关于误差规律性与影响特性的新概念Thenewconceptofmeasurementerrorsre

ContentslistsavailableatScienceDirectMeasurementjournalhomepage:’sregularitiesandeffectcharacteristicsYeXiaominga,b,⁎,LiuHaiboc,d,XiaoXuebine,LingMocaSchoolofGeodesyandGeomatics,WuhanUniversity,Wuhan,Hubei430079,ChinabKeyLaboratoryofPrecisionEngineering&IndustrySurveying,StateBureauofSurveyingandMapping,Wuhan,Hubei430079,ChinacInstituteofSeismology,ChinaEarthquakeAdministration,Wuhan,Hubei430071,ChinadWuhanInstituteofEarthquakeMetrologicalVerificationandMeasurementEngineering,Wuhan,Hubei430071,ChinaeWuhanUniversityLibrary,Wuhan,Hubei430079,ChinaARTICLEINFOKeywords:MeasurementerrorFunctionmodelRandommodelError’sregularitiesUncertaintyABSTRACTInseveralliteratures,theauthorsgiveanewthinkingofmeasurementtheorysystembasedonerrornon-clas-sificationphilosophy,whichcompletelyoverthrowstheexistingmeasurementconceptsystemofprecision,truenessandaccuracy.Inthispaper,byfocusingontheissuesoferror’sregularitiesandeffectcharacteristics,theauthorswilldoathematicinterpretation,andprovethattheerror’sregularitiesactuallycomefromdifferentcognitiveperspectives,arealsounabletobeusedforclassifyingerrors,andthattheerror’seffectcharacteristicsactuallydependonartificialconditionrulesofrepeatedmeasurement,andarestillunabletobeusedforclas-sifyingerrors.Thus,fromtheperspectivesoferror’sregularitiesandeffectcharacteristics,theexistingerrorclassificationphilosophyisstillincorrect;andanuncertaintyconceptsystem,whichmustbeinterpretedbytheerrornon-classificationphilosophy,naturallybecomestheonlywayoutofmeasurementtheory.1.IntroductionInseveralliteratures[1–3],theauthorsgiveanewthinkingofmeasurementtheorysystembasedonerrornon-classificationphilo-sophy.Themainlogicofthisthinkingisbrieflyintroducedasfollows:Theconceptoferrorisdefinedasthedifferencebetweenthemea-surementresultanditstruevalue.Becausethemeasurementresultisunique,andthetruevalueisalsounique,sotheerrorofthemea-surementresultistheonlyunknownandconstantdeviation.Forafinalmeasurementresult,theconstantdeviationconsistsoftwoparts:1,thedeviationΔAbetweenthefinalmeasurementresultandmathematicalexpectation,whichistheso-calledrandomerrorinex-istingtheory;2,thedeviationΔBbetweenmathematicalexpectationandtruevalue,whichistheso-calledsystematicerrorinexistingtheory.Becausebothdeviationsareunknownandpersistconstantde-viations,anddonothaveanydifferenceincharacteristics,therefore,havingnocharacteristicdifferencemustnotcauseanyclassificationdifference!ThestandarddeviationofdeviationΔAisgivenbythestatisticandanalysisofcurrentmeasurementdata.ThedeviationΔBisalsopro-ducedbymeasurement;itsformationprincipleisactuallythesameasthecurrentmeasurement;itsstandarddeviationcanbeobtainedbytracingbacktoitsupstreammeasurement.Thus,thestandarddeviationoftotalerroroffinalmeasurementresultisequaltothesynthesisofthetwostandarddeviationsaccordingtotheprobabilitylaws.Thistotalstandarddeviationisuncertainty,whichistheevaluationoftheprob-ableintervaloftheerroroffinalmeasurementresult(thisgiveamoreclearmeaningtotheuncertaintyconcept).Thisconstantdeviationtheoryiscompletelyoppositetotherandomvariationtheoryofexistingmeasurementtheory,thatis,intheopinionoftheauthors,itisobviouslyillogicalthatexistingmeasurementtheoryinterpretdeviationΔAasprecisionbutinterpretdeviationΔBastrue-ness,andtheerrorclassificationdefinitionandalltheconceptsofprecision,truenessandaccuracyshouldbeabolished.Forexample:in2005,theChinesesurveyingandMappingBureaugavethattheelevationresultofMountEverestis8844.43mwithstandarddeviationof±0.21m.Accordingtoexistingerrorclassifica-tiontheory,fromtheperspectiveoferror’sdefinition,theerrorofthisresultisasingleconstantdeviationandshouldbeclassifiedassys-tematicerror;however,fromtheperspectiveofstandarddeviation±0.21m,itshouldbeclassifiedasrandomerror.Thisisthelogicaltrapofexistingerrorclassificationtheory.Andtheinterpretation,accordingtoerrornon-classificationtheory,isthatthisresult'serror(thediffer-encebetweentheresultandthetruevalueatimplementingmeasure-ment)isanunknownconstant,andthatthestandarddeviationof±0.21misonlytheevaluationoftheprobableintervaloftheunknown⁎Correspondingauthorat:SchoolofGeodesyandGeomatics,WuhanUniversity,Wuhan,Hubei430079,China.E-mailaddress:[e-mail](H.Liu).Measurement126(2018)65–71Availableonline14May20180263-2241/©2018ElsevierLtd.Allrightsreserved.Tconstanterror.That'sit.ThedifferencebetweenthenewtheoryandtheexistingtheoryisshowninFig.1.Pleasenotethattheauthors’emphasizeontheconceptofconstantdeviationistofocusonthemeasurementresultinsteadoftheoriginalobservationvaluesbeforeformingfinalmeasurementresult.Ofcourse,theauthorsrecognizethattheremaybeindeedadiscreteerrorsamplesequencebeforethemeasurementresultisformed.However,thesediscreteerrorsamples,whichhavecertainnumericalvalue,arethemeasuredvaluesoferrors.Theybelongtothemeasurementresult,andnaturallycannotbemixedwiththeunknownerroroffinalresulttodiscusstheerrorclassification.Inad