00hybrid Controller

HIERARCHICALCONTROLOFHYBRIDPOWERSYSTEMS
By
MaraElenaTorresHernndez
Athesissubmittedinpartialfulfillmentoftherequirementsforthedegreeof
MASTEROFSCIENCE
in
ELECTRICALENGINEERING
UNIVERSITYOFPUERTORICO
MAYAGEZCAMPUS
2007
Approvedby:
_______________________________________________
EfrainONeill,Ph.D.Date
Member,GraduateCommittee
_______________________________________________
EduardoOrtiz,Ph.D.Date
_______________________________________________
CarlosCuadros,Ph.D.Date
_______________________________________________
MiguelVlezReyes,Ph.D.Date
President,GraduateCommittee
_______________________________________________
JoseColucci,Ph.D.Date
RepresentativeofGraduateStudies
________________________________________________
IsidoroCouvertierDate
Chairpersonofthedepartment
II
Abstract
In this thesis, we present a Hierarchical control of Hybrid Power Systems (HPS) that
consistofaWindTurbine,PhotovoltaicPanels,connectiontotheACutility,aBatteryBank,
and a zonal model of a load. The hierarchical control system consists of lower level
controllers for each generation and storage component developed using Sliding Mode
Control (SMC), and Model Predictive Control (MPC). The supervisory controller objectiveis
to supply the energy demand while maximizing usage of renewable sources, minimizing
connectiontothegrid,andeffectiveuseoftheavailablebatterystorage.Theperformance
of the system is studied by means of simulations. The thesis presents HPS model
development,controldesignandsimulationresultsunderdifferentoperatingscenarios.The
simulation results of the proposed hierarchical control shows a good performance of the
system.

III
Resumen
Enestatesis,sepresentauncontroljerrquicoparaunSistemaHbridodePotencia
(SHP) que consiste en una turbina de viento, un panel solar, conexin a la utilidad AC, un
banco de bateras y un modelo zonal de la carga. El nivel inferior del sistema de control
jerrquico est constituido por controladores con el objetivo de controlar cada sistema de
generacin de energa y de los componentes de almacenamiento; las tcnicas que se
utilizaron en este nivel de la jerarqua son control por deslizamiento y control predictivo
basadoenmodelos.Enelnivelsuperiorseencuentraelcontrolsupervisorcuyoobjetivoes
suplir la energa demandada por la carga mientras maximiza el uso de las fuentes
renovables, minimiza el uso de la utilidad AC y maneja de una forma efectiva el
almacenamiento de la energa de la batera. El comportamiento de todo el sistema es
estudiado a travs de simulaciones. La tesis presenta el modelo del SHP, el diseo del
control y los resultados de simulacin bajo diferentes escenarios de operacin. Los
resultadosobtenidosdelsistemapropuestomuestranunbuencomportamientodelsistema.

IV
Acknowledgements
I would like to thank my Advisor, Professor Miguel Vlez Reyes for all his support,
direction, and patience. Thanks for letting me work under your supervision. I will always
admireyouasaprofessor,asaprofessionalandasanexcellentperson.
IwouldliketothankDr.EfrainONeill,Dr.EduardoOrtiz,andDr.CarlosCuadrosfor
servingasmembersofmygraduatecommittee.
Thankstomyfamily,LuisandmyfriendsforbelieveinmeandgivingsupportwhenI
needed,especiallyindifficultsituations.ThankstotheCenterforPowerElectronicsSystems
(CPES); to the Electrical and Computer Engineering Department faculty and staff; and to all
of the people who directly or indirectly have helped through my graduate period because
thelittlethingsthatsomebodydoesonehadinfluencedmylife;sotoallofyouthanks.
Thankstootherauthorswhopermitmetopublishtheirfiguresonmythesis.
And the most important, I want to thank God for giving me the motivation and
strengthtofinishthisdream.
This work was primarily supported by the ERC Program of the National Science
FoundationundergrantEEC9731677.

V
TableofContents
Abstract......................................................................................................................................II
Resumen....................................................................................................................................III
Acknowledgements...................................................................................................................IV
Listoftables..............................................................................................................................IX
Listoffigures..............................................................................................................................X
1. Introduction........................................................................................................................1
1.1 Justification..................................................................................................................1
1.2 ResearchObjectives....................................................................................................2
1.3 SummaryofContributions..........................................................................................2
1.4 ThesisStructure...........................................................................................................3
2. LiteratureReview................................................................................................................4
2.1 Overview......................................................................................................................4
2.2 HybridPowerSystems(HPS).......................................................................................4
2.2.1 HybridPowerSystemsClassification...................................................................5
2.2.2 ElementsofHybridPowerSystems.....................................................................7
2.2.2.1 WindEnergyConversionSystem(WECS).....................................................7
2.2.2.2 PhotovoltaicPanel........................................................................................9
2.2.2.3 BatteryBank...............................................................................................10
2.3 HierarchicalControl...................................................................................................11
2.3.1 Controllevels......................................................................................................11
2.3.1.1 TheStrategicLevel......................................................................................12
2.3.1.2 TheTacticalLevel........................................................................................13
VI
2.3.1.3 TheOperationalLevel.................................................................................13
2.4 SlidingModeControl(SMC)......................................................................................13
2.4.1 BasicsConceptsofSMC.....................................................................................14
2.4.2 SMCinPowerSystems.......................................................................................15
2.4.2.1 DC/DCConvertersofaVariableStructuresystem.....................................15
2.4.2.2 SMCinDC/DCconverters...........................................................................16
2.5 ModelPredictiveControl(MPC)...............................................................................17
2.5.1 MPCtechnologies..............................................................................................18
2.5.2 MPCinPowerSystems.......................................................................................23
2.6 EnergyManagement(EM)inPowerSystems...........................................................24
2.6.1 EnergyManagementinHybridPowerSystems.................................................27
3. HPSSimulationModel......................................................................................................29
3.1 HybridPowerSystem................................................................................................29
3.2 WindTurbine.............................................................................................................32
3.3 PhotovoltaicPanel.....................................................................................................35
3.4 BatteryBank..............................................................................................................38
4. HierarchicalControllerforaHPS......................................................................................41
4.1 IndividualControlUnits.............................................................................................41
4.1.1 SMCforaWindTurbine.....................................................................................42
4.1.1.1 SlidingModecontrollerDesign..................................................................42
4.1.1.2 ModesofOperation....................................................................................43
4.1.1.3 SimulationResults......................................................................................46
4.1.2 PowerControlofaPhotovoltaicArrayusingSMC............................................52
4.1.2.1 SlidingModeControllerDesign..................................................................52
VII
4.1.2.3 ThecontrolLaw..........................................................................................55
4.1.3 ControlStrategyfortheBatteryBankandtheGrid..........................................57
4.1.3.1 ControllerDesign........................................................................................59
4.1.4 OnOffControlfortheLoad..............................................................................64
4.1.4.1 Thecontrollaw...........................................................................................64
4.2 SupervisorControlStrategy......................................................................................66
4.2.1 ModesofGeneration.........................................................................................69
4.2.1.1 Supervisorycontroller:Mode1..................................................................69
4.2.1.2 SupervisoryController:Mode2.................................................................70
4.2.2 OperationStrategy.............................................................................................72
4.2.3 SimulationResults..............................................................................................74
4.2.3.1 BatteryBankFullyCharged.........................................................................75
4.2.3.2 BatteryBankTotallyDischarged.................................................................80
4.2.3.3 Sufficientpowergeneration.......................................................................85
4.2.3.4 ZonalEPDS..................................................................................................89
4.2.3.5 Changingsourcepriorityonsupervisorycontroller...................................95
VIII
4.2.3.6 Changingsourcepriorityonsupervisorycontroller(4PVpanels).............99
4.3 Conclusions..............................................................................................................103
5. ControlforaHPSusingModelPredictiveControl.........................................................104
5.1 DynamicMatrixControl(DMC)...............................................................................104
5.2 PVControllerDesignusingDMC.............................................................................106
5.2.1 SimulationResults............................................................................................108
5.2.2 MPCPVagainstSMCPV..................................................................................110
5.3 WindControllerusingDMC.....................................................................................111
5.4 SupervisoryController.............................................................................................111
5.4.1 SupervisorControlwithbatterybankfullycharged........................................112
5.4.2 SupervisorControlwithBatteryBankTotallyDischarged...............................115
5.5 Conclusions..............................................................................................................118
6. ConclusionsandFutureWork.........................................................................................119
6.1 Summaryofthework..............................................................................................119
6.2 Conclusions..............................................................................................................119
6.3 FutureWork............................................................................................................121
References..............................................................................................................................123
Appendices.............................................................................................................................128

IX
Listoftables
Table1BasicsConfigurationonHybridPowerSystems...........................................................6
Table2WindturbineandPMSGparameters.........................................................................35
Table3PhotovoltaicModuleSpecifications............................................................................37
Table4ParametersoftheBatteryBank.................................................................................40

X
Listoffigures
Figure1structureofatypicalWindEnergySystem..................................................................8
Figure2StructureofatypicalsolarenergysystemconnectedtoaDCbus[50]......................9
Figure3ElectriccharacteristicforaPVcell............................................................................10
Figure4Multilevelcontrolofasystem...................................................................................12
Figure5SlidingModeIdea[23]...............................................................................................15
Figure6DC/DCConverter.......................................................................................................17
Figure7BasicstructureofMPC[10].......................................................................................20
Figure8Controllerstateatthek
th
samplinginstantadaptedfrom[39]................................21
Figure9Twoareapowernetwork..........................................................................................24
Figure10GenericPower/Energymanagementanddistributionsystemadaptedfrom[45].26
Figure11PrinciplesofEnergyManagement..........................................................................26
Figure12ElectricVehiclePowerFlow....................................................................................28
Figure13HybridgenerationSystem.......................................................................................30
Figure14Matlab/SimulinkSimulationmodeloftheHPS......................................................31
Figure15Simulinkmodelofthewindsubsystem...................................................................32
Figure16SimulinkmodelofWindTurbine.............................................................................33
Figure17SimulinkmodelofPermanentMagnetSynchronousGenerator............................34
Figure18Simulinkmodelofthesolarpanel..........................................................................36
Figure19CurrentVoltagecurveofPVarray..........................................................................37
Figure20BatteryModeladaptedfrom[53]............................................................................38
Figure21SimulinkmodelofLiIonBattery..............................................................................39
XI
Figure22OperationPointsofbothslidingsurfaces[54].......................................................44
Figure23Blockdiagramofcontrolledwindsubsystem.........................................................45
Figure24WindSpeed..............................................................................................................47
Figure25PowerCoefficientofwindturbine...........................................................................47
Figure26PowerReferenceofwindsubsystem.a)Mode1b)Mode2..................................48
Figure27AngularShaftSpeed................................................................................................48
Figure28Slidingsurfacesofwindturbine.a)1
st
modeb)2
nd
modeofgeneration..............49
Figure29Controlsignalsofwindsubsystem...........................................................................50
Figure30DCbusCurrents(Windsubsystem).........................................................................51
Figure 31 Zones of operation for control modes under sufficient power generation
conditions[15].........................................................................................................................54
Figure32Powersignalstodecidemodeofoperationofsolarsubsystem............................55
Figure33Slidingsurfacesofsolarsubsystem..........................................................................56
Figure34Controlsignalsofsolarsubsystem...........................................................................57
Figure35DCbuscurrents........................................................................................................58
Figure36Implementationofslopeonstrategyadaptedfrom[2].........................................61
Figure37BatteryBankandGridcurrentswhenthebatteryisdischarged............................63
Figure38BatteryBankandGridcurrentswhenthebatterybankisfullycharged................64
Figure39Controlsignalforloadzones....................................................................................66
Figure40GridandLoadcurrents.............................................................................................67
Figure41Supervisorycontrollerinteractionmodel...............................................................68
Figure42Schemeforinputsandoutputsinthesupervisorcontrol......................................69
Figure43Modetransitioncriteriaforthesupervisorcontrol.................................................72
Figure44Loadsheddingscheme............................................................................................73
XII
Figure45Matlab/SimulinkModelofsupervisorycontroller...................................................73
Figure46a)Windspeedandb)Celltemperaturevariations.................................................74
Figure47BatteryVoltageandSOC(BatteryBankFullyCharged)..........................................75
Figure48Loadcurrentdividedbyzones(BatteryBankFullyCharged).................................76
Figure49ControlsignalsofWindTurbineandPVarray(BatteryBankFullyCharged).........76
Figure50ControlsignalsandslidingsurfacesofPVarray.....................................................77
Figure51ControlsignalsandslidingsurfacesofWindTurbine.............................................77
Figure52DCbuscurrentsandoperationmodes(BatteryBankFullyCharged).....................79
Figure53BatteryvoltageandSOC(BatteryBankdischarged)................................................80
Figure54Loadcurrentdividedbyzones(BatteryBankDischarged)......................................81
Figure55ControlsignalsofWindTurbineandPVarray(BatteryBankDischarged).............81
Figure56Controlsignalandslidingsurfacesofsolarsubsystem...........................................82
Figure57Controlsignalandslidingsurfacesofwindsubsystem...........................................82
Figure58DCbuscurrentsandoperationmodes(BatteryBankDischarged).........................84
Figure59SOCofBatteryBank(Sufficientpowerregime).......................................................85
Figure60DCbuscurrentofHPS(sufficientpowergeneration)..............................................86
Figure61PowerreferenceandfictitiouspowerofPVarray(sufficientgenerationpower)..87
Figure62Windsubsystemdecisioncriteria(sufficientpowerregime)..................................87
Figure63PVarraycontrolsignals(sufficientpowergeneration)...........................................88
Figure64PVarrayslidingsurfaces(sufficientpowerregime)................................................88
Figure65Renewableenergycontrolsignals(sufficientpowerregime).................................89
Figure66VoltageandSOCofBatteryBank(a)totallydischarged,(b)fullycharged.............90
Figure 67 Control signals of wind and solar subsystems with the battery bank (a) totally
discharged,(b)fullycharged....................................................................................................90
XIII
Figure68Slidingsurfacesandangularspeedofwindsubsystemwithbatterybank(a)totally
discharged,(b)fullycharged....................................................................................................91
Figure69Controlsignalsofwindsubsystemwithbatterybank(a)totallydischarged,(b)fully
charged.....................................................................................................................................92
Figure70Slidingsurfacessolarsubsystemwithbatterybank(a)totallydischarged,(b)fully
charged.....................................................................................................................................92
Figure71Controlsignalsofsolarsubsystemwithbatterybank(a)totallydischarged,(b)fully
charged.....................................................................................................................................93
Figure72DCbuscurrentofHPS.TheBatteryBankis(a)totallydischarged,(b)fullycharged.
..................................................................................................................................................94
Figure73Batterybanksignalsa)Voltageb)StateofCharge..................................................95
Figure74PVarraycontrolsignals...........................................................................................96
Figure75Windturbinecontrolsignals...................................................................................96
Figure76Renewablesourcescontrolsignals.........................................................................97
Figure77DCbuscurrent.........................................................................................................98
Figure78Batterybanksignalsa)Voltageb)StateofCharge..................................................99
Figure79PVarraycontrolsignals.........................................................................................100
Figure80Windturbinecontrolsignals.................................................................................100
Figure81Renewablesourcescontrolsignals.......................................................................101
Figure82DCbuscurrent.......................................................................................................102
Figure83Multiobjectivecontroller......................................................................................106
Figure84DMCflowdiagramincludingconstraintsoncontrolsignal...................................107
Figure85DMCControlsignalofPVarray..............................................................................108
Figure86DCbuscurrents(PVarraywithDMC)....................................................................109
Figure87WindTurbineStepResponse.................................................................................111
XIV
Figure88VoltageandSOCofBatteryBank..........................................................................112
Figure89ControlsignalsPV(DMC)andWT(SMC)...............................................................113
Figure90DCBuscurrentofHPS(PVcontrolledbyDMC).....................................................114
Figure91VoltageandSOCoftheBatteryBank...................................................................115
Figure92ControlsignalsofHPS...........................................................................................116
Figure93DCbuscurrentsofHPS..........................................................................................117
Chaptei 1
1. Introduction
1.1 Justification
Hybrid Power Systems (HPS) are power systems that combine different electric
power energy sources. One of the main problems of the HPS is related to the control and
supervision of the power distribution system. The dynamic interaction between the grid
and/or the loads and the power electronic interface of renewable source can lead, to new
system, critical problems stability and power quality that are not common in conventional
powersystems.
Electronic Power Distribution Systems (EPDS) are power distribution systems where
theelectricpowerflowiscontrolledusingpowerelectronicconverters.EPDSarepresentin
many applications such as ship power systems, electric hybrid vehicles, hybrid power
systemsamongothers.Advancedcontroltechniquescanimprovetheperformanceofsuch
systemsbyimprovingenergymanagementcapabilityanditscapabilitytoadapttofaultsand
othersignificantchangesinoperatingconditions.
ThepurposeofthisresearchistodevelopcontrolstrategiesforahybridEPDSusing
Sliding Mode Control (SMC), and Model Predictive Control (MPC); SMC is very robust with
2
respecttosystemparametersvariationsandexternaldisturbances;MPChasshowntobea
very good control methodology for process control and other applications. MPC has been
applied in conventional power systems and to hybrid vehicles but no applications to HPS
EPDSisknown.
1.2 ResearchObjectives
ThemainobjectiveistodesignandimplementaHybridPowerSystemControllerfor
energymanagement.
Additionalobjectivesofthisworkwere:
ToimplementthemodelofaHybridPowerSystemusingMatlab.
TocomparetheperformanceofHybridPowerSystemwithSlidingModeControlagainst
othercontrolmethodologies.
Tovalidatetheproposedschemeusingsimulations.
1.3 SummaryofContributions
Here, we expanded the works of [1] and [2] in control of Hybrid Power Systems. In
[1]thesupervisorcontroldevelopedhadthreemodesofoperationanddidnotresultinan
acceptable performance of the battery bank. The latter [2] proposes a control for the
battery bank but the developments of other control strategies are not evident. Here, we
include other subsystems as traditional generators and add local shedding schemes when
thereisinsufficientgeneration.
3
The main contribution of this work is the development and validation of a

