資源描述:
《Unbiased Statistical Comparison of Creep and Shrinkage.pdf》由會員上傳分享,免費在線閱讀,更多相關(guān)內(nèi)容在學術(shù)論文-天天文庫。
1、ACIMATERIALSJOURNALTECHNICALPAPERTitleno.105-M69UnbiasedStatisticalComparisonofCreepandShrinkagePredictionModelsbyZdenekP.Ba?antandGuang-HuaLiThispaperaddressestheproblemofselectingthemostrealisticcomparisonsofthemainpredictionmodelsforcreepandcreepandshrinkageprediction
2、model,importantfordesigningshrinkageofconcreteandexplainwhyvariousnonstandarddurableandsafeconcretestructures.Statisticalmethodsofstatisticalindicatorshaveledtodubiousconclusions.ThesamestandardandseveralnonstandardtypesandaverylargeexperimentalfivemodelsasinReference7wi
3、llbestatisticallyevaluated:databasehaverecentlybeenusedtocompareandranktheexisting1.ModelB3,1995,whichwasapprovedastheinternationalpredictionmodels,butconflictingresultshavebeenobtainedby2721RILEMRecommendationandslightlyupdatedin2000variousinvestigators.Thispaperattempt
4、stoovercomethis9confusion.Itintroducesdataweightingrequiredtoeliminatethe(thismodelisarefinementofthe1978ModelBPandofitsimprovementasModelBP-KX13);biasduetoimproperdatasamplinginthedatabase,andthen2.ACImodel,8basedon1960sresearch,reapprovedbyexaminesBa?antandBaweja’sMode
5、lB3,theACImodel,theCEBmodel,andtwoofGardner’smodels.ThestatisticsofpredictionACICommittee209in2008;errorsarebasedstrictlyonthemethodofleastsquares,whichisthe3.ModelofComitéEuropéenduBéton,labeledCEB,standardandtheonlystatisticallycorrectmethod,dictatedbythewhichisbasedon
6、theworkofMüllerandHilsdorf3(itwasmaximumlikelihoodcriterionandthecentrallimittheoremoftheadoptedin1990byCEB,12updatedin1999,28andco-optedtheoryofprobability,aswellastherequirementofnoncorrelationin2002forEurocode2);oferrors.Severalnonstandardstatisticalmethodsthathave204
7、.GardnerandLockman’smodel,labeledGL;andrecentlybeeninventedtoevaluatecreepandshrinkagemodelsare19alsoexaminedandtheirdeficienciesarepointedout.Theranking5.Gardner’searliermodel,labeledGZ.Sakata’smodel,18,22whosescopeissomewhatlimited,ofthemodelsthatensuesfromtheleast-squ
8、areregressionstatisticsisshowntobequitedifferentfromtherankingsobtainedbytheaswellasthecrudeoldmodelsof