1/127ProcessCertification工序认证2/127统计基础1Content工序认证步骤23/127母体:所研究的统计量的总体;样本:母体的子集母体和样本Population&sample母体样本统计估算4/127数据的种类1.计量值:连续性可测量的数据(mm,kg,ton….)测量值Measurement:0.25625/127数据的种类Fail/NogoPass/Go很小VerySmall小Small中等Medium大Large很大VeryLarge2.计数值:离散型的计点或计件值1.2.3…(unit,piece…)3.属性值:判断物体可接受的程度6/127最大值与最小值之差Max-Min1.估计标准差的计算2.直方图的绘制全距/极差Range7/127平均值mean&中位数median平均值Mean:代表样本数据值的中心位置-(nN,Xµ)中位数Median:代表样本数据排序的中心位置中位数可理解为第二四分位数(Q2)---Quartile或第50百分位数(P50)---PercentileLSLUSLP50P25P75Q3Q2Q1平均值mean&中位数median9/127计数值统计数据缺点(defects):代表一个单位产品不符合要求的点数不良(defectives):代表一个单位产品有一个以上不符合要求的缺点不良件数np:5,8,12…缺点数c:13,2,50…不良率P%:5.00%,12.0%….单位缺点数DPU:0.23,0.02…百万机会缺点数DPMO:1200,2000…10/127DefectpermillionopportunityDPMO的计算DPMO=(总缺点数/总缺点机会数)*1000000=defects/totalopportunities*1000000总缺点机会包括以下几个方面:A.产品的复杂程度(零件数,功能数)B.制程的复杂程度(作业数,附着或连接数)C.软体的复杂程度(程序的行数)D.处理业务的复杂程度(业务表单处理的栏数)注:无附加值作业不能算机会数,如:物料的储运,检验与测试…11/127标准差StandardDeviation()代表离平均值的差距(nN,Ѕ)标准差12/127•“Sigma”是一个统计学的质量量测尺度•用于量测服务,制程和产品质量改善过程中的测量,分析,去提高和控制什么是Sigma?用于衡量不同产品或制程复杂程度的品质指标DPMO13/127Sigma和DPMO的关系345676681062102333.4.02SigmaDPMO(DefectsPerMillionOpportunitieswith1.5shift)14/127Sigma水平合格率DPMOSigma和DPMO的关系15/127相互换算练习:某一主机板的加工制程,有50个零件组成,50个焊点,经自动测试机检查100片中有29个缺点,请计算:DPU=DPO=DPMO=总缺陷数/总检验单位数=Defects/Units=29/100=0.29DPU/产品或制程机会数=0.29/100=0.0029DPU/产品或制程机会数X1000000=2900PPM=不合格数/总数量X100000016/127偏态Skewness&峰度Kurtosis偏态Skewness:描述数据分布的对称程度=0无偏0右偏或正偏0左偏或负偏峰度Kurtosis:描述数据分布的宽尖程度=0常态0较常态宽0较常态尖=000=00017/127一次合格率FTY&流通合格率RTY一次合格率FTY:FirsttimeyieldA:99.1%B:98.4%C:99.7%D:99.4%inputoutputFTY=output/inputRTY=A*B*C*D流通合格率RTY:rolledthroughoutyield制造能力制程能力18/127正态分布SPC的基本概念19/127质量统计数据data(单位:mm)3.883.843.843.883.813.783.883.883.903.843.933.833.973.853.893.903.883.813.943.833.883.853.843.833.913.923.943.793.903.813.843.903.893.833.863.773.873.773.903.943.843.903.853.843.903.983.903.803.833.843.883.933.783.813.843.793.813.733.753.843.823.873.883.853.863.743.883.803.843.983.853.873.833.833.893.833.913.853.773.933.833.943.893.933.833.883.853.793.953.993.913.903.873.763.833.853.893.803.833.853.933.893.963.813.903.813.873.823.843.863.893.923.903.773.803.853.843.773.793.793.863.873.843.873.873.943.903.863.973.853.843.913.843.903.863.893.793.783.863.843.843.873.793.833.903.913.783.843.893.793.883.703.763.793.803.843.863.833.893.883.903.863.813.893.863.973.863.923.863.873.893.793.813.833.893.833.783.753.843.833.973.873.843.873.863.753.873.833.903.783.973.733.823.903.833.953.893.833.993.80SPC的基本概念20/127统计数据的分布SPC的基本概念21/127直方图SPC的基本概