hierarchical control with two levels. The highest level is an online Supervisor control for a
modularHybridPowerSystemwithfivemodesofoperation.Thiscontrollerdeterminesthe
modeofoperationoftheallsubsystemsoftheHPS.Itisalsoeasytoimplement.
Other contribution is the MPC control of the PV array in the lowest level of the
hierarchicalstructure.
1.4 ThesisStructure
Chapter2providesaliteraryreviewanditisarrangedinfivesubtopics:HybridPower
Systems (HPS), Sliding Mode Control, Model Predictive Control, Energy Management in
Power Systems, and Hierarchical Control. Chapter 3 depicts the information related to the
model of the HPS describing the generalconstruction of each subsystem. Chapter 4 shows
thecontroldesignoftheHPSusingSlidingModetechniques,themodesinwhichthesystem
operates, and simulation results. Chapter 5 provides the control design of the Photovoltaic
Subsystem using Model Predictive Control techniques, and simulation results. Finally,
Chapter 6 presents the conclusions of the study and some recommendations for future
work.

4
Chaptei 2
2. LiteratureReview
2.1 Overview
Thissectionisdividedinfivesubsections;thefirstpart,givesabriefoverviewabout
hybrid power systems that includes their configurations, operation strategy, limitations,
and control requirements. The second part shows an examination of Hierarchical Control
and the different levels of control. The third part explains Sliding Mode Control Technique
and its applications in Power Systems, and Power electronics devices. In the fourth part a
review of model predictive control and its applications in power systems is presented.
Finally,thefifthpartdiscussesEnergyManagementtheoryappliedtoPowerSystems.
2.2 HybridPowerSystems(HPS)
Hybrid power systems are a combination of traditional generators, like diesel
generator,withrenewablesourceenergysuchaswindturbineandphotovoltaicpanels.The
mainobjectiveofthesesystemsistoextractmaximumpoweratlowcostswithgoodpower
quality,lowpollution,andreliablesupply[3].
5
2.2.1 HybridPowerSystemsClassification
HPSareclassifiedintwocategories:(i)gridconnectedHPS,whichareconnectedin
parallelwiththeutilitygridandcanbeusedatanylocation;and(ii)standalone(oroffgrid),
HPS,whichareusedtoattendloadsatremoteplacesbecausetheyareindependentofthe
utilitygrid[4,5].
The Grid Connected HPS generally use renewable generation power units as
photovoltaicpanelsandwindturbines,abatterybank(rarelyused)asabackupsource,and
the ac source is the utility grid. The basic configurations are Plantoriented or central
converter concept, moduleoriented or string converter concept, and moduleintegrated
converterconcept[6].
Stand Alone HPS are designed and sized to attend specific loads. The power units
commonly used are photovoltaic panels (DC source), wind turbines, and Diesel Generators
(AC sources), batteries are often used for backup power. Other power electronics
components like rectifiers, converters, and inverters are used to match the ac and dc
generation source with the voltage and frequency requirements of the load [5]. The basic
busconfigurationsofstandalonesystemsarepresentedonTable1.
The control system for HPS configurations should minimize fuel consumption by
maximizing power from the renewable sources. However, there are power fluctuations by
thevariabilityoftherenewableenergy,whichcausedisturbancesthatcanaffectthequality
ofthepowerdeliveredtotheload[3].
Table1
C
o
n
n
e
c
t
i
o
n
B
e
n
e
f
i
t
s
D
r
a
w
b
a
c
k
s
B
u
s
C
o
n
f
i
g
u
r
a
t
i
o
n
s
1BasicsConfig
All genera
loadareco
DC sourc
inverters t
and frequ
theACbus
It is
configur
the grow
increasin
needs[8
The syn
inverters
maintain
frequenc
the u
introduc
the use
increase
qualityp
Need 1
electrica
genset
configur
gurationonHy
ating source
onnectedtoan
ces need t
to match the
ency requirem
s[5].
a more
ation,whichfa
wth to mana
ng energy and
8].
nchronization
s and AC sou
n the volta
cy of the syst
ndesired ha
ced into the sy
e of inverters
the level o
problems.
10% to 18%
al energy fro
than
ation[8].
AC
bridPowerSys
and the
nACbus.
o have
voltage
ments of
All
loa
AC
cha
sou
modular
acilitates
age with
d power
D
d
r
f
e
T
n
v
g
t
a
g
T
t
of the
urces to
ge and
tem and
armonics
ystem by
s, which
f power
% more
om the
AC/DC
T
c
g
a
6
stems
generating s
ad are connect
sources nee
ange the AC
urce[5].
DC loads can
directly to the
reduces harm
from powe
equipmentint
The DC bus
need for f
voltage cont
generation so
to the bus a
application of
generatorsint
The fuel consu
to14%lower[8
Thepassesthro
conversion sta
generators p
affectsitseffici
D
source and th
ted to a DC Bu
ed rectifiers
source to D
be connecte
e DC bus, whic
monic pollutio
er electron
heload[5].
eliminates th
frequency an
trols of th
urce connecte
and enable th
variable spee
hesystem[5].
umption is 10
8].
oughtwopow
age, of the
power, whic
iency.
C
he
us.
to
DC
The sour
connecte
bus, and
loads are
[5].
ed
ch
on
nic
he
nd
he
ed
he
ed
0%
Both
through
inverte
flow be
increas
reliabili
continu
Fuel co
when
off[5].
wer
ac
ch

rces and load

d directly to
d the DC sou
connected to
buses are c
h a bid
r that permit
etween the tw
e the system
ity and
uity[5].
onsumption fal
equalization i
AC/DC
ds AC are
o the AC
urces and
aDCbus
connected
directional
ts power
wo buses,
m power
supply
ls by 30%
is turned
7
The control system should take control actions to maintain the power quality
conditions and power balance with the generators in combination with their energy
converters,thesolarandwindgeneratorscouldbecontrolledtosupplymaximumpoweror
providethepowernecessarytopreservethepowerbalanceattheload.Thesolarandwind
generationarecontrolledintwowaysMaximumpowerconversionandPowerregulation,
accordingtothegenerationconditionsandtheload[1].
2.2.2 ElementsofHybridPowerSystems
NextwewilldescribethedifferentelementsinaHPS.ItisdividedintoWindEnergy
ConversionSystem,PhotovoltaicPanels,andtheBatteryBank.
2.2.2.1 WindEnergyConversionSystem(WECS)
ThepurposeofWECSistoextractthepowerfromthewindandconvertittoelectric
power. The principal elements of typical WECS are the wind turbine, a generator like a
synchronous generators, permanent magnet synchronous generators, induction generators
(includingthesquirrelcagetypeandwoundrotortype),oraninterconnectionapparatusas
powerelectronicsconverters,tointerconnectthesystemtothebus[7].
Wind turbines are classified into the horizontal axis type (with two or three blades,
operatingeitherupwindordownwind),andtheverticalaxistype[7].
8
Figure1structureofatypicalWindEnergySystem.
The wind turbine could be designed for a variable speed or fixed speed operation.
Constant speed wind turbines can produce 8% to 15% less energy output as compared to
their variable speed counterparts. Nevertheless, they require power electronic converters
togiveafixedvoltagepowerandfixedfrequencytotheirloads[7].Controlsareincludedto
holdorregulaterotationalspeed,andoneoftheprincipalobjectivesistomaximizepower.
On other hand, winds turbines typically have at least three different control
actuators:bladepitch,generatortorque,andmachineyaw.Bladepitchisthemosteffective
methodofcontrollingaerodynamicsloads[9].
The nonlinear behavior of a wind turbine can make control design difficult. In pitch
control,thecontrolinputsgainsaretypicallythepartialderivativeoftherotoraerodynamic
torque with respect to blade pitch angle. These input gains vary with rotor speed, wind
speed, andpitch angle. If the control of a wind turbine is designed at one operating point,
maygivepoorresultsatotheroperatingpoints.Infact,thecontrollerpossiblywillresultin
unstable closedloop behavior for several operating conditions. To avoid this, for each
turbine operating points, regional controllers could be designed, and switched from one
regiontoanother.Thisgenerallyusesasswitchingparameterwindspeedorpitch[9].
The study of switching between controllers for wind turbines was done by Kraan in
[10]. Several problems such as undesirable switching transients were reported. However,
Bonge
transi
isano
theau
2.2.2.
electr
in wh
solar
gener
severa
atypic
ers in [11]
entsaremin
Theadapt
othermethod
uthorsrepor
.2 Photovo
The photo
icalenergy.
ich current
cells produ
ration,theym
Cells could
almodulesc
calsolarene
Figu
describes
nimizedbeca
ivecontrol,
dtocontrol
rtedaccepta
ltaicPanel
ovoltaic pan
Theexpres
through the
uce direct c
mustbelink
d be electri
couldbeput
ergyconvers
ure2Structure
the use of
ausethenex
investigated
thewindtu
ablesimulati
l
nels are the
ssionphotov
e device is to
current, so
kedbyaninv
cally connec
ttogetherto
sionsystem
ofatypicalso
9
controller
xtcontroller
dbyFreema
urbine,inwh
onsresults.
principal co
voltaicdenot
otally due to
when they
vertertocon
cted togeth
oformarray
[14].
olarenergysyst
conditionin
rtobeactiva
anandBalas
hichareada
omponent t
testheoper
o the transd
y are used
nverttheDC
er to form
ys.Figure2
temconnected
ng, in whic
atedisarran
in[12]and
ptedtochan
to convert s
ratingmode
duced light e
for grid co
CtoAC.
a photovolt
showstheb
dtoaDCbus[
h the switc
ngedforthis
Bossanyiin
ngingcondit
solar energy
ofaphotod
energy [14].
onnected p
taic module,
basicstructu
50]
ching
task.
[13],
tions;
y into
diode
The
ower
, and
ureof
10
The electric characteristic for a PV cell is presented on Figure 3, Points A and B

correspondtooperationundersufficientpowergenerationconditionsforagivenreference
power; on this mode, is desirable to operate on the right hand side of the PV array
characteristic (point B) because it allows a wider range of power regulation. Point C
represents the Maximum Power Operation Point (MPOP), so the system keeps the stored
energy as much as possible; the PV work on this point when the isolation regimes are
insufficient[15].
Figure3ElectriccharacteristicforaPVcell.
2.2.2.3 BatteryBank
Batteries convert the chemical energy stored inside into electrical energy. Battery
applications are huge, thats the reason that they are available in different sizes, voltage,
amphourratings,liquidorgel,ventedornonvented,etc[16].
Thebatterybankissupposedtobedesignedsothebatteriesdonotdischargemore
than 50% of their capacity on a regular basis. Discharging up to 80% is acceptable on a
11
restricted basis, such as an extended utility outage. Completely discharging a battery can
reduceitseffectivelifeordamageit[16].
The charged period of the battery bank could differ depending upon the availability
of other charging sources, the nature of the load and other factors. If renewable energy
(wind, solar, etc) powered the system, the charged period of the battery depends on the
weather orseasonal variations among others. However, the batteriesare not justused for
storage;theyaretooabufferforallthechargingenergywhichisbroughtintothem[16].
2.3 HierarchicalControl
The hierarchical system theory goes back to 1970s by Mesarovic, Macko, and
Takahara's in which the main characteristic is the fact that the decisionmaking process has
beendivided.Thereareanumberofdecisionmakerunitsinthestructure,butonlysomeof
them in a straight line access the control system. The decisionmaker units that define the
tasks and coordinate are at a higher level on the hierarchy; the lower levels have direct
contactwiththeprocess[17].
ThedesignedcontrolforHPSdevelopedinthisthesisisakindofhierarchicalcontrol
withfourdecisionmakerunits.
2.3.1 Controllevels
The control problem of a complex system can be divided into different levels; each
level has different control objectives that are handled by different controllers. When a
distur
norma
functi
decisi
contro
2.3.1.
overa
`produ
days[
rbance occu
aloperation
Figure 4
ons.
The contr
on units to
olunitstoea
.1 TheStra
This is the
ll operation
uction'or`s
[18].
rs, the obje
asquicklya
shows a h
rol strategy
decide the
achpowerg
ategicLevel
e highest lev
. The operat
hutdown'[1
Opera
(In
Con
ective is typ
aspossible[1
hierarchical
Figure4Mult
developed
operationa
generationsu
l
vel of the s
ting conditio
17],andthe
Tacticalle
(Supervi
contro
ationallevel
ndividual
ntrolUnit)
12
pically to br
17].
control str
tilevelcontrol
in this wo
l mode of t
ubsystemis
structure; th
on of the sy
timeconsta
Strategic
evel
sor
ol)
Operationalleve
(Individual
ControlUnit)
ring the ent
ructure with
ofasystem.
rk, is over
the HPS are
overoperat
his level mad
ystem can be
antsaretypic
Level
el
Tactical
(Superv
contr
Operation
(Individ
Control
tire system
h three dif
two lower
over tactic
tionallevel.
de the decis
e characteri
callyinther
level
visor
rol)
nallevel
dual
Unit)
safely back
fferent leve
level; the m
al level, and
sions conce
zed as `start
rangeofhou
k into
els of
make
d the
rning
tup',
ursto
13
2.3.1.2 TheTacticalLevel
Thisleveltakesthelocaldecisionsconcerningthelocaloperation.Thislevelactslike
asupervisorcontrolwhichswitchesbetweendifferentoperationsmodesorcouldshutdown
the controlled subsystem. Typically, the time constants at this level are in the range of
minutesorseconds[18].
The operation of a supervisor in hierarchical control has two steps. First, an
observation step which collects information concerning the controlled system and its
environment. Second the decision step which uses this information and prior information
withthepurposeofselectdesirablecontrol[17].
2.3.1.3 TheOperationalLevel
Onthisleveltheactualcontroloftheplantisbeingperformed;thecontrolobjectives
foreachoperationmodehavetobedefined.Typicallythetimeconstantsareintherangeof
milliseconds [18]. Over this level, we developed the individual control units for each
subsystemintheEPDSusingSlidingModeControlorModelPredictiveControl.
2.4 SlidingModeControl(SMC)
ThebasicsconceptsanddefinitionspresentedinthissectioncomefromSlidingMode
Control theory developed 30 years ago in Russia by [19] , and to applied to on power
electronicsconverters20yearsagoby[20,21].
14
Before discussing SMC techniques it is necessary to understand Variable Structure

