插电式HEV对于大气影响
expensiveanalysisorsimulationprocess;metamodelingreferstothetechniquesandprocedurestoconstructsuchamodel[39].Inouroptimizationframework,asetofpolynomialmetamodelswasusedtoexpresstheobjectivefunctionandtheconstraints.Inpar-ticular,fueleconomy,GHGemissions,and0–30mphand0–60mphaccelerationtimeswereevaluatedthroughsimulationinAutonomieoveragridofvaluesformotor/generatorandbatterysizes.Thenmultivariatepolynomialfunctionswere ttothedatausingleastsquares.
3.1RegressionModel.Theleastsquaresmethodisafunda-mentalapproachforparameterestimation.Ifthemodelhasthepropertyofbeinglinearintheparametersthentheleastsquaresestimatecanbecalculatedanalytically[40].Weassumethatthemodel,wewishtoidentifyisintheform
^ðiÞ¼u1ðiÞÁa1þu2ðiÞÁa2þÁÁÁþunðiÞÁany
(1)
n
1X1
eðiÞ2¼kEk2Rða;nÞ¼
2i¼12
(7)
Ourobjectiveistoderivethevectorofthemodelparametersa
thatmakestheerrorbeequaltozero,thatis
^¼YÀUÁa¼0E¼YÀY
T
(8)
whereUðnÞ¼½uTð1ÞuTð2Þ:::uTðnÞ .Consequently,the
solutionoftheleastsquaresproblemisgivenbysolvingEq.(8)
Y¼UÁa,UTÁY¼UTÁUÁa,
ðUTÁUÞÀ1ÁUTÁY¼ðUTÁUÞÀ1ÁðUTÁUÞÁa,
a¼ðUTÁUÞÀ1ÁUTÁY
IfthematrixUTUisnonsingular,thenthesolutionofEq.(9)isauniqueminimumfortheleastsquaresproblem[40].
3.2OptimizationObjectiveFunctionandConstraints.Inouroptimizationproblemformulation,thevectorofthedesignvariablesxconsistsofthemotor/generatorsize,x1,andthebatterysize,x2.Thesetofpolynomialmetamodelsisafunctionofthevectorx.Acubic ttingfunctionofthefollowingformprovidesanappropriate ttingtothediscretesimulationdatapoints[39]forthePHEVperformanceindices(1)fueleconomy,ffueleconomy(mpg);(2)GHGemissions,fCO2(kg-CO2);(3)0–30mphaccelerationtime,t0À30(s);and(4)0–60mphaccelera-tiontime,t0À60(s)
32222
fðx1;x2Þ¼a1x31þa2x2þa3x1x2þa4x1x2þa5x1þa6x2
(9)
wherei¼1;2;:::;n;n2@indexesthenumberofsimulationdata
^istheoutputofthemodel;a1;a2;:::;anaretheparame-points;y
tersofthemodeltobedetermined;andu1;u2;:::;unareknownfunctionsthatmaydependonotherknownvariables.ThemodelinEq.(1)canbewritteninthevectorformasfollows:
^ðiÞ¼uTðiÞÁay
(2)
whereuTðiÞ¼½u1ðiÞu2ðiÞ:::unðiÞ anda¼½a1a2:::an T.The
modelinEq.(1)istheregressionmodel,andthefunctionsui;i¼1;2;:::;narecalledthe“regressionvariables.”Thesimula-tiondatapointsderivedfromAutonomiecorrespondtopairsofthemeasuredandregressionvariablesfðyðiÞ;uðiÞÞ;i¼1;2;:::;n;n2@g.Theproblemisformulatedsoastominimizethefollow-ingleastsquarescostfunctionwithrespecttotheparametersofthemodela1;a2;:::;an:
nnÂÃ21X1X2
^ðiÞ ¼Rða;nÞ¼½yðiÞÀyyðiÞÀuTðiÞÁa
2i¼12i¼1
(3)
þa7x1x2þa8x1þa9x2þa10(10)
Themeasuredvariableyislinearinparametersai,andthecost
functionisquadratic.Consequentlytheproblemadmitsananalyt-^bethevectorofthemeasuredvariablesicalsolution.LetYandY
andoutputofthemodel.respectively
Y¼½yð1Þ;yð2Þ;…;yðnÞ
and
^¼½y^ð1Þ;y^ð2Þ;…;y^ðnÞ TY
(5)
T
Toidentifytheappropriateorderofpolynomialmetamodelthat
