestforEqualVariancesforC10P值=0.4650.05,H0成立,分散同质。1)输入数据2)选择命令A3837323336344135442930354045ABoxplotofA(withHoand95%t-confidenceintervalforthemean)[]X_HoOne-SampleT:ATestofmu=36vsmunot=36VariableNMeanStDevSEMeanA1035.904.381.39Variable95.0%CITPA(32.76,39.04)-0.070.9443)选择图表4)OK1)运用假设验证判断A业体的物量精度是否达到36,并说明其理由P值=0.9440.05,H0成立,已经达到36。38A37A32A33A36A34A41A35A44A29A41B36B37B33B38B34B45B29B48B39B1)使用的Data2)选择命令3)选择数据列4)选择区分列5)选择分散同质6)选择图表7)OKAB304050C11C10BoxplotsofC10byC11(meansareindicatedbysolidcircles)Two-SampleT-TestandCI:C10,C11Two-sampleTforC10C11NMeanStDevSEMeanA1035.904.381.4B1038.005.641.8Difference=mu(A)-mu(B)Estimatefordifference:-2.1095%CIfordifference:(-6.84,2.64)T-Testofdifference=0(vsnot=):T-Value=-0.93P-Value=0.365DF=18BothusePooledStDev=5.05P值=0.3650.05,H0成立,平均值没有差异。One-wayANOVA:C10versusC11AnalysisofVarianceforC10SourceDFSSMSFPC11122.122.10.860.365Error18458.925.5Total19481.0Indiv