Systems (VSS), as implied by the name, are systems whose structures are changed
deliberately through the transient, according to a predetermined structure control law, to
accomplishthecontrolobjectives[22].
2.4.1 BasicsConceptsofSMC
SlidingModeControlisatypeofVariableStructureControlcharacterizedbyasetof
feedback control laws and a decision rule, and could be seen as an arrangement of
subsystems,whereeachsubsystemhasapredeterminedcontrolstructureandisapplicable
for specified regions of system behavior, on this control scheme, the feedback is not a
continuousfunctionoftime[23].
The sliding mode control operates in a basic method as follows: a sliding surface is
defined with the equilibrium point, and the system is forced to be held into the sliding
surface (existence condition), and then the system must reach the equilibrium point
(stability) [24]. This control scheme implies a selection of a surface such that the system
trajectory exhibits desirable behavior when confined to this surface and the selection of
feedback gains so that the system trajectory intersects and stays on the surface [25]. The
existenceconditionandstabilitymustbeverifiedtoguaranteethecorrectlyoperationofthe
controlledsystem.
The main idea of the SMC is illustrated in Figure 5. First, drive system to stable
surface(reachingphase),thenslidetoequilibrium(slidingphase).
15
Sliding mode control has two advantages: the dynamic behavior of the system may
be tailored by the particular choice of switching function, and the closedloop response
becomesinsensitivetoanexternaldisturbanceandparametersvariations[23].
Figure5SlidingModeIdea[23]
2.4.2 SMCinPowerSystems
The application of the sliding mode control technique in Power Electronics devices
suchaDCDCconvertersshowsthatthiscontrolapproachcouldgivegoodresultsintermsof
robustnesstowardloadandinputvoltagevariations,whilemaintainingadynamicresponse.
SMCwasusedinHybridPowerSystemstocontrolaphotovoltaicarrayin[15],anda
wind turbine in [26] by Valenciaga et al. The objective was to control the operation of the
wind subsystem to complement the photovoltaic generation, so the power demand is
satisfied.
2.4.2.1 DC/DCConvertersofaVariableStructuresystem
DC/DCconvertersareaffinenonlinearsystemsdescribedby[27]:
x = (x) +g(x)u (2.1)
16
x e +
n
u e +
The control signal u is discontinuous, and could take two values (0 or 1). Its
discontinuous points correspond to changes on converter structure, for this reason, DC/DC
convertersareavariablestructuresystem[27].
The principle of the switching control law is to force the nonlinear plants state
trajectory against a prespecified surface S in the state space and to preserve the plants
state trajectory on this surface for following time [28]. The control switching law is defined
asfollow:
u = _
1 i s(x) u
u i s(x) < u
(2.2)
Where s(x) is called the switching boundary and its determined by a surface of n1
dimension,andnisthestatedimension.
S = {x e +
n
: s(x) = u] (2.3)
Sistheswitchingsurface,whichisalsocalledaslidingsurface[22].
Themostimportanttaskistodesignaswitchedcontrolthatwilldrivetheplantstate
totheswitchingsurfaceandmaintainitonthesurfaceuponinterception.
2.4.2.2 SMCinDC/DCconverters
Onthissection,wearegoingtoexplainhowtocontrolwithSMtechniquetheoutput
currentofaDC/DCconverter(Figure6).
17
Figure6DC/DCConverter.
To create a sliding surface, in order to achieve the control objective, we choose the
switchingfunctionas[27]:
S(x) = i
o
-i
oc]
(2.4)
wherei
oc]
isthedesiredoutputcurrent.
WesupposethatthesystemhasanequilibriumstateX
e
withi
o
= i
oc]
forauvalue
betweenoneandzero.
Tocreateaslidingmoderegimetheconditionstobesatisfiedare[27]:
ds
dt
=
d
c
dt
-
d
crc]
dt
< u i u = 1 (2.5)
ds
dt
=
d
c
dt
-
d
crc]
dt
> u i u = u
2.5 ModelPredictiveControl(MPC)
The MPC method is based on a prediction model of the system response to obtain
the control action optimizing the future behavior of a plant by minimizing an objective
functioninanonlinemode[29].
There are many applications areas where MPC is successfully in use at the present
time such as the process industry like chemicals, automotive, food processing, metallurgy,
aerospace, pulp and paper, and power systems [30,31] among others. The good
18
performance of these applications shows the capability of the MPC to accomplish highly
efficientcontrolsystemsabletooperateduringlongperiodsoftime[32].
MPC can be used to integrate issues of optimal control, control of processes with
time delays, stochastic control and multivariable control. The advantages over other
methodsare[29,32]:
Theconceptsareveryintuitiveandthetuningisrelativelyeasy.
Itintroducesfeedforwardcontroltocompensatemeasurabledisturbances.
MPCpermitsconstraints,andthesecanbeincludedduringthedesignprocess.
Onlinecomputationwithlowercomputationalrequirements.
Drawbacksare[32]
Theneedforanaccuratemodeloftheprocess.
Allthecomputationshavetobecarriedoutateverysamplingtime.
2.5.1 MPCtechnologies
This section shows the model predictive control technologies that are commercially
available and that have large impact on the industrial world, the various MPC algorithms
differ among themselves in the model used to represent the process and the noise model
andcostfunctiontobeminimized.
19
The history of industrial MPC began with the Linear Quadratic Gaussian (LQG)
Controller, in which, the process can be described by a linear statespace model driven by
Gaussiannoise,andtheinitialstateisassumedtobeGaussianwithnonzeromean.
x
k+1
= Ax
k
+Bu
k
+6w
k
(2.6)
y
k
= Cx
k
+
k
TheobjectivefunctionincludesseparatestateandinputweightmatricesQandRto
allow for tuning tradeoffs and penalizes expected values of squared input and state
deviationsfromtheorigin.
[ = x
k+]
1
Qx
k+]
]=1
+u
k+]
1
Ru
k+]
(2.7)
The optimal input u
k
, is computed using an optimal state feedback law u
k
= k
c
x
whereK
c
iscalculatedbysolvingamatrixRiccatiequation[33].
The LQG controller has stabilizing properties. In LQG theory, it is not simple to
includeconstraintsontheprocessinputs,statesandoutputs[33,34].
Grieder et al. in [35] presented an algorithm to calculate the solution to the
constrained infinitetime, linear quadratic regulator (CLQR) combining reachability analysis
with multiparametric quadratic programming to get the optimal piecewise affine (PWA)
feedback law. The algorithm reduces the time necessary to compute the PWA solution for
theCLQRwhencomparedtootherapproachesmakingCLQRanattractivesolutionevenfor
fast processes. This situation leads to the expansion, in industry, of a general model based
controlmethodology,callednowMPC,inwhichthedynamicoptimizationproblemissolved
onlineateachcontrolinterval[36].
20
Process inputs are calculated so as to optimize the outputs over a time interval
knownasthepredictionhorizon[37].Theprocessmodel,thatdescribestheplantdynamic,
couldtakeanymathematicalform,andisusedtopredictthefutureplantoutputs,basedon
currentandpastvaluesandontheoptimalproposedfuturecontrolactions.Theoptimizer
calculates these actions taking account the constraints and the cost function, thats the
reason because the future constraints violations could be predictable and prevented. This
structureisshowinFigure7,[33].
Figure7BasicstructureofMPC[10].
The MPC control action at time k is obtained by solving the optimization problem
givenby
min
x(k),u(k)
[(x(k), u(k)) = |y(k +]|k) -w(k +])]
2
+z(])|u(k +] -1)]
2
N
c
]=1
N
p
]=1

(2.8)
Subjectto
u
mn
u
k
u
mux
-u u
k
u
y
mn
Cx
k
y
mux
x(k +1|k) = Ax(k +]|k) +Bu(k +]|k)

21
wherewisthereferencetrajectory,zisthecontrolweightingfactor,thepredictedoutputs
y depend on the future control signals u, and on the known values up to instant k (past
inputsandoutputs),whicharethosetobesenttothesystemandtobecalculated.Thefirst
inputoftheoptimalcontrolsequenceissenttotheprocessatthesametimeasthecontrol
signals
1
calculated are rejected, and the problem is solved again at the next time interval
usingupdatedprocessmeasurements[32,33].
Figure 8 shows the state of a SISO MPC system that has been operating for many
sampling instants. The current instant is represented by the integer k, y represents the
measuredoutputandushowsthecontroleffort[32,37,38].
Figure8Controllerstateatthek
th
samplinginstantadaptedfrom[39].
In the late seventies, various articles appeared showing an interest in MPC by
industry. Richaletet al. presented the first description of MPC control applications in 1976
[33]. His publications presenting Model predictive Heuristic Control (MPHC) and the
1
Thenumberofcontrolintervalsoverwhichthemanipulatedvariablesaretobeoptimizediscalledthecontrol
horizon[37].
22
software that developed was named IDCOM (Identification and Command), which permit
inputandoutputconstraints,impulseresponsemodelfortheplant,quadraticperformance
objectiveoverafinitepredictionhorizon[32].
In 1979, Cutler and Ramarker presented the Dynamic Matrix Control (DMC) as an
unconstrained multivariable control. This algorithm uses the linear step response of the
plant. Optimal inputs are calculated as the solution to a least squares problem using a
quadraticperformanceobjectiveoverafinitepredictionhorizon[32,33].In1982,Garciaand
MorarishowedthattheDMCalgorithmwasclosedloopstablewhenthepredictionhorizon
wassetlongenoughtoincludethesteadystateeffectofallcomputedinputmoves[34].
In 1983, Cutler et al. described the Quadratic Program DMC (QDMC) in which input
and output constraints appear explicitly, and the solution can be accomplished readily via
standardcommercialoptimizationcodes.Thisalgorithmrepresentsasecondgenerationof
MPC[33].
The second generation did not permit the combination of multiple objectives into
one objective (function), and did not allow the designer to reproduce the correct
performance requirements. These are the reasons why third generation of MPC appeared
with IDCOM algorithms. The third generation MPC used two separate objective functions,
one for the outputs and one for the inputs [40]. Hard and soft constraints were
incorporated, and the quadratic output objective function was minimized subject to
constraintdegree[37].
23
The Shell Multivariable Optimizing Controller (SMOC) proposed by (Marquis &

Broustail,1998;Yousfi&Tournier,1991)combinesstatespacemethodswiththeconstraint
handling features of MPC. This algorithm is equivalent to solving the LQR problem with
inputandoutputconstraints,exceptthatitisstillformulatedonafinitehorizon,butitdoes
nothavethestabilizingpropertiesoftheLQRalgorithm[33].
The fourth generation is represented by DMCplus and Robust Model Predictive
Control(RMPCT),whichpresentanautomaticwayfortuning.Theusercouldenterdirectly
the estimates of model uncertainty, and compute the tuning parameters to optimize the
performance for the worstcase model mismatch. This generation provides some
mechanism to recover from an infeasible solution, distinguishes between several levels of
constraints(hard,soft,ranked),allowsforawiderrangeofprocessdynamicsandcontroller
specifications, and addresses the issues resulting from a control structure that changes in
realtime[33].
2.5.2 MPCinPowerSystems
There are several studies of model predictive control in power applications such as
areapowernetworks,andvehicularelectricpowersystems.
Camponogaraetal.in[30]presentedapowersystemapplicationofSCDMPCintwo
or more autonomously controlled areas (see Figure 9), each area usually consist of various
generators and loads, for studying Load Frequency Control (LFC). They assign MPC
controllers to control the generator power output. The purpose of LFC is to keep the
freque
throug
centra
(AGC)
realize
agent
one a
coope
strate
neighb
exper
netwo
2.6
others
ency deviati
ghthetielin
Venkat et
alized MPC,
. They sho
edthrougho
Hines P, e
s, to the pro
gent is at e
eratewithit
egyissucces
bor networ
iments also
orks.
EnergyM
The main
s secondary
ion of the sy
neatzero.
t al. in [41
communica
ow that the
outdistribute
et al. in [42
oblem of ar
each node o
sadjacents
ssfulaslong
k arrives at
o revealed
Managem
goal of Ene
y objectives
ystem at ze
Figure
1] use the
ation based
performanc
edMPCstra
2] and [43]
rresting casc
f a power n
subsystemsi
asthetime
t a steady s
the value
ment(EM)
ergy Manag
are to imp
24
ro and to p
e9Twoareap
same exam
MPC, and S
ce benefits
tegies.
show a new
cading failur
network to c
inmakingit
eamongMP
state before
of even si
)inPowe
gement is to
prove energy
reserve the
powernetwork
mple to com
Standard Au
obtained w
w applicatio
res in power
control a sin
sdecisions.
PCiterations
e the next
mple collab
rSystems
o minimize
y efficiency
variation of
k
mpare the
utomatic Ge
with centraliz
on of DMPC
r systems.
ngle variable
Theirresul
sissufficient
control acti
boration sch
s
costs or ma
and to red
f the power
performanc
neration Co
zed MPC ca
C methods,
On this stra
e using MPC
tsimplytha
tlylargetha
ion occurs.
hemes in a
aximize ben
duce energy
r flow
ce of
ontrol
an be
using
ategy,
C and
atthe
atthe
The
agent
nefits;
y use,
25
develop and keep efficient monitoring, report, and management strategies for intelligent
energyusage,andreducetheimpactinterruptioninenergysupplies[44].
ThefunctionofEnergyManagementistoprioritizerealtimepowerdemandfromthe
loads and distribute power resources available from the generation and storage devices in
anoptimizedapproachformaximumefficiencyandperformance[45].
The power management system could be divided into the following subsystems:
powergeneration,energystorage,powerbus,electricalload,powerelectronics,andPower
ManagementController(PMC).Figure10showsagenericpower/energymanagementand
distribution system. For a particular configuration, not all components or subsystems are
required,andsomesmallchangesinthesystemtopologymightbenecessary[45].
Figure11presentstheprinciplesoftheenergymanagement.Themajortaskforthe
poweranenergymanagementsystemistoprioritizetheloadpowerrequestandtoallocate
limitedpowerresources.Itisnotpracticaltoofferapermanentpowercapacityhigherthan
theaveragepowerdemand.Supplyandstoragemustmeetbriefpeaksinpowerneeds[45],
whicharethereasonwhyineachtimeintervalofoptimization,themanagementdetermines
the operation mode of all components of the power system. It is best if the objective
functionbecomesminimal[46].
26
Figure10GenericPower/Energymanagementanddistributionsystemadaptedfrom[45].
Figure11PrinciplesofEnergyManagement.
27
2.6.1 EnergyManagementinHybridPowerSystems
EnergymanagementisperformedinHPSdeterminingonlinetheoperationmodeof
generation subsystems, switching from power regulation to maximum power conversion;
theenergybalancedependsoftotaldemand,andgeneration[1].
Valenciaga and Puleston in [1] designed a supervisor control, using robust sliding
mode control, with three modes of operation: in mode 1 the wind subsystem is in power
regulation against the solar subsystem is off, and the battery bank is in recharge cycle; in
mode2thewindsubsystemisinmaximumpowerconversionatthesametimeasthesolar
subsystemisinpowerregulation,andthebatterybankisinrechargecycle;inmode3,both
windandsolarsubsystemsareinmaximumpowerconversion,andthebatterybanksupply
power to the load. On this scheme, the objective is to control the operation of the wind
subsystemtocomplementthephotovoltaicgeneration,sothepowerdemandissatisfied.In
thiswork,theStateofChargeofthebatterybankandloadsheddingarenotconsidered.
Thecontrolofwindturbineandthephotovoltaicarraywasdevelopedbeforein[26]
and[15]respectivelyusingslidingmodetechniques.However,MPChasnotbeenappliedto
control of HPS, but West M. et al. in [47] applied MPC techniques for EM in hybridelectric
vehicle (HEVs) drivetrain incorporating numerous energy/power sources. This strategy is
used to control the power drawn from a battery pack and a supercapacitor peak power
buffer,toprovideanallelectricdrivetrain.TheschemeusedisshowninFigure12.Similar
studies to control the electric power system with energy management for HEVs have been
28
proposedbyKootetal.in[48]toreducethefuelconsumptionandemissionsoveradriving
cycle.
Figure12ElectricVehiclePowerFlow.
On this technique, the design procedure starts defining the cost function, such as
minimizing the fuel consumption and emissions over a driving cycle. To find the optimal
control, they useda variant of MPC employingzone control, and adding an additional slack
variableointothecostfunctionofgeneralpredictivecontrolgivenby
[(u, k) = |y(k +]|k) -w(k +]) +o(k +])]
2
+z(])|u(k +] -1)]
2
N
c
]=1
N
p
]=N
c
(2.9)
Subjectto
(k)
(k)
|o(k)| o
mux
where
(k)
(k) include all the inputs, outputs and the states constraints, and
|o(k)| o
mux
isanadditionalconstraintontheslackvariable[47].