念正态分布SPC的基本概念23/127Sigma与正态分布概率关系Sigma&Probabilityσ范围合格率缺陷率±1σ68.27%31.73%±2σ95.45%4.55%±3σ99.73%0.27%±4σ99.99%63PPM±5σ99.99994%0.57PPM±6σ99.9999998%0.002PPMSPC的基本概念24/127工序能力SPC的基本概念25/127射击/打靶的游戏shotgameSPC的基本概念26/127精确度Precision准确度Accuracy射击能力的分析ShotCapabilityAnalysisSPC的基本概念?CL精确度准确度3LCLUCL43124规格中心21SPC的基本概念过程工序能力的定义准确度(Ca)---CapabilityofAccuracyX-Target(USL-LSL)/2Ca=---Ca反映制程中心与规格中心偏离的状况。Target目标值LSL下规格线USL上规格线XMean平均值准确度(Ca)---CapabilityofAccuracyX-Target(USL-LSL)/2Ca=X-Target(USL-LSL)/2Ca=---Ca反映制程中心与规格中心偏离的状况。Target目标值LSL下规格线USL上规格线XMean平均值Target目标值LSL下规格线USL上规格线XXMean平均值6CpLSLUSL-=2dR=—Cp反映制程的潜在能力指数。精确度(Cp)---CapabilityofPrecisionLSLUSL6X过程工序能力的定义30/1272dR=USLTargetLSLX—Cpk反映制程的实际能力指数。Cpk=(1-|Ca|)*Cp=Min(Cpu,Cpl)CMinUSLXorXLSLpk=--[()()]3CpuCpl过程工序能力的定义31/127练习部件规格:USL=6.0LSL=4.0标准值=5.0sigma=0.25X=5.1(USL-LSL)/(6sigma)=(6.0-4.0)/(6*0.25)=1.45Min{(USL-X)or(X-LSL)/(3sigma)=Min{(6.0-5.1)or(5.1-4.0)}/(3*0.25)=1.30Cp=Cpk=工序能力与不合格率的关系工序能力不合格品率P工序能力判断2≥CP1.67P0.6PPM过剩1.67≥CP1.330.6PPM≤P≤60PPM充分1.33≥CP160PPM≤P≤0.27%正常1≥CP0.670.27%≤P≤4.45%不足CP≤0.67P≥4.45%严重不足CpSpec.不合格率1.00(+-)3sigma0.27%1.33(+-)4sigma63PPM2.00(+-)6sigma2.0PPB•如果工序能力不足或不能接受:1.减小sigma(变异)2.调整均值3.修正设计规范(客户标准)问题34/127时间Time测量值Measurement控制中心CL控制上线UCL(x+3S)控制下线LCL(x-3S)控制图35/127预防与检测过程PROCESS原料Material人people机Machine法MethodsEnvironment.环Measure测量测量Measuring结果result好pass不好fail不要等产品做出来后再去看它好不好而是在制造的时候就保证把每一个过程做好36/127制程process人People机Equipment物Material法Methods环Environment产品Product或Or服务Service顾客Customers客户变化的需求和期望值IdentifyingChangingNeedsandExpectations客户声音Voiceofcustomer统计方法StatisticalMethods过程声音Voiceoftheprocess输入Input过程/系统Process/System输出Output过程控制系统的反馈模型37/127控制图的原理68.26%95.45%99.73%μ+1σ+2σ+3σ-1σ-2σ-3σUCLLCLUSLLSL38/127x3+UCLLCL3-USLLSL控制图的原理39/127建立控制图的步骤A收集数据B计算控制限D过程控制解释E过程能力解释C绘制控制图40/127RX-控制图A收集数据B计算控制限D过程控制解释E过程能力解释C绘制控制图样品测量S1S2S3S4S5115.97615.94615.95415.90615.934215.91715.92415.91515.94715.906315.95415.94515.95715.90615.967415.90615.93615.92415.92515.947515.92515.90515.95415.95615.967615.93715.93715.90615.91515.924715.96515.94315.94815.89515.957815.89615.93815.92715.91515.934915.97515.94715.93415.94315.9041015.92415.92415.91715.94515.9071115.93715.95815.90415.96515.9461215.93515.91815.94515.91415.9261315.95515.92515.96815.90715.9451415.90515.93415.93715.92515.9251515.95415.95715.93615.96415.9051615.93515.93515.92715.93415.9031715.95515.90715.95415.97515.9241815.93515.91515.94615.92415.9171915.95615.94715.89515.94715.9642015.91515.92815.91415.93515.957工序批样本数据总体抽样测量4