SPC-PRESENTATION-PRIMARY1[1]

1STATISTICALPROCESSCONTROL(SPC)PartI─FundamentalsR.S.Chen统计过程控制(基础部分)2WhatIsAProcessComposedOf?什么叫过程？方法物料机器人环境测量输出3PROCESSBEHAVIOR过程状态Precise,notAccurate精密，不准确1515Accuracy—Theclosenessoftheagreementbetweentheresultsofmeasurementandthetruevalue准确度—测得值与真值或设定值之差Precision—Thedegreeoftheagreementbetweenrepeatedmeasurementsofthesametestmaterial精密度—同一样品多次测试的重复性41515PROCESSBEHAVIOR过程状态Accurate,notPrecise准确，不精密51515PROCESSBEHAVIOR过程状态AccurateandPrecise准确精密6WhichProcessesWouldYouChoose?你会选择哪两个过程？15151515Precise,notAccurate精密，不准确Accurate,notPrecise准确，不精密AccurateandPrecise准确精密7PrecisionAndAccuracy过程的精密度和准确度Precise,notAccurate精密，不准确Accurate,notPrecise准确，不精密AccurateandPrecise准确精密8ADistributionCurveOfAConsistentProcess正态分布―连续过程的概率分布e21222)(x(-∞x∞)f(x)=对于连续过程而言，若一随机变量由众多偶然因素(commoncauses)决定，且其中无主导因素，则该变量通常均服从正态分布规律．9AConsistentProcess―NormalDistribution正态分布•Abell-shapedsymmetricalcurve对称分布曲线•Canbedescribedbytheprocessaverageandstandarddeviation过程性状可用过程均值和标准偏差表征•Canpredictwherethedatawillfallunderthecurve可预测过程数据将落在曲线包含的范围内标准偏差过程均值10PredictWhereDataWillFallUnderTheCurve过程数据的正态分布=过程均值=过程标准偏差(s=过程样本标准偏差）N=过程个体的数量(n=过程样本个体的数量)11StatisticallyControlAProcess用统计技术对过程进行控制受控失控可预测不可预测12ThePurposeConductingSPC统计过程控制的目的Toachievequalityexcellence实现质量卓越Operatingontargetwithminimumvariability按设定目标以最小的变异度运行SPCisaproactiveapproachtomanagequalitySPC是前瞻性的质量管理手段13ControlChart统计过程控制的工具―控制图控制上限控制下限控制下限控制上限测量值移动极差时间时间均值平均移动极差14Example:ARunChartOfACarbonationProcess例：一碳酸化过程(含汽量)的单值图含汽量样品序号15Wheretheprocessistargeted：Average-Thesumoftheobservationsdividedbythenumberofobservations平均值Measuresofvariationordispersion-HowvariabletheoutputoftheprocessisStandardDeviation―Anumericalvaluethatmeasuresthespreadingorvariationofthedata离散性－标准差－精密度Twoprincipalmeasures两个关键值16ProcessCapabilityStudy―ToAcquireTheTwoMeasures过程能力分析的目的─获取两个关键值Theprocessaverageandthedispersionoftheprocessarenecessaryelementstomakeupcontrolcharts.Processcapabilitystudyisthewayforacquiringthetwomeasures.过程均值及离散度是控制图的必要参数，过程能力分析是获取这二个数据的方法。理想理想实际（假设目标是100）频率频率17PrerequisitesToPerformingAProcessCapabilityStudy过程能力分析的前提条件Precision(σ)isthekeytoprocesscapability.Inordertoseparateprocessvariabilityfromallothers,itisnecessarytostabilizetheprocessandkeepalltheprocesselementsinthe“standardized”operationalstatus.☺keepingoneanalystwithonesetofdataanalysis;☺sufficientlyaccurateandprecisetestingequipmentthatisincalibrationstatus;☺sameshiftandsamecontainerofmaterials;☺allmachinesingoodmaintenancestatus,….