29
Chaptei S
3. HPSSimulationModel
The first step is the implementation of a simulation model of Hybrid Power System
using Simulink; this software was selected because it is a special package for modeling,
simulating,andanalyzingdynamicsystems,supportslinearandnonlinearsystems,modeled
ineithercontinuoustime,sampledtime,orboth[49].
Allmodelswereselectedfromdifferentsources:thewindturbinemodelispresented
in [26], the photovoltaic panels model was developed by OrtizRivera in [50], the battery
Bankwaspresentedin[51],andtheloadmodelistheDCZonalElectricalDistributionSystem
(DCZEDS), which is the DC part of the an Integrated Power System (IPS) in [52], all
subsystemsoftheloadhavealocalcontroller.
3.1 HybridPowerSystem
The Electric Generation Hybrid System (EGHS) that will be used on this work
combines wind energy, solar sources and traditional sources. Each unit constitutes one
subsystemtocontrol;theoverallsystemisshowinFigure13. Thehybridpowersystemto
be studied here can be divided in six components; power generation (Wind turbine, PV
30
Panels, and the Grid), energy storage (Battery Bank), power bus, electric load (three zone
EPDS),powerelectronics(DC/DCconverters),andthePowerManagementController(PMC).
Figure13HybridgenerationSystem
The wind generation unit comprises a windmill, a multipolar permanentmagnet
synchronousgenerator(PMSG),arectifier,andadc/dcconvertertointerfacethegenerator
with the dc bus. The solar unit comprises several panels coupled to the dc bus through a
DC/DCconverter.Thedcbuscollectstheenergygeneratedbybothunitsanddeliversitto
theloadand,ifnecessary,tothebatterybank.
The voltage of the dc bus is set by the battery bank, which comprises leadacid
batteriescoupledinaserial/parallelarray;theconverterscontroltheoperationpointofthe
windturbineandPVPanels.
31
Theloadsservedconsistofthreesimilarzones.Eachzoneisconnectedtothebusby
aDC/DCconverter.
ThePMCisacentralcontrolunitforthepower/energymanagementanddistribution
systemtocontrolandcoordinatethesystem.Thiscontrollersendstothepowerconverters
thecontrolsignalsandreceivingthesensorsignalsandthestatusreportfromtheseunits.
Figure 14 shows the MATLAB/SIMULINK simulation model of the HPS; a lot of sub
modelsareputtogethertocomposethemainmodel.ToputintopracticetheEMwehave
tocontroltheoverallsystemandimplementthegeneralEMalgorithmonthePMC.
Figure14Matlab/SimulinkSimulationmodeloftheHPS.
wi nd speed
u_wi nd
u_PV
cel l Temp
modos
i _Wi nd
i _Load3
i _Load2
i _Load1
i _Load
i DG
i _sol ar
i _battery
cti vateWi
acti vateP
acti vateDG
of f Z1
of f Z2
of f Z3
iLoad
iLoad1
iLoad2
iLoad3
Zone Load
Wind Speed
Win dref
activ ate wind
iw
u wind
Wi nd Subsystem
SOC
modos
SetPointPV
SetPointGrid
SetPoint wind
Supervi sor Control
Temperature
On PV
ipv ref
io
u PV
Sol ar Subsystem
Graphs
[Off1]
[DGref]
[i _sol ar]
[i _Wi nd]
ti vateWi nd
[Off3]
[Off2]
[acti vateDG]
[i DG]
[i pvref]
[i _Load]
cti vatePV
[i wi ndref]
iw
io
il
iDG
ib
DC Bus
Vin
IDG ref
iout
Conventi onal Energy
Cel l Temperature
vb
ib
v b
SOC
Battery Bank
V
3ph-Source
32
3.2 WindTurbine
The wind turbine model used in this work was proposed in [26] by Valenciaga et al.
Onthismodel,theturbineislinkedtothebatterybankthroughadiodebridgerectifierand
a DC/DC converter. Figure 15 shows the MATLAB/SIMULINK simulation model of the wind
subsystem.
Figure15Simulinkmodelofthewindsubsystem.
The mechanical power generated by a turbine is proportional to the air density (),
the power coefficient of the rotor (C
p
), the cube of the wind speed (), and the swept area
(A)asshowbelow.
P
t
= C
p
(z)pAu
3
(3.1)
Thepowercoefficientvalue,dependsupontheaerodynamicsoftherotorblades,the
bladesangle,andthewindvelocity,andcouldbedescribedintermsofthetipspeedratioz
whichisgivenby:
z =
r
m
u
(3.2)
where,r isthebladelength,and
m
istheangularshaftspeed.
Thewindturbinetorqueisgivenby:
I
t
=
P
t
m
=
1
2
C
t
(z)pAru
2
(3.3)
Torque
Vs
3
wm
2
iw
1
Wind Turbine
Vw
Wm
Tt1
Power Converters
Is
Vb
Ux
Vs
Iw
PMSG
Vs
Tt
Is
Wm wind speed
3
ux
2
vb
1
33
where, C
t
(z) =
C
p
(x)
x
,isthetorquecoefficientoftheturbine.
The expression of the electrical angular speed, corresponding to the minimum shaft
speedbelowwhichthesystemcannotgenerateisgivenby:
c Im
=
I
s
m
(3.4)
where,
m
is the flux linked by the stator windings, and I
s
is the line voltage on the
Permanent Magnet Synchronous Generator (PMSG). Figure 16 depicts the
MATLAB/SIMULINKsimulationmodelofthewindturbine.
Figure16SimulinkmodelofWindTurbine.
The PMSG dynamic model in a rotor reference frame is given by the following
equations[26],andtheMatlab/SimulinksimulationmodelisshowninFigure17.
t
q
= -
R
s
I
i
q
-
c
i
d
+
m
I
-
n:
b
i
q
u
x
SSI_i
q
2
+i
d
2
(3.5)
t
d
= -
R
s
I
i
d
+
c
i
q
-
n:
b
i
d
u
x
SSI_i
q
2
+i
d
2
(3.6)
c
=
P
2[
_I
t
-
S
2
P
2
m
i
q
] (3.7)
Tt
2
Pt
1
TSR
R*u[2]/u[1]
Mechanical Power
1/2*u[1]^3*rho*pi *R^2*u[2]
Divide
Cp Vs TSR
TSR Cp
Cp
Wm
2
Vw
1
34
where i
q
and i
d
are, respectively, the quadrature current and the direct current; L and R
s
aretheperphaseinductanceandresistanceofthestatorwindings;PisthePMSGnumberof
poles;Jistheinertiaoftherotatingparts;
m
isthefluxlinkedbythestatorwindings;and
u
x
isthecontrolsignal.
Thevoltage I
s
isexternallyimposedbytheDC/DCconverterasafunctionoftheduty
cycle,andisdescribedby:
I
s
=
n:
b
SS
u
x
(3.8)
If we assume an ideal static conversion, the current at the output of the DC/DC converter
couldbedeterminedasshowninthenextequation.
i
w
=
n
2S
_i
q
2
+i
d
2
u
x
(3.9)
Figure17SimulinkmodelofPermanentMagnetSynchronousGenerator.
Wm
2
Is
1
we_dot equation
f(u)
we
[we]
we
[we]
magnitude
calculation
f(u)
iq_dot equation
f(u)
iq 1
[iq]
iq
[iq]
iq
[iq]
integrator
1
s
id_dot equation
f(u)
id 1
[id]
id
[id]
id
[id]
change
wm
-K-
Integrator 2
1
s
Integrator 1
1
s
Tt
2
Vs
1
35
TheparametersusedinthesimulationsareinTable2.
Table2WindturbineandPMSGparameters.
PMSGnominalpower
5KW
P 28
R
s
0.3676
L 3.55mH
m
0.2867Wb
J 7.856Kgm
2
R 1.84m
3.3 PhotovoltaicPanel
Thephotovoltaicpowermodelthatweusedinthisworkwasproposedin[50].This
model depends on several cell parameters an on variable environment conditions such as
thetemperatureoverthesolarpanelT,thecharacteristicconstantfortheIVcurvesb,the
percentageofeffectiveintensityofthelightoverthesolarpanels,theopencircuitvoltage
V
oc
, the short circuit current I
sc
, and a shading linear factor . This model was selected
because it uses the electrical characteristics provided by the solar panel data sheet, and its
electrical behavior could be modeled by a nonlinear current source connected with the
intrinsiccellseriesresistance.Inthismodel,theIVandPVrelationofthesolarpanelare:
i
p
= I
mux
- I
mux
c
_
v
p
b(yu+1-y)(v
mcx
+:
)
-
1
b
]
(3.10)
P
p
= I. i
p
= I
mux
I
- I
mux
I
c
_
v
p
b(yu+1-y)(v
mcx
+:
)
-
1
b
]
(3.11)
Where
o =
E
E
N
(3.12)
maxim
array
conve
where
levelo
The shadi
mumtomin
foraneffect
ThePVpa
erterwhichi
e is the b
onthePVpa
Figure18s
ng linear fa
imumintens
tiveintensity
nelisdevelo
sdescribed
battery volta
anelarrayte
showstheM
ctor is th
sityoflight.
yoflightles
opedaround
bythefollow
age, is the
erminals,and
MATLAB/SIM
Figure18S
36
e percent o
Theopen
ssthan20%o
daDCbusan
wingequatio
e current inj
d istheco
MULINKsimu
Simulinkmode
of maximum
circuitvolta
overthesol
nditisconn
ons[15]:
jected on th
ontrolsignal
lationmode
elofthesolarp
m voltage los
agemarkof
arpanelis
nectedthrou
he DC bus,
.
elofthesola
panel.
ss fro
fthesolarp
[50].
ughaDC/DC
is the vo
arpanel.
(3.13)
om a
anels
(3.14)
(3.15)
(3.16)
buck
(3.17)
(3.18)
oltage
37
ThedatasheetfortheSLK60M6panelispresentedinappendixA,andsummarizedon
Table 3. The maximum power generated by the panel is 279.7W; the array that we used
havetwopanelsinseriestosetthevoltageandtwopanelsinparalleltosetthecurrent.
Table3PhotovoltaicModuleSpecifications.
SLK60M6

Isc 7.52A
Voc 37.2V
Iop 6.86A
Vop 30.6V
b 0.07292
TCi 2.2mA/C
TCv 127mV/C
Vmin 32.55V
Vmax 37.312V
ThetotalpowergeneratedbythePVarrayis559.48W,themaximumcurrentis15A,
andthemaximumoperationvoltageis74.4VasshowninFigure19.
Figure19CurrentVoltagecurveofPVarray
3.4
mode
curren
shown
the le
relatio
righth
Battery
The batte
lingthestea
The batte
nt controlle
ninFigure2
The state
eft hand si
onship of th
handsideof
Thestates
Bank
ry model th
adystateand
ery model is
d current s
20.
F
of charge (S
ide circuit;
e battery w
fthediagram
spacemode
hat we used
dthedynam
s modeled
source and
igure20Batte
SOC) of the
the mode
was represen
m.
lofthecircu
38
was develo
micbehavior
by two circ
a nonlinea
eryModeladap
battery is re
l of the tr
nted by two
uitisgivenb
oped by [53
ofthebatte
cuit diagram
r voltage co
ptedfrom[53]
epresented
ransient be
RC circuits
by[51]:
3]. This mod
ery.
ms which ar
ontrolled vo
.
bya large c
havior and
and series r
del is capab
re coupled
oltage sourc
apacitor
voltagecu
resistance on
ble of
via a
ce as
in
rrent
n the
(3.19)
39
where C
1L
and R
1L
are the capacitance and resistance in the long transient RC circuit, C
1S
andR
1S
arethecapacitanceandresistanceintheshorttransientRCcircuit,R
S
istheseries
resistance,gisthenonlinearSOCfunction.Theinputuisthecurrententeringthebattery,
andtheoutputyisthevoltageacrossthebatteryterminals.
The model was implemented in Simulink as shown in Figure 21. The relationship of
nonlinear SOC was implemented via a lookup table with a set of ten values range from full
chargetocompletedischarge[51].
Figure21SimulinkmodelofLiIonBattery.
All the parameters in the model are multivariable functions of SOC, current, cycle
numberandtemperature,andcanbecalculatedfollowingthenextequations[53].
C
CAP
= S6uu - Copocity -
1
(cyclc) -
1
(tcmp)
(3.20)
SOC
Vts
Vtl
SOC
2
vb
1
Vsoc VSOC
Lookup Table
1
s
1
s
1
s
Rs
-1/Rtl
-1/Rts
1/Ctl
1/Cts
1/Ccap
ib
1
40
whereCapacityisisthenominalcapacityinAhrandf
1
(cycle)andf
2
(temp)arecyclenumber
andtemperaturedependentcorrectionfactors,respectively.Forthissimulationmodel,the
f
1
(cycle)andf
2
(temp)aresettoone,andthebatterycapacityis22Ahr.
TheopencircuitVoltageV
OC
ischangedtodifferentcapacitylevels.
I
0C(S0C)
= -1.uS1c
-35S0C
+S.68S +u.21S6S0C -u.1178S0C
2
+u.S2u1S0C
3
(3.21)
All the extracted RC parameters are constants around 20%100% SOC and vary
exponentiallysurroundedby0%20%SOCcausedbytheelectrochemicalreactioninsidethe
battery.
R
S(S0C)
= u.1S62c
-24.37S0C
+u.u7446 (3.22)
R
1S(S0C)
= u.S2u8c
-29.14S0C
+u.u4669 (3.23)
C
1S(S0C)
= -72S.9c
-13.51S0C
+7uS.6 (3.24)
R
1L(S0C)
= 6.6uSc
-155.2S0C
+u.u4984 (3.25)
C
1L(S0C)
= -6uS6c
-27.12S0C
+447S (3.26)
TheparametersusedinthesimulationoftheBatterybankareinTable4.
Table4ParametersoftheBatteryBank
C
CAP
80000F
C
TS
600F
C
TL
5000F
R
TS
0.05
R
TL
0.05
R
S
0.05
V
SOC0
48V

41
Chaptei 4
4. HierarchicalControllerforaHPS
ThemainobjectiveoftheHPSistosatisfytherequirementsoftheelectricalloadsat
the same time as maximizing the utilization of renewable energy sources while optimizing
theoperationofthebatterybankandtheconventionalgenerators.
The developed control strategy consists of a hierarchical two level structure:
individual control units for the individual energy conversion systems and a supervisory
controllerwhichdeterminesthereferencecurrentsfortheDC/DCconvertersandtheSOCof
thebatterybankforenergymanagement.
4.1 IndividualControlUnits
Each energy conversion system has an individual control unit in order to operate it
according to the system needs. The system energy requirements are delivered to these
units through signals coming from the supervisory controller. However, the supervisory
controller takes into account the limits of the energy conversion systems to which it is
deliveringacommandsignal.
42
On this work, there are three different individual control units. The wind and the
solar subsystems are controlled using SMC following [15,26]. The control strategy for the
batterybankwasdevelopedin[2].
4.1.1 SMCforaWindTurbine
The wind subsystem acts under supervisor control commands whichever supplying
the total load or tracking the operation point of maximum power conversion. This
subsystem is turned on in all modes of operation because is designated as the principal
energysource.Theimpulseforthisdesignchoicewasthatitwastakenfromtheperspective
of applications in geographical areas with affluent wind regimes. However, we could
considerate the solar subsystem as the main generator source and the wind subsystem as
thesecondaryrole.
4.1.1.1 SlidingModecontrollerDesign
Themainobjectiveistocontrolthepowerproducedbythewindsubsystemintothe
DCbustosatisfythetotaldemand(theloadandthebatterychargerequirements I
bc]
)[26].
Inthiscase,thecontrolisjustusedtorejecttheunwantedinternalforcesthatdraw
thesystemapartfromtheslidingsurface.
43
4.1.1.2 ModesofOperation
Forthissubsystem,therearetwomodesofoperationasdescribedin[26].Thefirst
occurs when the wind subsystem could generate enough power to satisfy the total power
demand.
P
w c]1
= I
b
(i
L
+I
bc]
) (4.1)
Thesecondmodeofoperationoccurswhenthemaximumenergycapturedfromthe
windisnotenoughtosatisfythetotaldemand.Forthiscase,wemakeasapowerreference
theoptimumpowerextractedbytheturbineminuslosses.
P
w c]2
= P
w opt
-P
Iosscs
P
w c]2
= K
opt
m
3
-
S
2
(i
q
2
+i
d
2
)r
s
(4.2)
where
K
opt
=
C
t
(z
opt
)pAR
3
2z
opt
2

