外文翻译基于优化的牛顿拉夫逊法和牛顿法的潮流计算.doc
《外文翻译基于优化的牛顿拉夫逊法和牛顿法的潮流计算.doc》由会员分享,可在线阅读,更多相关《外文翻译基于优化的牛顿拉夫逊法和牛顿法的潮流计算.doc(14页珍藏版)》请在冰点文库上搜索。
外文翻译--基于优化的牛顿—拉夫逊法和牛顿法的潮流计算
英文文献
PowerFlowCalculationbyCombinationofNewton-RaphsonMethodand
Newton'sMethodinOptimization.
AndreyPazderin,SergeyYuferev
URALSTATETECHNICALUNIVERSITY?
UPI
E-mail:
pav@//0>.,usv@//.
Abstract--Inthispaper,theapplicationoftheNewton'smethodin
optimizationforpowerflowcalculationisconsidered.Convergence
conditionsofthesuggestedmethodusinganexampleofathree-machine
systemareinvestigated.Itisshown,thatthemethodallowstocalculate
non-existentstatepointsandautomaticallypullsthemontotheboundary
ofpowerflowexistencedomain.Acombinedmethodwhichiscomposedof
Newton-RaphsonmethodandNewton'smethodinoptimizationispresented
inthepaper.
IndexTerms?
Newtonmethod,Hessianmatrix,convergenceofnumerical
methods,steadystatestability
Ⅰ.INTRODUCTION
problemisthebasisonwhichother
flowthepowerofThesolution
problemsofmanagingtheoperationanddevelopmentofelectricalpower
systemsEPSaresolved.Thecomplexityoftheproblemofpowerflow
calculationisattributedtononlinearityofsteady-stateequations
systemanditshighdimensionality,whichinvolvesiterativemethods.The
basicproblemofthepowerflowcalculationisthatofthesolution
feasibilityanditerativeprocessconvergence[1].
Thedesiretofindasolutionwhichwouldbeontheboundaryofthe
existencedomainwhenthegivennodalcapacitiesareoutsidetheexistence
domainofthesolution,anditisrequiredtopullthestatepointback
ontothefeasibilityboundary,motivatestodevelopmethodsand
algorithmsforpowerflowcalculation,providingreliableconvergenceto
thesolution.
ThealgorithmforthepowerflowcalculationbasedontheNewton's
methodinoptimizationallowstofindasolutionforthesituationwhen
initialdataareoutsidetheexistencedomainandtopulltheoperation
pointontothefeasibilityboundarybyanoptimalpath.Alsoitispossible
toestimateastaticstabilitymarginbyutilizingNewton'smethodin
optimization.
AsthealgorithmbasedontheNewton'smethodinoptimizationhas
considerablecomputationalcostandpowercontrolcannotberealizedin
allnodes,thealgorithmbasedonthecombinationoftheNewton-Raphson
utilized
betoofferedisoptimizationinmethodNewton'stheandhodsmet
forcalculatingspeed,enhancingthepowerflowcalculation.
II.THEORETICALBACKGROUND
A.Steady-stateequations
Thesystemofsteady-stateequations,ingeneral,canbeexpressed
asfollows:
whereisthevectorofparametersgivenforpowerflow
calculation.Inpowerflowcalculation,realandreactivepowersareset
ineachbusexceptfortheslackbus.Ingenerationbuses,themodulusof
voltagecanbefixed.WX,Yisthenonlinearvectorfunctionof
steady-stateequations.VariablesYdefinethequasi-constantparameters
associatedwithanequivalentcircuitofanelectricalnetwork.Xisa
requiredstatevector,itdefinessteadystateofEPS.Thedimensionof
thestatevectorcoincideswiththenumberofnonlinearequationsofthe
system1.Therearevariousknownformsofnotationofthesteady-state
equations.Normally,theyarenodal-voltageequationsintheformofpower
balanceorintheformofcurrentbalance.Complexquantitiesinthese
equationscanbepresentedinpolarorrectangularcoordinates,which
leadstoasufficientlylargevarietyformsofthesteady-stateequations
notation.Therearevariablemethodsofanonlinearsystemofsteady-state
equationssolution.Theyareunitedbytheincrementalvectorof
independentvariablesΔXbeingsearchedandtheconditionof
convergencebeingassessedateachiteration.
B.TheNewton'smethodinoptimization
Anotherwayofsolvingtheproblemofpowerflowcalculationis
relatedtodefiningazerominimumofobjectivefunctionofsquaressum
ofdiscrepanciesofsteady-state
equations:
2?
Thefunctionminimum2isreachedatthepointwherederivatives
onallrequiredvariablesareequaltozero:
3
Itisnecessarytosolveanonlinearsetofequations3tofindthe
solutionfortheproblem.Calculatingthepowerflow,whichismadeby
thesystemofthelinearequationswithaHessianmatrixateachiteration,
isreferredtoastheNewton's
methodinoptimization[4]:
4
TheHessianmatrixcontainstwoitems:
5
Duringthepowerflowcalculation,thedeterminantofHessian
matrixispositiveroundzeroandnegativevalueofadeterminantof
Jacobian.Thisallowstofindthestatepointduringthepowerflow
calculation,wheninitialpointhasbeenoutsideoftheexistencedomain.