tsthediscretesimulationdatapointswell,thenormofresidualsgivenbythefollowingwasused
krk¼
nXi¼1
(4)
!1=2
^ðiÞÞ2ðyðiÞÀy
(11)
andletEbethevectoroftheerroreðiÞbetweenthemeasured
variableandoutputofthemodel
E¼½eð1Þ;eð2Þ;…;eðnÞ T
(6)
^ðiÞ¼yðiÞÀuTðiÞÁa.SubstitutingEq.(6)inwhereeðiÞ¼yðiÞÀy
Eq.(3),thecostfunctioncanbewrittenas
ThenormofresidualsforeachmodelcorrespondingtothePHEVperformanceindiceswereplottedagainsttheorderofthepolynomialmetamodel,asshowninFigs.2–5.Inallcases,athirdorderpolynomialyieldsthesmallestvalueofthenormofresid-uals,andahigherorderdoesnotseemappropriateto tthesimu-lationdata.
Thepolynomialcoef cientsoftheregressionmodelusedforeachoftheoutputvaluesto tasetofdiscretesimulationdatapointsoveragridofvalueswerederivedusingleastsquares.Thevectorofknownfunctions,uT,inEq.(10)is
uTðiÞ¼½x31ðiÞx32ðiÞx21ðiÞÁx2ðiÞx1ðiÞÁx22ðiÞ2
x21ðiÞx2
x1ðiÞÁx2ðiÞx1ðiÞx21 (12)
wherex1andx2arethedesignvariables.Therangeofvalues
forthemotor/generatorsize,x1,usedtoderivethesimulationdatapointsincontinuouspowerisx1¼f60;80;100;120gkW.Similarly,therangeofvaluesforthebatterysizeinnumberof041201-4/Vol.135,DECEMBER2013
modulesisx2¼f6;7;8;9;10g,eachofwhichincludes118cellsinseriesand1cellinparallel(Table2).Asaresult,thesimulationdatasetiscreatedoveragridof20differentinputsði:e:;i¼1;2;:::;20Þ.
TransactionsoftheASME
2010 Abby Types of new energy vehicles ? hybrid electric vehicle (HEV) ?...To develop this can reduce the emission of greenhouse gases and increase ...
Topic 1-Electric and Hybrid Vehicles-Introduction...in the greenhouse gases (CO2, methane, NOx, ... which could result in zero emission vehicles ...
Greenhouse gas emission and global warming concerns...Requirements for Plug-in Hybrid Vehicles 1 2 1....electric vehicles and FreedomCAR charge-depleting ...
plug-in hybrid electric vehicles (PHEVs) and ...emission characteristics of greenhouse gases (GHG) ...Greenhouse Gases, Regulated Emissions, and Energy ...
Electric vehicles(电动汽车)_高等教育_教育专区。...in the direction of efficient and low emission ...? reduce greenhouse gas emissions. ? Improve fuel...
plug-in hybrid, and battery electric vehicles. ...and greenhouse gases which are calculated over the...Cars with a CO2 emission above the threshold of...
and emission Disadvantage of Parallel Architecture ?...Less greenhouse gas Hybrid Pros – ? Fuel ... Hybrid Electric Vehicles,” SAE 1998 Useful ...
plug-in hybrid electric vehicles an intrinsically ... that rely on low greenhouse gas emission ...energy department of the Flemish institute for ...
Plug-in Hybrid Electric Vehicles / 30 Ethanol ...emission of 10 tons of CO2 over a vehicle’s ...and greenhouse gas pollutants and reduce our ...
potential to reduce greenhouse-gas emissions and utilize renewable energy ...electric vehicles and plug-in hybrid electric vehicles (Shiau et al., 2010...
热门文档