(4.3)
z
opt
isthetipspeedratiothatmaximizesthepowerextractedtothewind.
The boundary between the operation modes of the wind subsystem is given by the
angular shaft speed of the boundary point, and is obtained by equating the mechanical
powerreferencesofbothmodesofoperation(Equation4.1and4.2),thatis:
P
w c]1
+
S
2
(i
q
2
+i
d
2
)r
s
= K
opt
mSw
3
(4.4)
resultingthefollowingfunctionforthedelimitingelectricalangularspeed:
44
m Sw
=
_
P
w c]1
+
S
2
(i
q
2
+i
d
2
)r
s
K
opt
3
(4.5)
Figure22showstheoperationpointsofbothslidingsurfaces.Thedecisiontoworkin
one mode of operation or the other can be taken by comparing the measured speed with
theboundaryspeed
m Sw
asfollow:
m

m Sw
= First moJc o opcrotion
m
<
m Sw
= ScconJ moJc o opcrotion
Figure22OperationPointsofbothslidingsurfaces[54].
a) SufficientWindRegime
On this mode of operation, the main objective of the wind subsystem is to regulate
thepowerproducedbythewindturbinefollowing P
w c]1
.
To accomplish this objective, the sliding surface b
1
was written in terms of currents
onthebatterybank.
45
b
1
= P
w c]1
-P
w
= I
b
( I
bc]
+i
L
-i
w
) (4.6)
TheswitchedcontrolsignalproducesanunwanteddeviationinTe,andahighripple
on the current of the DC/DC converter. To avoid this problem, following [26], we added an
integrator to the input u
x
, channel turning u
x
into a new state variable and the integrator
input signal w becomes the new input to the system. In consequence, the algebraic
dependencebetweentheinputofthesystemandtheslidingsurfaceisbroken.
The dynamic model of the extended system is shown in Equation 4.7 [26] and the
blockdiagramispresentedinFigure23.
Figure23Blockdiagramofcontrolledwindsubsystem
x = _
c
u
x
_ = _
1
(x)
2
(x)
_ +_
g
1
(x)
g
2
(x)
_ w (4.7)
_
c
u
x
_ = _
P
2[
(I
t
-I
cIn
)
u
_ +j
u
1
[ w
where the electrical torque of PMSG is given by I
cIn
= K(
c
-
cIm
)and the switched
controllawisgivenby:
46
w =
`
1
1
1
1
w
1
+
=
-|
1
(x)|
4
_
-18I
2
m
2
u
x
c
+2nSI
2
u
x
2
m
:
b
S
m
R
s
+9
m
2
R
s
2
-S6C +1uB -2A
+1_
2
i b
1
u
w
1
-
=
|
1
(x)|
4
_
18I
2
m
2
u
x
c
+2nSI
2
u
x
2
m
:
b
S
m
R
s
+9
m
2
R
s
2
-S6C +1uB -2A
+1_
2
i b
1
< u
(4.8)
where
A = n
2
I
2
:
b
2
u
x
2
B = nSI
2
m
:
b
c
u
x
C = I
2
m
2
c
2
= 27
m
2
R
s
2
-1u8C +24B -4A
b) InsufficientWindRegime
On this mode, the system is not able to operate on the sliding surface h
1
, and it is
necessary to incorporate a secondary sliding surface h
2
whose main objective is to extract
themaximumpoweravailablefromthewindturbine.
b
2
= P
w
-P
w c]2
(4.9)
andtheswitchinginputsignalisgivenby
w = _
w
2
+
> u i b
2
u
w
2
-
< u i b
2
< u
(4.10)
4.1.1.3 SimulationResults
Inordertotesttheoperationofwindsubsystem,twoscenariosweresimulatedand
evaluated.Overthefirst25seconds,thereisasufficientwindforthewindturbinetowork
well(theminimumwindspeedrecommendedforthisturbineis10m/s),between30and38
47
secondsthewindspeedwasreducedtosimulateinsufficientwindpowermode(SeeFigure
24).Forallscenarios,theloadisavariablecurrentsource.
Figure24WindSpeed
Figure 25 depicts the power coefficient of the wind turbine. If the wind speed was
reduced under the operation limit of wind turbine, the power coefficient decreases to zero
for a few seconds until the wind speed increases again, in other words, the rotation of the
rotorwouldstopbecausethewindspeedattherotorfrontwouldreducetozero.
Figure25PowerCoefficientofwindturbine
Figure 26a shows that the power reference of the wind turbine to follow the first
generationmode,andFig.26btofollowthesecondgenerationmode.Figure27presentsthe
decisioncriteriatochangethemodeofoperation;iftheangularshaftspeedoftheboundary
48
point
m Sw
ishigherthantheangularshaftspeed
m
,themodeofoperationismode2or
insufficientwindgeneration;else,themodeofoperationismode1.
Figure26PowerReferenceofwindsubsystem.a)Mode1b)Mode2
Figure27AngularShaftSpeed
49
Figure28showstheslidingsurfacesofbothmodesofoperationofwindsubsystem;
Fig28apresentstheevolutionoftheslidingsurfaceh1,andFig.28bshowstheevolutionof
h2;theslidingsurfacesmustbezerofortheirrespectivemodeofoperation.
Figure28Slidingsurfacesofwindturbine.a)1
st
modeb)2
nd
modeofgeneration.
Figure 29 shows the control signals for the wind subsystem; Fig. 29a presents the
signal u
x
, Fig 29b shows the switching signal w
1
;
due to the design based on passivity

considerations, exist the variation of the switching control values w
1
+
and w
1
,. Fig 29c
presents the switching signal w
2
; the reaching time depends on the tuning of the switching
controlvaluesw
2
+
andw
2
.FinallyFig29ddepictsthetotalcontrolsignalw.Itisimportant
to mention that such minimum effort switching control law result in a chattering reduction
oftheoutputcurrentoftheDC/DCconverter.
50
Figure29Controlsignalsofwindsubsystem
Finally, Figure 30 depicts the currents present on the DC bus. Fig. 30a presents the
windcurrent,wecouldobservewhenthereisnotenoughpowerfromthewindsource,and
the wind turbine cant follow the setpoint. Fig. 30b shows the PV current, it was set to a
constant current of 10A, on this case, the wind subsystem is complementing the
51
photovoltaicgeneration.Fig.30cshowstheloadcurrent,whichwhenaddedtothebattery
reference (20A) makes the total demand. Fig. 30d illustrates the current on the battery
bank, we could observe when there is enough power to supply the total demand it is on
recharge cycle and set the current to 20A, but when the power is insufficient, it supplies
powertotheload.Thishappensbetween22to38secondsandbetween70to83seconds.
Figure30DCbusCurrents(Windsubsystem)
52
4.1.2 PowerControlofaPhotovoltaicArrayusingSMC
The solar subsystem acts under supervisor control commands using whichever
regulating power or tracking the PV to maximum power operation point. This subsystem
starts its operation when the wind power generation is not enough to satisfy the total
demand. This subsystem is functioning on all modes except mode 1, because it is
establishedassecondaryenergysource.
4.1.2.1 SlidingModeControllerDesign
The main objective of SMC is to control the power generated by the PV array to
satisfy the total power demand (The load and the Battery Bank). For an efficient recharge
and to maximize the battery life, the recovery cycle is expected to be done under a given
constantcurrent I
bc]
.
ForthePVarray,wesettwomodesofoperationfollowing[15].Thefirstoccurwhen
thePVarrayiscapableofgeneratingenoughpowertosatisfythetotalpowerdemand.
P
c]
= :
b
(i
L
+I
bc]
-i
w
)
(4.11)
ThesecondmodeofoperationoccurswhenthePanelisunabletogenerateenough
power to supply the power reference. In this mode, the PV operates on the Maximum
PowerOperationPoint(MPOP).
53
InMPOP,theinstantaneousandincrementalconductanceofthearrayhasdifferent
signandthesameabsolutevalue.Thisconditionarisesfrom[50]:
oP
p
oI
= i
p
-
Ii
p
- II
mux
b(yo +1 -y)(I
mux
+
(4.12)

ThepreviousequationisthederivativeofEquation3.11withrespecttothevoltage.
Taking the derivatives of equation 3.10 we obtained the dynamic equations for Current of
thePVarray[50].
oi
p
oI
=
i
p
- I
mux
b(yo +1 -y)(I
mux
+
(4.13)

The decision to work in one mode of operation or the other is taken by comparing
the power that the array would generate if operating at the MPOP and P
c]
. This is
illustrated on Figure 31, and could be expressed as a fictitious power P
]
= -
I
p
v
p
I
p
2
in the
followingway[15]:
P
]
P
c]
= First moJc o opcrotion
P
]
< P
c]
= ScconJ moJc o opcrotion
a) SufficientPowerGenerationConditions
Inthismode,thecelltemperatureandtheisolationaresufficienttosatisfy P
c]
.To
accomplishthisobjective,theslidingsurfacewaswrittenintermsofcurrentsonthebattery
bank[15].
b
1
= i
L
+I
bc]
-i
w
-i
o
(4.14)
54
Figure31Zonesofoperationforcontrolmodesundersufficientpowergenerationconditions[15].
andtheswitchcontrolsignalis:
u = _
1 i b
1
u
u ib
1
< u
(4.15)
b) InsufficientPowerGenerationconditions
The control objective is to operate the PV cell at the MPOP because the cell
temperatureandtheisolationareinsufficienttosatisfy P
c]
.InordertosatisfytheEquation
4.12,theslidingsurfacemustbe[15]:
b
2
=
oI
p
oI
p
+
I
p
I
p
(4.16)
andtheswitchedcontrolsignalmustbe:
u = _
u i b
2
u
1 ib
2
< u
(4.17)
inthiscase,thebatterybankrechargingismadeatI
b
insteadof I
bc]
.
55
4.1.2.3 ThecontrolLaw
The final control law comprises the two modes of operation that could be
summarizedasfollow[15]:
u =
`
1
1
1
1
1
1
i
I
p
v
p
I
p
2
P
c]
tbcn _
1 i b
1
u
u ib
1
< u
i
I
p
v
p
I
p
2
< P
c]
tbcn _
u i b
2
u
1 ib
2
< u
(4.18)
withb
1
= i
L
+I
bc]
-i
w
-i
o
and b
2
=
I
p
v
p
+
I
p
v
p
.
The performance of the PV controller was evaluated through computer simulations.
Figure 32 shows the fictitious power against the power reference. If the fictitious power is
lowerorequaltoreferencepower,thePVcontrolleractsinmode2elseitactsinmode1.
Figure32Powersignalstodecidemodeofoperationofsolarsubsystem.
56
Figure 33a depicts the sliding surfaces h1 for the first mode of generation, and Fig.
33bshowsh2forthesecondmodeofgeneration.Wecouldobservethemodeofoperation
bylookingattheslidingsurfaces.
Figure33Slidingsurfacesofsolarsubsystem.
Figure34showsthecontrolsignalsofthePVarray.Fig34ashowsthecontrolsignal
u1 for the PV array when it is under first mode of generation and Fig 34b shows control
signal is u2 for the second mode of operation. Fig 34c shows the signal u corresponds to
totalcontrolactionofPVarray(u1+u2).Thereachingmodeoperates,keepingtheswitched
inapermanentposition,closedoropen,sothatthefastestrateofconvergenceisassured.
57
Figure34Controlsignalsofsolarsubsystem.
Finally, Figure 35 depicts the currents present on the DC bus. Fig. 35a presents the
wind current fixed at 20A, on this case, the solar subsystem is complementing the wind
generation.Fig35bshowsthePVcurrent,wecouldobservethesecondmodeofoperation
because the reference signal is higher than the generated signal, and it means that is not
enough power from the PV array to satisfy the power demand. Fig 35c shows the load
current, which adding the battery reference (20A) makes the total demand. Fig 35d
illustrates the current on the battery bank, we could observe that when there is enough
power to supply the total demand, it is on recharge cycle and set the current to 20A, but
whenthepowerisinsufficientitsuppliespowertotheload.
4.1.3 ControlStrategyfortheBatteryBankandtheGrid
Themainobjectiveofthiscontrolunitistomakethedecisiontocharge,ordischarge
thebatterybank,andtomanagethecurrentprovidedbythegrid.
58
Figure35DCbuscurrents
Here, the reference bank current I

bc]
could take two values. If the battery bank
current I
b
ispositive,itchargesthebatterybankand I
bc]
issettoapositivevalue.Ifthe
batterybankisfullycharged, I
bc]
mustbereducedtozero[2].
When the battery bank current is negative, it means that the battery bank is
supplyingenergy.IftheenergyavailableontheBBisinsufficient,theGridisabletosupply
thepowerneededbytheload[2].
59
4.1.3.1 ControllerDesign
First, we have to define the magnitude of the charge and discharge currents; they
always have to be less or equal to these values. Second, the State of Charge (SOC) of the
BatteryBankissupposedtoalwaysstaywithindefinedupperandlowerlimits.Theselimits
aregivenby:
I
bc]+
= Hoximun Cborging Currcnt
I
bc]-
= Hoximun iscborging Currcnt

mux
= uppcr Iimit o S0C

mn
= Iowcr Iimit o S0C
For the Battery Bank, we set two modes of operation. The first occurs when the
batterycurrentispositive(chargingmode),andthesecondoccurswhenthebatterycurrent
isnegative(dischargingmode).
a) I
b
ispositive
In this mode, the power generated by the renewable energy system is higher than
thepowerrequiredbytheload,sotheextrapowerishandledasfollow[2]:
The extra power will be stored in the Battery Bank as long as SOC is less than

mux
and I
b
isasmalleramountthan I
bc]+
.
60
If I
b
> I
bc]+
,thentheBatterybankischargedby I
bc]+
,andtheextrapoweris
reducedfromthePVandwindsubsystemsthroughtheindividualcontrolunits.
If S0C =
mux
, means that the battery bank is fully charged and the extra
powermustbereducedfromthePVandWindsubsystems,i.e.thereferencefor
thecontrollersissettozero.
b) I
b
isnegative
Inthiscase,theloaddemandishigherthanthepoweravailablefromtheWindand
PV subsystems, so the battery bank and the Grid must supply the extra power needed.
Depending upon the SOC of the Battery Bank and the magnitude of I
b
, supplying the extra
demandiscompleteaccordingtothenexthierarchicalorder[2]:
OnlythroughtheBatteryBankaslongastheseconditionsaresatisfied:
S0C >
mn
(4.19)
|I
b
| I
bc]-
Through the Battery Bank and the Grid: this occurs when the load demand is
higher than the battery discharge limitations, so the Grid is able to provide the
extracurrent.Theconditionsthathastobesatisfiedare:
S0C >
mn
(4.20)
|I
b
| > I
bc]-
61
Only through the Grid: if the battery bank is completely discharged, this
is S0C =
mn
, the Grid must supply the extra power needed. The strategy
implemented is called slopeon strategy (see Figure 36), in which the grid start
level occurs when the SOC is equal to a level higher than Q
min
, defined by the
battery; when that level Q
min
*
X is reached, the grid starts taking the load
progressively from the battery. The grid current starts increasing with the same
slopesothatthetotalcurrentcontributionfromincooperationbatterybankand
thegridmaking I
b
= u[2].
Figure36Implementationofslopeonstrategyadaptedfrom[2].
62
The performance of the Battery Bank and the Grid was evaluated through Matlab
simulations.Forthisresults,theupperandlowerlimitofSOCare:

mux
= 1

mn
= -1
But,wearegoingtoconsiderthatthebatteryisfullychargedat76%ofSOCandthe
DieselStartPointisat0.76%ofSOC,thatis:

mux
= u.761S

mn
- X = -u.761S
Themaximumchargingcurrent,andthemaximumdischargecurrentwassettothese
values:
I
bc]+
= 2uA
I
bc]-
= -2uA
Figure37(a)showstheBatteryBankcurrent,Fig.37bshowstheGridcurrentandFig.
37cshowstheStateofChargeoftheBatteryBank.Onthisexample,theBatterywastotally
discharged, and the Grid supplies the extra power demand; on recharge cycle, the battery
was recharged until 20A. At 3.8, 8.5, and 13 seconds, the battery bank starts to discharge;
the grid is able to supply energy at 4, 8.7, and 13.5 seconds because the battery bank
63
reachesthemaximumdischargecurrent.At0,5.8,and10.1seconds,thegridsupplyenergy
becausetheSOCis
mn
- Xorlower.
Figure37BatteryBankandGridcurrentswhenthebatteryisdischarged.
Figure38ashowstheBatteryBankcurrent,Fig.38btheGridcurrentandFig.38cthe
StateofChargeoftheBatteryBank.Onthisexample,theBatterywasfullycharged,andthe
grid suppies energy only when the battery bank reaches the maximum discharge current.
Between 1.6 to 3.5 seconds I
b
=0 and the renewable generators only produce the power
necessary to satisfy the load demand (Battery is not a part of the total demand), when the
SOCislowerthanQ
max
,thebatteryisrechargedat20A.
64
Figure38BatteryBankandGridcurrentswhenthebatterybankisfullycharged.
4.1.4 OnOffControlfortheLoad
This control was developed by us to connect or disconnect a particular load or zone
from the total power demand. The main reason to do that is to protect the generation
sourceswhenthereisinsufficientgenerationcapacitytomeetthedemand.
If the total demand is higher than the total generated power, the loads will be
disconnectedonasequenceorderdependingofsomeprioritiesestablishedapriori.
4.1.4.1 Thecontrollaw
Inoursimulation,thelowestpriorityloadiszonethree,andthehighestiszoneone
(zoneonealwaysneedstostayconnected).
65
0
mux
is the maximum current generated by the Grid. The Load currents for each
zonearei
L1
, i
L2
, and i
L3
.
Thecriteriatodisconnectloadthreeisdescribedasfollow:
i( i
L1
+i
L2
+i
L3
) -i
gcn
> 0
mux