TheconvergencedomainofthesolutionoftheNewton'soptimization
methodislimitedbyapositivevalueoftheHessianmatrixdeterminant.
Theiterativeprocessevenforasolvableoperatingpointcanconverge
toanincorrect
solutionifinitialapproximationhasbeenoutsideconvergence
domain.Thisallowstoestimateastaticstabilitymarginofthestate
andtofindthemostperilouspathofitsweighting.
III.INVESTIGATIONSONTHETESTSCHEME
ConvergenceoftheNewton'smethodinoptimizationwithafull
Hessianmatrixhasbeeninvestigated.Calculationsweremadebasedon
programMathCADforanetworkcomprisingthreebusestheparametersof
whicharepresentedinFigure1.Dependantvariableswereanglesofvectors
ofbusvoltage1and2,independentvariableswerecapacitiesinnodes
1and2,andabsolutevaluesofvoltagesofnodes1,2and3werefixed.
Fig.1?
TheTestscheme
InFigure2,theboundaryofexistencedomainforasolutionofthe
steady-stateispresentedinangularcoordinatesδ1-δ2.Thisboundary
conformstoapositivevalueoftheJacobiandeterminant:
AsaresultofthepowerflowcalculationbasedontheNewtonmethod
inoptimization,theanglevalueshavebeenreceived,thesevalues
correspondingtothegivencapacitiesinFig.2generationispositiveand
loadingisnegative.
Forthestatepointswhichareinsidetheexistencedomain,the
objectivefunction2hasbeenreducedtozero.Forthestatepointswhich
areontheboundaryoftheexistencedomain,objectivefunction2hasnot
from
differedcapacitiesofvaluescalculatedtheandzerotoreducedbeen
thegivencapacities.
Fig.2?
DomainofExistenceforaSolution
Fig.3-BoundaryofexistencedomainInFig.3,theboundaryofthe
existencedomainispresentedincoordinatesofcapacitiesP1-P2.State
pointsoccurringontheboundaryoftheexistencedomain6havebeenset
bythecapacitieswhichwereoutsidetheexistencedomain.Asa
resultofpowerflowcalculationbyminimization2basedonthe
Newton'smethodinoptimization,theiterativeprocessconvergestothe
nearestboundarypoint.Itisduetothefactthatsurfacesoftheequal
levelofobjectivefunction2incoordinatesofnodalcapacitiesare
propercirclesforthreemachinesystemhavingthecentreonthepoint
definedbygivenvaluesofnodalcapacitiesThegraphicinterpretationof
surfacesoftheequallevelofobjectivefunctionforoperatingpoint
statewith13000MWloadingbus1and15000MWgeneratingbus2ispresented
inFig.3.
Hessianmatrixisremarkableinitsbeingnotsingularonthe
boundaryofexistencedomain.ThedeterminantofaHessianmatrix5is
positivearoundzeroandanegativevalueoftheJacobianmatrix
determinant.Thisfactallowsthepowerflowtobecalculatedevenfor
theunstablepointswhichareoutsideexistencedomain.Theiterative
processbasedonthesystemofthelinearequations4solutionhas
Naturally,
iteration.3-5withinpointstabilitycriticalthetoconverged
theiterativeprocessbasedonNewton-Rapsonmethodisdivergentforsuch
unsolvableoperatingpoints.
Theconvergencedomainofthemethodunderconsiderationhasbeen
investigated.Whatismeantisthatnotallunsolvableoperatingpoints
willbepulledontothe
boundaryofexistencedomain.Acertainthresholdhavingbeen
exceededtheiterativeprocesshasbeguntoconvergetotheimaginary
solutionwithanglesexceeding360Itisnecessarytonotethattoreceive
acriticalstabilityoperatingpointincasewheninitialnodalcapacities
aresetoutsidetheboundaryoftheexistencedomain,thereisnonecessity
tomakeanyadditionaltermsastheiterativeprocessconvergesnaturally
tothenearestboundarypoint.
Pullingtheoperationpointontofeasibilityboundaryisnotalways
possiblebytheshortestandoptimalpath.Thereareanumberof
constraints,suchasimpossibilityofloadconsumptionincreaseatbuses,
constraintsofgenerationshedding/gainingatstations.Loadfollowing
capabilityofgeneratorunitsisvarious,consequentlyforfasterpulling
theoperationpointontothefeasibilityboundaryitisnecessaryto
carryoutthispullingprobablybylonger,butfasterpath.
Thealgorithmprovidespossibilityofpathcorrectionofpulling.
Itiscarriedoutbyusingoftheweightingcoefficients,whichdefine
degreeofparticipationofeach
nodeintotalcontrolaction.ForthispurposediagonalmatrixAof
theweightingcoefficientsforeachnodeisincludedintotheobjective
function2:
AlldiagonalelementsoftheweightingcoefficientmatrixAshould
begreater-thanzero:
Wheninitialapproximationliesintothefeasibilitydomain,
coefficientsarenotinfluenceonthecomputationalprocessandonthe
result.
Inthefigure4differentpathsofthepullingthesameoperation
pointontofeasibilityboundarydependingontheweightingcoefficients
arepresented.Pathsarepresentedfortw