精密度是过程能力的关键,为了排除所有其它变化对过程精密度的影响，需要将过程要素稳定在“标准”的运行状态．☺由一位分析员对一套数据进行测试;☺测试仪器符合准确度和精密度的要求，且处于校正状态；☺同一班次，同一缸物料;☺所有设备均处于良好的维护保养状态，…．18Step-By-StepApproachToPerformingAProcessCapabilityStudy过程能力分析的步骤(例：制造)1.Identifyaproductionrunofsufficientlengthtorepresenttheprocessstudied.确定一个代表被分析过程的生产时间足够长的批次．2.Dividetheruntimeby30forrandomlycapturingatleast30samplesbyusingarandomnumbertable.Theactualsamplingintervalwillbesometimesmoreandsometimeslessthantheaverageaccordingtotherandomnumbertable.以该批次的生产时间除以至少30个样品，得采样的平均时间间隔．实际采样间隔按随机数表决定，有些会大于平均间隔．有些则会小于平均间隔．例：生产时间约8小时，则平均采样间隔约为：480分锺÷30=16分锺19ARandomNumberTable20Step-By-StepApproachToPerformingAProcessCapabilityStudy过程能力分析的步骤(例：制造)3.Assignasingletechniciantoconfirmandmonitortoensurethatproperstart-upandrunprocedureshavebeenused.指定一位主管对过程进行监控，以确保开车及运行程序正确恰当。4.Designadatacollectingsheetthatisalsotherecordtobemaintainedtogetherwiththedottedcontrolcharts.设计一份数据表记录测试结果及采样相关信息。5.Beginsamplingwhenthetechnicianandoperatorsarethroughwiththestart-upoftheproductionandtheprocessisrunningsmoothly.当主管及操作人员顺利启动过程，且过程运行平稳后开始采样。6.Testandrecordthetestresultsintheorderthatsamplewascollected.按采样顺序测试样品和记录测试结果．21例：数据记录表(Brix)日期时间数据目标记录22Step-By-StepApproachToPerformingAProcessCapabilityStudy过程能力分析的步骤(例：制造)7.Analyzeandcalculatethedata分析和计算数据7.1分析数据InformationFromHistogram直方图给出的信息Constructahistogramtodeterminethetypeandshapeofthedistribution．绘制频率直方图，以确定数据分布的类型和形状Thefrequencyhistogramrevealsthestatisticalpatternofthedataset,itgraphicallyshowsthecentrallocationandspreadoftheindividualvalues.频率直方图呈显该组数据的统计学性质，其中心位置，及数值的离散性．23InformationFromHistogram直方图Ahistogramcangraphicallyshowthreethings:位置：Avisualindicationofthecentraltendencyorlocationofthedata分布形状：Avisualindicationoftheshapeordistributionofthedata离散性：AvisualindicationofthespreadorvariationofthedataClassIntervalWidth(w)StartValue(CS)ClassInterval#n分组上限值UpperBoundary(CB#n)HistogramClassIntervalWidthandBoundaries组距起始值24AGuideforpreparingfrequencyhistogram怎样绘制频率直方图♦1.Numberofinterval(nc):Squarerootofsamplesize(n)—A“ruleofthumb”♦2.Calculatesamplerange(R):R=Highvalue-Lowvalue♦3.Calculateclassintervalwidth(w):w=Rnc♦4.Determinemeasurementaccuracy:=Measuringresolution1.区间数(nc):样品总数的平方根2.样本范围(R):R=最高值-最低值3.区间宽度(组距)(w):w=R/nc4.计量精度(ε):ε=分辨率Step-By-StepApproachToPerformingAProcessCapabilityStudy过程能力分析的步骤(例：制造)25Step-By-StepApproachToPerformingAProcessCapabilityStudy过程能力分析的步骤(例：制造)HowAFrequencyHistogramIsConstructed怎样绘制频率直方图5.StartvalueforthefirstClassinterval:StartValue(CS)=Lowestvalue(2)6.Calculateupperboundaryforeachclassinte