IooJ
3
= 0
clsc
IooJ
3
= 0n
Thecriteriatodisconnectloadtwoisdescribedasfollow:
i( i
L1
+i
L2
) -i
gcn
> 0
mux

IooJ
2
= 0
clsc
IooJ
2
= 0n
i
L
-i
gcn
< 0
mux
In this example, the grid only could generate 20A, so if the total demand is higher
thanthemaximumcurrentgenerated,theloadzoneswillbedisconnected.
Figure 39 shows the control signals for each load zone; Fig 39a is always zero,
because the first zone is on all the time; Fig 39b depicts the onoff signal for zone two, the
66
signal is set to one only if the third zone is already off. Fig 39c presents the onoff control
signaltothirdloadzone,ifthesignalissettoone,theloadisturnedoff.
Figure39Controlsignalforloadzones.
Figure 40a shows the Grid current; for this example, the grid was turned on all the
time.Fig.40bdepictsthetotaldemand(withoutcontroltechniques),andthetotalcurrent
demandpresentonthesystem.Fig.40cpresentsthecurrentdemandedforeachzone.
4.2 SupervisorControlStrategy
Thesystemiscomposedbyawindsubsystem,asolarsubsystem,aconnectiontote
grid, a battery bank, and loads as described before. The main objective of the supervisor
controlistosatisfytheloadpowerdemandandtomaintainthestateofcharge(SOC)ofthe
batterybank.
67
Figure40GridandLoadcurrents.
The supervisor control determines the operation mode of each generation
subsystem. Fundamentally, these operation modes are determined by the energy balance
betweenthetotaldemand(loadandbatterybank)andthetotalgeneration(wind,solar,and
grid).
68
To design the supervisory controller, we select the wind subsystem as the main
generator role; the complementary roles are in charge of solar subsystem and the grid
respectively.
The complete diagram of HPS including the interaction with the supervisory
controller is presented in Figure 41. The supervisor inputs are measure variables as the
currentsandvoltagesoutputsofthesubsystemsintheHPSandtheSOCofthebatterybank.
Thesupervisoroutputsarethesignalstoactivateordeactivateeachsubsystem.Thegeneral
schemeforinputsandoutputsisshowninFigure42.
Figure41Supervisorycontrollerinteractionmodel.
69
Figure42Schemeforinputsandoutputsinthesupervisorcontrol.
4.2.1 ModesofGeneration
Toaccomplishtheobjective,weselectfivemodesofoperationasfollow.Themain
idea of the first three modes was proposed in [1], the latest two modes were added to
accomplishthesupervisortechniqueofHPS,andwasproposedin[2].Mode3wasmodified
tomatchthebatterybankcontrolstrategywiththesupervisorcontrol.Mode4andmode5
were modified because they were proposed for Diesel generators, but we changed it by
connectiontothegrid.
4.2.1.1 Supervisorycontroller:Mode1
Mode one occurs when the wind generation is enough to satisfy the total demand.
Thesolarsubsystemandthegridareinactiveevenasthebatterybankisinrechargemode
at maximum charging current if it is discharged or at zero current if it is fully charged, and
70
the wind subsystem is set to power regulation. This mode is running until the maximum
availablewindpowerisexceedbythetotalpowerdemand[1].
HoJc 1: i {
m

mSw

`
1
1
1
1
1
1
1
1winJ Subsystcm - _
Powcr Rcgulotion
P
w c]1
= I
b
(i
L
+I
bc]
)

Solor Subsystcm - Inocti:c
Bottcry Bonk - _
Rccborgc Cyclc
I
brc+

0riJ - Inocti:c
Zonc IooJ - Inocti:c

4.2.1.2 SupervisoryController:Mode2
Onthismode,thewindgenerationisnotenoughtosatisfythetotaldemand,anditis
set by the supervisor control to operate at the point of maximum energy conversion. The
solar subsystem is set to follow a power reference required to complement the wind
subsystem and together satisfy the total power demand. The battery bank is part of the
total power demand, because is in recharge cycle, and its current is always the battery
referencecurrent[1].
HoJc 2: i _

m
<
mSw
-
oI
p
oI
p
I
p
2
P
c]

`
1
1
1
1
1
1
1
1
1
1
winJ Subsystcm - _
Hoximum Powcr Con:crsion
P
w c]2
= K
opt
m
3
-
S
2
(i
q
2
+i
d
2
)r
s

Solor Subsystcm - _
Powcr Rcgulotion
P
S c]
= :
b
(i
L
+I
bc]
-i
w
)

Bottcry Bonk - _
Rccborgc Cyclc
I
brc+

0riJ - Inocti:c

On this mode, the wind and the solar subsystems are set to operate at their
maximumenergyconversionpoints,andthebatterybankissettosupplypowertotheload
instead to receive energy. This mode is maintained as long as the state of charge of the
71
battery is greater than a minimum required [1] or the battery current is higher than the
maximumdischargingcurrent[2].
HoJc S: i
`
1
1
1
1

m
<
mSw
-
oI
p
oI
p
I
p
2
< P
c]
|I
b
| I
brc-
i
I
<i
gcn

`
1
1
1
1
1
1
1
1
1
1
winJ Subsystcm - _
P
w c]2
= K
opt
m
3
-
S
2
(i
q
2
+i
d
2
)r
s

Solor Subsystcm - _
HP0P trocking
oI
p
oI
p
+
I
p
I
p
= u

Bottcry Bonk - Ablc to supply powcr to tbc looJ
0riJ - Inocti:c

In this case, the load demand is higher than the power available from renewable
sources,andthebatterybankcurrentisequalorhigherthanmaximumdischargingcurrent.
Thesupervisorcontrolturnsonthegridtoprovidetheextracurrent[2].Onthismode,the
batterycurrentisalwaysthemaximumdischargingcurrent.
On this mode, the OnOff control for the load may be activated if the total current
demandishighestthanthetotalcurrentgeneratedbytherenewableenergyandthegrid.
HoJc 4: i
`
1
1
1
1
1
1

m
<
mSw
-
oI
p
oI
p
I
p
2
< P
c]
|I
b
| > I
brc-
S0C >
mn
i
I
>i
gcn

`
1
1
1
1
1
1
1
1
1
1
winJ Subsystcm - _
P
w c]2
= K
opt
m
3
-
S
2
(i
q
2
+i
d
2
)r
s

Solor Subsystcm - _
HP0P trocking
oI
p
oI
p
+
I
p
I
p
= u

Bottcry Bonk - _
Supplying powcr to tbc looJ
I
brc-

0riJ - Ablc to suply powcr to tbc looJ
Zonc IooJ - Acti:c
On this mode, the total demand is more than the power available from wind and
solarsubsystems,andthebatterybankisdischarged.Thesupervisorcontrolallowsthegrid
72
tostartstakingtheloadgraduallyfromthebatterytosupplythepowerdemand[2].Onthis
mode,thebatterycurrentisalwayszero.
The OnOff control for the load may be activated if the total current demand is
highestthanthetotalcurrentgeneratedbythesystem.
HoJc S: i
`
1
1
1
1

m
<
mSw
-
oI
p
oI
p
I
p
2
< P
c]
S0C
mn
i
I
>i
gcn

`
1
1
1
1
1
1
1
1
1
1
winJ Subsystcm - _
P
w c]2
= K
opt
m
3
-
S
2
(i
q
2
+i
d
2
)r
s

Solor Subsystcm - _
HP0P trocking
oI
p
oI
p
+
I
p
I
p
= u

Bottcry Bonk - _
Complctcly iscborgcJ
I
b
= u

0riJ - Acti:c
Zonc IooJ - Acti:c

4.2.2 OperationStrategy
TheoperationstrategyforthesupervisorcontroloftheHPSisshowninFigure43as
astatetransitiondiagramforeachmodeofthesupervisorycontroller.
Figure43Modetransitioncriteriaforthesupervisorcontrol.
The supervisory controller can shed some loads if the total power demand is higher
than what the overall system could supply. This strategy was not included as an operation
mode because it depends basically of the load demand and not on the availability of the
generation sources. Figure 44 shows the transition criteria to apply shedding load. This
criteri
Matla
Q
-0
Qm
0.7
Discharg
SO
2
Charge B
1
ion occurs o
b/Simulinkd
modos 1
Subsystem
Q
m
i
n
min
.755
max
7613
Batte
Ibref
Qma
SOC
ge Battery
3
OC
2
Battery
1
only if the sy
diagramfor
Figure45
c
h
a
r
g
e

