ibm预测性维护质量spcqews成立.pptx

IBMPredictiveMaintenanceQuality:SPCQEWSInception

SPC-StatisticalProcessControl

计量型数据控制图Parametric(Variable)Data：Xbar-R图(均值-极差图)：应用于样本个数较少，多组间数据控制Xbar-S图(均值-标准差图)：应用于样本个数较多，多组间数据控制X-MR图(单值-移动极差图)：应用于单组数据控制Xbar-R图X-MR图

SPC-StatisticalProcessControl

计数型数据控制图AttributeData：P图(不合格品比率图)nP图(不合格品数图)c图(不合格数图)u图(单位不合格数图)P图u图

SPC的选择过程

QEWSInspectionObjective:Toidentifyaninstanceofagivenmanufacturingprocessdeviationleadingtoqualityimpactsascapturedbyapplyingstatisticalproceduresontheobservedfailurerateduringinspections,earlierthanmanyoftheconventionalquality/inspectionstatisticalprocesscontrolsystemslikep-chart.Features:AbsoluteFailureRateplotFailureRateEvidencePlot-CUSUMofthefailureratesaboveacceptablelimit(mean+/-giventolerance)ThresholdoftheCUSUMofthefailurerateasobtainedbysimulationusingcustomizedacceptancelimitsandprobabilityoffalsealarmsperpart.IdentificationofVintagesnotundercontrol(surpassingCUSUMthreshold)IdentificationofthemaximumCUSUMvintage

6QEWSParametric(Variable)DataFormatExampleofasinglesetofdata:numberofrowsmustbe

between50and1000

7QEWSDemonstrationAttributeDataFormatExampleofasinglesetofdata:numberofrowsmustbe

between50and100

InputdataforInspection

PMQneedsmasterdatasuchasProduct(Part),ProductParametersandProductionBatchtoprocessinspectiondataBatchModeDataneedtobeprovidedinaggregated(atdaylevel)csvformatperproductHistoricaldataneedtobeprovidedQEWSalgorithmwillbeprocessedforgiveninputdataandQEWSresultswillbecalculatedontheprovideddatasetRealTimeModeDataneedtofedinformofPMQeventsDataneedtobeprovidedforeventtypePRODUCTIONandINSPECTIONQEWSAlgorithmneedsfollowingproductparametersforInspectionLAM0LAM1PROB0INSPECT_NO_DAYSProductKPIandProfiletableswillbeupdatedwithQEWSgeneratedinspectionresults

QEWS–InspectionChart5/22/