B
a
t
t
e
r
y
M
o
d
o
ery supervisor
+
ax Ibref
ystem is ope
thesupervis
Figure44
5Matlab/Simu
i
iw
io
il
SO
Qm
Ibr
S
SO
Qm
Dis
i_Load
i_Wind
i_Wind
vb
i_Load
i_Load
i_solar
73
erating in m
sorycontrol
Loadshedding
linkModelofs
ac
iwref 1
il
Iwref
iwneed
iwind
iwin
i
DG ref
OC
min
ref -
DG ref
DG
Subsystem1
OC
min
scharge Batter
DGmax
Pref sol
vb
iL
iw
Ibref
Pref
ipvref
mode 4 or m
ler.
gscheme
supervisoryco
tivate Grid
3
dref
ndref
ipvref
ipvref
DGref
Subsystem3
DGmax
Off 1
Off 2
Off 3
Psref
Psref
vpv
Psref
ipv
i_Wind
i_solar
W
W
ellTemp
ode 5. Figu
ontroller.
igen
iw
io
i ge
PV decision
ipv
vpv
T
Psref
sw
Offf 3
Off 3
Offf 2
Off 2
Off 1
Off 1
Mode S
wm
WmS
ac
Wm
WmSW
ure 45 show
activate
4
activate PV 2
en igen
elector Wind
SW
ctivate wind
ws the
e wind
4
74
4.2.3 SimulationResults
This section presents the results of the full system including all modes of operation,
all Individual Control Units, and the Supervisor Control. Two scenarios were evaluated to
represent all the possible situations on the system behavior. First scenario was simulated
with the battery bank fully charged and the second scenario with the battery bank totally
discharged.
Both scenarios were simulated with the same parameters and perturbations to
compare the performance of each subsystem and to evaluate the behavior of Supervisor
Control.
Figure46showsa)thewindspeed,withvariationsbetween1012m/sec,andb)the
celltemperatureofPVarraywhichvariesbetween3050C.Thethreezonesoftheload
wereselectedtoforcetheHPSandtheSupervisorControltooperateinallpossiblemodes.
Figure46a)Windspeedandb)Celltemperaturevariations
75
4.2.3.1 BatteryBankFullyCharged
Thiscorrespondstothefirstscenariotoevaluate.Theinitialconditionofthebattery
bankisfullycharged(SOC=0.7616);theupperlimittoconsiderateafullychargedbatteryis
0.7613, so when the SOC of the battery is higher than this value, the battery reference
currentfortherenewablesourcetochargethebatteryiszero.
Figure 47 shows a) the battery bank voltage and b) the SOC of the battery bank, on
this graph we could observe the battery fully charged before four seconds, after this time,
the battery still charged, but not at its maximum capacity. If the slope is positive, means
thatthebatterybankisonchargingmode,alsoitisondischargingmode.
Figure 48 depicts the current of each load zone. As we said before, the highest
priority zone is the zone 1. The interruption on the current means that the corresponding
zonewasdisconnected.
Figure47BatteryVoltageandSOC(BatteryBankFullyCharged).
76
Figure48Loadcurrentdividedbyzones(BatteryBankFullyCharged).
Figure 49 shows a) the wind turbine and b) the Photovoltaic panel control signals.
WecouldobservethatPVarrayisnotcontrolledwhentheHPSisinmode1.Theindividual
control signals and sliding surfaces of solar subsystem and wind subsystem are shown in
Figure50andFigure51respectively.
Figure49ControlsignalsofWindTurbineandPVarray(BatteryBankFullyCharged)
77
Figure50ControlsignalsandslidingsurfacesofPVarray
Figure51ControlsignalsandslidingsurfacesofWindTurbine
Figure52showstheDCbuscurrentofHPS.Fig.52apresentsthewindcurrent(Set
Pointandgeneratedcurrent),attime1.2to6.8,13.8to35,and60to81seconds.Itcanbe
78
observedthatthereferencesignalisequaltogeneratedsignal.Thismeansthatthesystem
is operating at mode 1 as shown in Fig. 52f; the maximum current generated by the wind
subsystemis75A.Fig.52bpresentsthesolarcurrent.Whenthereferencesignalisequalto
a generated signal and both signals are different from zero, it means that the system is
operating at mode 2 else, system operates at mode 3 as shown in Fig. 52f. The maximum
currentproducedbythepanelis17A.Fig.52cshowsthebatterybankcurrent.Becausethe
battery initial condition was fully charged, when there are sufficient generation conditions,
thebatteryreferencecurrentissettozero,alsoissetto20A.Ifthebatterycurrentdoesnot
exceedthemaximumdischargingcurrent(20A),itmeansthatthesystemisworkingwithin
thethreefirstmodes,elseitisworkingonmode4asshowningraphsix.Fig.52dshowsthe
grid current. This signal is activated only if the maximum discharge current of the battery
bank was exceeded. This happens under this scenario because the battery bank is fully
charged. The maximum current that the grid could produce is 25A, so when the grid
referenceishigherthanthisvalue,theonoffcontrolmadebysupervisorcontrolisactivated
to turn off one zone of the total load demand as shown in Fig. 52e. Fig. 52f presents the
operation modes of supervisor control. On this scenario, mode five is impossible because
thebatterybankisfullycharged.
79
Figure52DCbuscurrentsandoperationmodes(BatteryBankFullyCharged).
80
4.2.3.2 BatteryBankTotallyDischarged
This corresponds to the second scenario to evaluate. The initial condition of the
battery bank is totally discharged (SOC = 0.755); the limit to allow the grid generation
becausethebatteryisconsidereddischargedis0.755,sothebatteryreferencecurrentfor
the renewable source to charge the battery bank is the maximum charging current, in this
case20A.
Figure 53 shows the battery bank voltage and the SOC of the battery bank. On this
graph we could observe the battery totally discharged. If the slope is positive, means that
thebatterybankisonchargingmode,alsoitisondischargingmode.Iftheslopeisconstant,
means that SOC is at lower limit so all the extra power needed by the load is given by the
Grid,andthebatterycurrentiszero.
Figure53BatteryvoltageandSOC(BatteryBankdischarged)
81
Figure 54 depicts the current of each load zone. As we said before, the highest
priorityzoneiszone1.Theinterruptionsonthecurrentsignalmeansthattherespectively
zone was disconnected. Figure 55 shows the wind turbine and the Photovoltaic panel
controlsignals.
Figure54Loadcurrentdividedbyzones(BatteryBankDischarged).
Figure55ControlsignalsofWindTurbineandPVarray(BatteryBankDischarged)
82
The individual control signals and sliding surfaces of solar subsystem and wind
subsystemareshowninFigure56andFigure57,respectively.
Figure56Controlsignalandslidingsurfacesofsolarsubsystem
Figure57Controlsignalandslidingsurfacesofwindsubsystem
83
Figure58showstheDCbuscurrentofHPS:Fig.58apresentsthewindcurrent(Set
Pointandgeneratedcurrent).Between2to6,16to22,24to31,and64to81seconds,we
can observe that the reference signal is equal to generated signal. As shown before, that
meansthatsystemisoperatingatmode1aspresentedingraphsix.Fig.58billustratesthe
solar current with the same behavior of the previous example. Fig. 58c shows the battery
bankcurrent.Becausetheinitialconditionwastotallydischarged,whenthereareisenough
generation from the renewable sources, the battery reference current is set to maximum
charge current (20A), and the system works under the first three modes. If the battery
current exceeds the maximum discharging current (20A), the system works on mode 4. If
theSOCofthebatteryisunderQ
min
,itmeansthatthesystemworksundermode5asshown
in Fig. 58f. Fig. 58d shows the grid current. This signal is activated only if the maximum
dischargecurrentofthebatterybankwasexceededortheSOCofthebatterybankischarge
to Q
min
. The maximum current that the grid could produce is 25A, so when the diesel
referenceishigherthanthisvalue,theloadsheddingcontrolisactivatedtoturnoffoneor
two zones of the total load as shown in Fig. 58e. Fig. 58f presents the operation modes of
supervisor control. This scenario activates mode five because this mode depends on the
lowerlimitofSOCofthebattery.
84
Figure58DCbuscurrentsandoperationmodes(BatteryBankDischarged).
85
4.2.3.3 Sufficientpowergeneration
In this example, we set a constant wind speed (12m/sec) and a constant cell
temperature (30 C) to simulate sufficient power regime for all system. The SOC of the
batterybankistotallydischargedasshowninFigure59.Thebehavioroftheoverallsystem
is similar to that described before. The main difference is that supervisor control moves
betweenthethreefirstmodes.Thegridonlygeneratespowerinthebeginningbecausethe
windgenerationwasnotenough.
Figure59SOCofBatteryBank(Sufficientpowerregime)
Figure 60 shows the DC bus currents, the order of the graphs are the same as the
previousexample,andhaveasimilarperformance.Figure61presentsthepowerofthePV
array. If fictitious power is lower or equal to the reference power means that system is
workinginmode3.Figure62showstheangularfrequencyofwindsubsystem,asdescribed
before, it acts like a decision criteria to work in mode 1 or turn on the PV array to work in
mode2or3.
86
Figure60DCbuscurrentofHPS(sufficientpowergeneration)
87
Figure61PowerreferenceandfictitiouspowerofPVarray(sufficientgenerationpower)
Figure62Windsubsystemdecisioncriteria(sufficientpowerregime)
88
Figure63andFigure64depictthePVarraycontrolsignalsforeachgenerationmode,
and their respective sliding surfaces. Figure 65 shows the control signals of renewable
sources.
Figure63PVarraycontrolsignals(sufficientpowergeneration)
Figure64PVarrayslidingsurfaces(sufficientpowerregime)
89
Figure65Renewableenergycontrolsignals(sufficientpowerregime)
4.2.3.4 ZonalEPDS
Inordertoevaluatethesupervisorycontrollerbehaviorwitharealload,wesimulate
thecontrollerwithazonalEPDS,whichistheDCpartoftheIntegratedPowerSystem(IPS)in
[52].
Here,thereweretwozones. Thefirstzonehasamotordrive,andthesecondzone
has a constant power load based on the classical buck converter. The power electronics
converters control the power flows among the sources, loads, and power buses, following
thecommandsofthePMC.
AftertheloadthereisaBoostconverterinordertoincreasetheoperationvoltageto
the load. On this example, we simulate two scenarios: a) the battery bank is totally
90
discharged,andb)thebatterybankisfullycharged.Figure66showsthebatteryvoltageand
SOCofthebatterybankforeachcase.
Figure66VoltageandSOCofBatteryBank(a)totallydischarged,(b)fullycharged
Figure 67 shows the control signals for wind and solar subsystems for each state of
thebatterybank.InFig.67athebatterybanktotallydischarged,andthewindcontrolsignal
is higher than Fig. 67b (battery bank fully charged), this means that in case a, the wind
subsystemhastogeneratemorepowertosatisfythetotaldemand.
Figure67Controlsignalsofwindandsolarsubsystemswiththebatterybank(a)totallydischarged,(b)fully
charged
91
Figure 68 shows the sliding surfaces of wind subsystem. The second mode of
operationonlywasactivatedifthebatterybankisdischarged,becausethereferencecurrent
forthewindturbineishigherthanthecaseb.TheangularshaftspeedgraphandFigure69
confirmsthisstatement.
Figure68Slidingsurfacesandangularspeedofwindsubsystemwithbatterybank(a)totallydischarged,(b)
fullycharged
Onotherhand,Figure70andFigure71showtheslidingsurfacesandcontrolsignals
forthesolarsubsystem,respectively.Onlyfortheleftsidecase,thesupervisorycontrolhas
toextractthemaximumpowerfromthePVsubsystem.
92
Figure69Controlsignalsofwindsubsystemwithbatterybank(a)totallydischarged,(b)fullycharged.
Figure70Slidingsurfacessolarsubsystemwithbatterybank(a)totallydischarged,(b)fullycharged
93
Figure71Controlsignalsofsolarsubsystemwithbatterybank(a)totallydischarged,(b)fullycharged.
Figure72showstheDCbuscurrentofHPS.Thisexampleworksbetweenmodesone
andthree,becausethetotaldemandissatisfiedbytherenewableenergy.Thebatterybank
does not supply energy to the load. In case a, the battery bank is in recharge mode at
maximum recharging current (20A). In case b, the battery bank is fully charged, so the
battery current is 0. The operation mode in case b is one, because the wind subsystem
generatelesspower(thereferenceislowerthancasea).
94
Figure72DCbuscurrentofHPS.TheBatteryBankis(a)totallydischarged,(b)fullycharged.
95
4.2.3.5 Changingsourcepriorityonsupervisorycontroller
In this example, the hierarchical structure of the operation modes of supervisory
controller was changed. The main generation role is in charge of the PV array and the
secondgenerationroleisinchargeofthewindsubsystem.Thesimulationsweremadewith
thesamesimulationparametersofsection4.2.3.2.
Figure 73 shows a) the battery voltage and b) SOC of the battery bank. Figure 74
shows a) the control signal u1 of PV array on first mode of generation, and b) the control
signal u2 to the second operation mode of PV array. Figure 75 shows the wind turbine
controlsignalsfora)firstmodeofgenerationb)secondmodeofgeneration,andc)thetotal
control signal w. Figure 76 shows the control signals for the renewable sources a) wind
turbine,andb)PVarray.
Figure73Batterybanksignalsa)Voltageb)StateofCharge
96
Figure74PVarraycontrolsignals
Figure75Windturbinecontrolsignals
97
Figure76Renewablesourcescontrolsignals
Figure 77a shows the wind current, which in this case is the complementary
generation source. Fig. 77b is the PV array current. In this case, it is the main generation
role, so always is operating at MPOP to satisfy the demand. Fig. 77c is the battery bank
current. The behavior is the same as described in previous examples. Fig. 77d is the grid
current.Fig.77eisthetotaldemandcurrent,andFig.77fshowstheoperationmodeofthe
supervisorycontroller.
98
Figure77DCbuscurrent
99
4.2.3.6 Changingsourcepriorityonsupervisorycontroller(4PVpanels)
In this simulation example, the hierarchical structure is the same as the one
presented in the previous section. The main difference here is that the simulations were
madewithfourpanelsinseriesandfourpanelsinparallelinorderforthePVtosatisfythe
loaddemandbyitself.
Figure 78 shows a) the battery voltage and b) SOC of the battery bank. Figure 79
shows a) the control signal u1 of PV array on first mode of generation; note that the solar
subsystemisoperatinginbothoperationmodes.Fig.78bshowsthecontrolsignalu2tothe
secondoperationmodeofPVarray.Figure80showsthewindturbinecontrolsignalsfora)
first mode of generation b) second mode of generation, and c) the total control signal w.
Figure 81 shows the control signals for the renewable sources a) wind turbine, and b) PV
array.
Figure78Batterybanksignalsa)Voltageb)StateofCharge
100
Figure79PVarraycontrolsignals
Figure80Windturbinecontrolsignals
101
Figure81Renewablesourcescontrolsignals
Figure 82a shows the wind current, which, in this case, is the complementary
generation source. Note that there is a time delay between when the wind turbine starts
deliveringenergytothesystemandwhenthegenerationcommandisgiven.Fig.82bisthe
PVarraycurrent.Inthiscase,itisthemaingenerationrole,soalwaysisoperatingatMPOP
to satisfy the demand. Fig. 82c is the battery bank current. Fig. 82d is the grid current.
Whiletheturbinestarts,thedeficiencyingenerationissuppliedbythebatteryandthegrid.
Fig. 82e is the total demand current, and Fig. 82f shows the operation mode for the
supervisorycontroller.
102
Figure82DCbuscurrent
103
4.3 Conclusions
ThischapterpresentedtheperformanceofaHPScontrolledbyaHierarchicalControl
capableofmanagingalltheenergyresourcesofthesystem.
The results presented on the previous section, demonstrated that the designed
controliscapabletodeterminingtheoperationmodeofeachsubsystempresentedonHPS.
The criteria, to make the decisions, were based on measurable system variables such as
shaftspeed,currents,SOC,andpower.
To accomplish the different control objectives (i.e., power regulation or maximum
power conversion) in both (wind and solar subsystems), we use sliding mode control laws,
fortheenergymanagementofthebatterybankweusedtheSOCofthebattery.
Inthesupervisorycontrolscheme,weassumedthatthemaingenerationrolewould
be carried out by the renewable energy, i.e., wind and solar subsystem respectively, while
thetraditionalgenerationplayacomplementaryrole.

104
Chaptei S
5. Control for a HPS using Model Predictive

Control
As described on previous section, the main objective of the HPS is to satisfy the
electricalloadsatwhileasmaximizingtheuseofrenewableenergysourcesatthesametime
thattheoperationofthebatterybankandthegridareoptimized.
The control strategy is the same as described on chapter 4. It consists of a
hierarchical two level structure: The main difference is that the PV individual control unit
was controlled using MPC techniques, while the wind subsystem was controlled using SMC
asshownbefore.Theoverallresultsareshowninsection5.3.
5.1 DynamicMatrixControl(DMC)
In 1979, Cutler and Ramaker developed DMC; this type of MPC used as prediction
modelthestepresponse[29].Thepredictorwillbe
y(t +k|t) = g
u(t +k -i) +(t +k)

k
=1

(5.1)
where, u(t)=u(t)u(t1), f(t+k) is the free response of the system, and G is the dynamic
matrixgivenby
105
0 =
l
l
l
l
l
l
g
0
u
g
1
g
0
u
u
. . .
g
N
c
-1
g
N
c
-2
.
g
N
p
-1
.
g
N
p
-2
g
0
.
g
N
p
-N
c1
1
1
1
1
1
(5.2)
Theobjectivefunctionandtheinputconstraintsaregivenby
[(N
p
, N
c
) = |y(t +k|t) -w(t +k)]
2
+ z(k)|u(t +k -1)]
2
N
c
k=1
N
p
k=1

(5.3)
subjectto
u
mn
u
k|t
u
mux
, t k
invectornotation,theobjectivefunctionisgivenby
[ = u
1
|0
1
0 +zI]u -2c
o
1
0u +c
o
1
c
o
(5.4)
whereuisthecontrolsignal,GistheDynamicMatrix,zisthecontrolweightingfactor,and
c
o
istheerrorbetweenthepredictiony,andthereferencetrajectoryw.
Iftherearenoconstraints,thecontrollawiscalculatedbyderivingthecostfunction
andsettingittozero.
o[
ou
= u
(5.5)
andthecontrollawisgivenby
(5.6)a
5.2
contro
hasto
linear
contro
differe
the s
Put in pra
andapplyit
PVCont
Thissectio
ollerispart
obedefined
We have t
, for that re
ollerasshow
The dynam
enceonthis
step respon
ctice, the re
totheproce
trollerDe
onwilldiscus
oftheopera
.
to define di
eason, the d
wninFigure
mic matrix G
sstrategyis
se of a no
egular meth
ess.Ifuviola
esignusin
ssthecontro
ationallevel,
fferent oper
different con
83.
Figure
G is built fr
thatthema
onlinear sys
106
od to use M
atesthecon
ngDMC
olstrategies
,andtheco
ration point
ntrol objecti
83Multiobjec
rom the ele
atrixGneeds
stem varies
MPC is to cal
nstraintsitis
sforthediffe
ntrolobject
ts because t
ives are con
ctivecontroller
ements of th
stobecalcu
s depending
lculate the c
settoitslim
erentoperat
iveforeach
he PV array
ntrolled by a
r.
he step res
ulatedsever
g on the o
control law
mits[32].
tionmodes.
operationm
y behavior is
a multi obje
ponse, the
raltimesbec
operating p
(5.6)
using
This
mode
s non
ective
main
cause
point.
107
Essentially, identical algorithm employed in linear systems can be in use, but G has to be
updated every time the process characteristics change. On this scheme, the prediction
matrixGissuppliedbythestepresponseselector.
The flow diagram of the constrained DMC algorithm implemented is presented on
Figure84.
Figure84DMCflowdiagramincludingconstraintsoncontrolsignal.
We set seven operation points, each one is part of a different control objective, to
developtheDMCcontrollerforthePVarray;thissetofpointswasselectedfromIVcurveof
thepanelshowninFigure3.
108
The power reference occurs when the PV array is capable of generating enough
powertosatisfythetotalpowerdemandandisgivenbyequation4.11,whilstthesetpoint
isgivenbythepowergeneratedbythePVpanelgivenbytheequation3.11.
5.2.1 SimulationResults
Figure 85 shows the duty cycle of the DC/DC converter for the PV array. When the
panel operates at maximum power conversion, the duty cycle is set to 0.8, because this
pointcorrespondtoMPOP,alowerdutycyclemeansthatthepanelisgoingtogenerateless
power.
Figure85DMCControlsignalofPVarray.
Figure 86 depicts the currents present on the DC bus. Fig. 86a presents a constant
wind current fixed at 20A, on this case, the solar subsystem is complementing the wind
generation. Fig. 86b shows the PV current. We could observe when there is not enough
powerfromPVarraytosatisfythepowerdemand,sothesystemworksonsecondmodeof
109
operation, also the PV follows the reference. Fig. 86c depicts the total demand (load
current,andbatteryreferencecurrent(20A)).Fig.86dillustratesthecurrentonthebattery
bank. We could observe when there is enough power to supply the total demand it is on
recharge cycle and set the current to 20A, but when the power is insufficient, it is able to
supplypowertotheload.
Figure86DCbuscurrents(PVarraywithDMC)
110
5.2.2 MPCPVagainstSMCPV
The natural behavior of SMC for DC/DC converters include the switching signal of
PWM, in other words, it is not necessary to include a PWM before the DC/DC converter.
Furthermore,MPCworkwiththemodel;ifthepredictedmodeldoesnothavethePWM,the
controllercantpredictthebehaviorofthesystem.
Onotherhand,wehavetoconsideratethePWMrestrictionsfortheMPCcontroller,
becausethecontrolsignalcouldnotbefasterthanswitchingfrequencyofPWM,sowehave
to limit the bandwidth for the control. In SMC the switching frequency is inherent to the
model.
Both examples (SMC and MPC) were simulatedunder the same circumstances. The
controlsignalofMPCPVhasnotaswitchingform;itsetstheaverageofthedutycycleofthe
DC/DCconverterasshowninFigure85.Ontheotherhand,theSMCPVcontrolsignalhasa
switchingfortogeneratethedutycycleoftheDC/DcconverterasshowninFigure34.
The behavior of MPC controller and SMC controller for the PV array is similar, the
main difference occurs at 3 seconds. The MPC controller for PV array does not follow the
referencecurrent.ThissituationispresentedbecausetheMPCcontrollerisnotanoptimal
control,andthetransitionstatesbetweenoperationpointsarenotbeenconsidered.
111
5.3 WindControllerusingDMC
We have to define different operation points because the wind turbine behavior is
the most non linear in the system, for that reason, the different control objectives are
controlledbyamultiobjectivecontrollerasdescribedbefore.
We try to do the same technique described in the previous section to control the
wind subsystem, but we had several problems to get the prediction matrix G because the
subsystemstepresponsehasatimedelaywithalotofnoisethatproduceabadbehaviorof
thecontrolsystem(seeFigure87).
Figure87WindTurbineStepResponse
The controller designed does not work appropriately, only works under some
operationpoints.
5.4 SupervisoryController
TheperformanceofthePVcontrollerwasevaluatedthroughcomputersimulationsin
two scenarios. The first scenario presents the PV panel controlled by a DMC with the
112
supervisorycontroller(windsubsystemwascontrolledbySMC),andtheBatteryBankisfully
charged.Thesecondscenariohasthebatterybanktotallydischarged.
5.4.1 SupervisorControlwithbatterybankfullycharged
TheinitialconditionoftheSOCofthebatterybankisSOC=0.7616,thismeansthat
the battery bank is fully charged, so the fifth operation mode of the supervisory controller
must never be activated. Figure 88 shows the battery bank voltage and the SOC of the
batterybank.
Figure88VoltageandSOCofBatteryBank.
Figure 89 shows the wind control signal with SMC techniques, and the solar control
signalusingDMCmethod.OnthisgraphwecanobservethatthePVcontrolsignalismore
softly,andthecontroleffortislowerthanthePVcontrolsignalwithSMCtechniques.
113
Figure89ControlsignalsPV(DMC)andWT(SMC).
Figure 90 depicts the currents present on the DC bus. Fig. 90a presents the wind
current. On this case, the wind subsystem is controlled by SMC controller. Fig. 90b shows
the PV current, the solar subsystem is complementing the wind generation and it is
controlled by a DMC controller. Fig. 90c illustrates the current on the battery bank. We
couldobservethatwhenthereisenoughpowertosupplythetotaldemand,andthebattery
bankisfullycharged,itisonrechargecycleandsetthecurrentto0Aor20A.Howeverwhen
the power is insufficient, it is able to supplies power to the load. Fig. 90d shows the grid
current. The grid supply power to the load only if the battery discharging current is higher
than 20A. Fig. 90e depicts the total demand (load current, and battery reference current
(20A)).Fig.90fshowstheoperationmodesofthesupervisorycontroller.
The main difference between PV controller described in section 4.2.3.1 and this, is
the PV array generated current using this control technique, the ripple on PV current is
avoided. The principal reason is that the duty cycle is sent to the DC DC converter in an
averageformandnotasaswitchingsignal.
114
Figure90DCBuscurrentofHPS(PVcontrolledbyDMC)
115
5.4.2 SupervisorControlwithBatteryBankTotallyDischarged
Thiscorrespondstothesecondscenariotobeevaluated.Theinitialconditionofthe
battery bank is totally discharged (SOC = 0.755); the battery reference current for the
renewablesourcetochargethebatterybankisthemaximumchargingcurrent,inthiscase
20A.
Figure91showsthebatterybankvoltageandtheSOCofthebatterybank.Noticeon
this graph that the battery is totally discharged. The battery behavior is the same as
describedinprevioussections.
Figure91VoltageandSOCoftheBatteryBank
Figure92showsthecontrolsignalofwindsubsystemusingSMCtechniquesandthe
dutycycleoftheDC/DCconverterforthePVarrayusingDMCtechniques.
116
Figure92ControlsignalsofHPS
Figure 93 depicts the currents present on the DC bus. Fig. 93a presents the wind
current.Inthiscase,thewindsubsystemiscontrolledbySMCcontroller.Fig.93bshowsthe
PV current, where the solar subsystem is complementing the wind generation and it is
controlled by a DMC controller. Fig. 93c illustrates the current on the battery bank. It is
observed that when there is enough power to supply the total demand it is on recharge
cycle and set the current to 20A, but when the power is insufficient, it is able to supply
powertotheload.Fig. 93dshowsthegridcurrent.Thegridsuppliespowertotheloadin
two cases: (i) if the battery discharging current is higher than 20A, or (ii) if the SOC of the
battery bank is under Q
min
. Fig. 93e depicts the total demand (load current, and battery
reference current (20A)). The sixth graph shows the operation modes of the supervisory
controller.
117
Figure93DCbuscurrentsofHPS
118
5.5 Conclusions
ThecontrolimplementedtoPVarraywithDMCisasuboptimalcontrol,becausethe
strategy proposed here is an adaptation of a linear method; over each operation point the
controlstrategyislinear,thetransientbetweenonepointtootherisnotconsidered.
The battery behavior does not change if we use DMC technique to control the PV
array, because the control laws are the same as described in section 4.1.3 so using one
techniqueorotherthebatterybankactsinthesameform.
On other hand, the DMC control is later than SMC for the same simulation, but the
controleffortislower.
We try to do the same control technique with wind subsystem, but we had several
problemstogetthepredictionmatrixbecausethesubsystemstepresponsehasatimedelay
withalotofnoisethatproduceabadbehaviorofthecontrolsystem.
To solve the optimization problem and to predict the process output in a nonlinear
MPC, elevated computational capacity is necessary. This makes the approach not practical
inanumberofapplications.

119
Chaptei 6
6. ConclusionsandFutureWork
6.1 Summaryofthework
OnthisworkwedevelopedaHierarchicalControloftwolevels.Thehighestlevelisa
supervisory control with five modes of operation and a load shedding scheme in case of
insufficient generation, in this case, we expanded the work of Valenciaga and Omari. The
lowestlevelshavelocalcontrolunitstocontrolthewindturbine,PVpanels,andthebattery
bankusingSMCandMPCtechniques.
6.2 Conclusions
AsimulationmodelforaHybridPowerSystemwasdevelopedtofacilitatethestudy
of the dynamics of HPS, and help in the development of control strategies to manage the
systempowerflowsunderdifferentgenerationandloadconditions.Thesimulationmodelis
modular in nature; each component is represented as a selfcontained module (Wind
Turbine,PVpanels,Grid,BatteryBank,andLoads).Thebalanceofenergyismaintainedby
the controller to keep the system stable. The subsystems could be connected to form a
completeHPS.
120
ThewindspeedandPVarraycelltemperaturescanbemodifiedtosimulatetheday
variations.TheBatteryBankwassimulatedintwostates(chargedordischarged).Theload
wasavariablecurrentsourcetoforcethesystemtooperateinallmodes.Differentcontrol
strategies(SMCandMPC)wereimplementedtoachievetheobjective.
The control strategy for the EPDS is a hierarchical structure with two levels; the
highest level is the supervisory controller which proved to be capable of handling and
synchronizing the operation of the subsystems that compose the HPS. The lower level is
compoundbythelocalcontrolunitsforeachsubsystem.
SlidingModeControlandModelPredictiveControltechniqueshavebeenconsidered
to accomplish the control objectives at the subsystem level, particularly wind and solar
subsystems. Nevertheless the applicability of the decision framework of the supervisor
controlisnotlimitedtothoseparticularcontrollaws.Incontrast,thistechniqueisgeneral,
and allows the inclusion of any other local control law appropriate to attain the
abovementionedcontrolobjectives.
Onthiswork,theprincipalgenerationrolewasinchargeofwindturbine,butthisis
notatallarigidrestrictionofthecontrolscheme.Thedecisionalgorithmofthesupervisor
control could be easily changed to set the solar subsystem as the main energy source and
the wind subsystem as the secondary. The traditional generation sources either a fuel
generatororACutilityisalwaysgoingtoplayasecondaryrole.
121
6.3 FutureWork
Futureworkcanbecommittedtowards:
Implementarobustmodelofwindturbine,becausethemodelusedherehadseveral
inconsistencies. The model that we used is a linear model of the nonlinear wind
turbineinordertoapplythemoderncontroldesigntheory.However,thefullmodel
was reduced under some practical assumptions, for example, a factor omitted from
the design modeling is the wind shear (the effect that rough terrain has on the
turbulenceonthewind)
To develop the dynamic models with multiple resolutions to study different time
scalesovertheHPS.
o Testing the supervisor control using a larger version of a zonal system to
represent the total demand. The supervisor control was tested with three ideal
variable current sources, and the average zonal system, which not forced the
supervisorytoworkinalloperationmodes.
IntroducingPowerQualityissuesaspartofthedecisionmakingprocessofsupervisor
control. The supervisory controller must required quality of the supply, in terms of
variations of voltage, frequency, and probability of loss of load. It is an essential
conditionforthechoiceofwhichgeneratorsshouldbeinservice.
122
o The load management (added or removed from the system) could be based on
frequencyvariations.
o TheloadmanagementmustbemoresophisticatedIncludecontrolovertheload
(variabledemand)bythesupervisor.
o Avoid starting utility grid or diesel generators in situations of fast variations of
power.
EvaluatestrategieswheretheHPSsuppliespowertothegridandnetmetering,and
includeitasanextraoperationmode.

123
References
1.Valenciaga,FandPuleston,PF."Supervisorcontrolforstandalonehybridpowersystem
using wind and photovoltaic energy". IEEE transaction Energy Conversion. Vol. 20. No2, pp.
398405,June2005.
2. Omari, Osama, Ortjohann, Egon and Morton, Danny. "An Online Control Strategy for DC
Coupled Hybrid Power Systems". IEEE Power Engineering Society General Meeting. pp. 18,
June2007.
3.DeLemosPereira,Alexandre."ModularSupervisoryControllerforHybridPowerSystems".
Phd.Thesis.RisoNationalLaboratory,Roskilde.UniversityofDenmark,June2000.
4.Roscia,MandZaninelli,D."Sustainabilityandqualitythroughsolarelectricenergy".10th
InternationalconferenceonHarmonicsandqualityofPower.Vol.2,pp.782782,2002.
5. Ross, M and Turcotte, D. "Bus configuration in Hybrid Systems". Hybrid info Semiannual
newsletteronphotovoltaichybridpowersystemsinCanada.7,pp.25,Summer2004.
6. Meinhardt, M and Cramer, G. "Past, Present and Future of grid connected Photovoltaic
and HybridPowerSystems". IEEE Power Engineering Society Summer Meeting. Vol. 2, pp.
12831288,July2000.
7.Chang,Liuchen."WindEnergyconversionSystems".CanadianReview.pp.1216,2002.
8.Ross,M,Turcotte,DandRoussin,S."ComparisonofAC,DC,andAC/DCbusconfigurations
forPVHybridSystems".SESCIConference.2005.
9.Wright,AlanD."ModernControlDesignforFlexibleWindTurbines".TechnicalReporton
NationalRenewableEnergyLaboratory(NREL).pp.1223,July2004.
10. Kraan, I. "Control Design for a Flexible Wind Turbine" (in Dutch). Deflt University of
Technology.1992.
11. Bongers, P.M. "Modeling and Identification of Flexible Wind Turbines and a
FactorizationalApproachtoRobust".Ph.DthesisDepartmentofMechanicalEngineeringand
MarineTechnology,DelftUniversityofTechnology.1994.
124
12. Freeman, J.B and Balas, M.J. "An Investigation of VariableSpeed HorizontalAxis Wind
Turbines Using Direct ModelReference Adaptive Control". Proceedings of the 37th AIAA
AerospaceSciencesMeetingandExhibit.pp.6676,1999.
13. Bossanyi, E.A. "Adaptive Pitch Control for a 250 kW Wind Turbine". Proceedings of the
9thBWEAWindEnergyConference.pp.8592,1987.
14. Wikipedia, the free encyclopedia. "Photovoltaics". (September 2007). [Online]
http://en.wikipedia.org/wiki/Photovoltaics.
15.Valenciaga,F,Puleston,P.FandBattaiotto,P.E."PowerControlofaPhotovoltaicarrayin
ahybridelectricgenerationsystemusingslidingmodetechniques".IEEProc.ControlTheory
Appl.Vol.148No6,pp.448455,November2001.
16. Xantrex. "Battery Banks for Inverter Systems". Smart Choice for Power. pp. 122. April
2006.
17. Uykan, Zekeriya. "Hierarchical Control and Multimedia". Helsinki University of
Technology, Control Engineering Laboratory. (September 2007). [Online]
http://www.control.hut.fi/Hyotyniemi/publications/97_report106/UYKAN/node1.html.
18. Henriksen, Lars C. "Model Predictive Control of a Wind Turbine". Ph.D Thesis. Technical
UniversityofDenmark.pp.1146.2007.
19. Utkin, V.I. "Variable Structured Systems with Sliding Modes". IEEE Transaction on
Automaticcontrol.1977,Vol.22.N2,pp.212222.April1977.
20. SiraRamirez, H. "Sliding Motions in Bilinear Switched Networks". IEEE Transaction on
CircuitsandSystems.August1987,Vol.34.N8,pp.919932.August1987.
21. SiraRamirez, H and Ilic, M. "A Geometric Approach to the Feedback Control of Switch
Mode DCtoDC Power Supplies". IEEE Transaction on Circuits and Systems. Vol. 35. N10,
pp.12911298.October1988.
22.Venkataramanen,R."SlidingModeControlofPowerConverters".Ph.DThesis.California
InstituteofTechnology.pp.1260.1986.
23. Young, K.D, Utkin, V.I and Ozguner, U. A control engineers guide to Sliding Mode
Control.IEEETransactiononControlSystemTechnology.Vol.7,pp.328342,May1999.
24. Vazquez, N, et al. "Sliding Mode Control for DC/DC Converters: A new Sliding Surface".
IEEEInternationalSysmposiumonIndustrialElectronics.Vol.1,pp.422426.June2003.
125
25. Ha, Q.P, et al. "Dynamic Output Feedback SlidingMode Control Using Pole Placement
and Linear Functional Observers". IEEE Transaction on Industrial Electronics. Vol. 50 No. 5.
October2003.
26. Valenciaga, F, Puleston, P.F and Battaiotto, P.E. "Power Control of a Solar/Wind
Generation System without Wind Measurement: A Passivity/sliding mode approach. IEEE
TransactiononEnergyConvertion.Vol.18No4,pp.501507,December2003.
27. Calvente, F. "Control en Modo Deslizante aplicado a Sistemas de Acondicionamiento de
Potencia de Satlites". Ph.D Thesis of Universidad Politecnica de Catalua. pp. 1162, July
2001.
28. M., Qian. "Sliding Mode Controller Design for ABS System". MS Thesis of Virginia
PolytechnicInstituteandStateUniversity.pp.4251.April1997.
29. Cutler, Charles R, Ramaker, Brian L and Prett, David M. "Dynamic Matrix Control
Method".Patentnumber4349869Houston,Texas,UnitedState,September14,1982.
30.Camponogara,E,JiaD.Krogh.B,Talukdar,S."DistributedModelPredictiveControl".IEEE
ControlSystemsMagazine.Vol.22,pp.4452,February2002.
31. Rovnak, J A and Corlis, R. "Dynamic Matrix Based Control of fossil power plants". IEEE
transactiononenergyconversion.Vol.6.1991.
32.Camacho,EduardoandBordons,Carlos."ModelPredictiveControl".NewYork:Springer
Verlag,Seconded,pp.277,2006.
33. Qin, Joe S. and Badgwell, Thomas A. "A survey of industrial model predictive control
technology". [ed.] Elsevier Science Ltd. Control Engineering practice 11. s.l.: Control
Engineeringpractice11,2003,pp.733764.
34.Garcia,CE,Prett,DMandMorari,M."Modelpredictivecontrol:theoryandpracticea
survey".Automatica(JournalofIFAC).Vol.25,pp.335348,1989.
35. Grieder, P, et al. "Computation of the constrained infinite time linear quadratic
regulator".Automatica.Vol.40N4.April2004.
36.Rawlings,JamesB."Tutorialoverviewofmodelpredictivecontrol".IEEEControlSystems
Magazine.Vol.20,pp.3852,2000.
37. Bemporad, Alberto, Morari, Manfred and Ricker, Lawrence. "Model predictive control
toolboxforusewithMatlab".s.l.:MathWorks,User'sguide,Version2,2006.
126
38. De Prada, Cesar. "Instrumentacion y control de procesos: Fundamentos de control

predictivo de procesos". Los Manuales de Ingeniera Qumica. Universidad de Valladolid
(1997).[Online]http://www.alcion.es/Download/ArticulosPDF/iq/gratis/08articulo.pdf.
39.R.R.Negenborn,B.DeSchutter,J.Hellendoorn.MultiAgentModelPredictiveControl:A
Survey.TechnicalReport:04010.DelftCenterforSystemsandControl,TUDelft,Delft,The
Netherlands,December2004.
40. Pannocchia, G and Rawlings, J B. "Disturbance Models for OffsetFree Model Predictive
Control".AIChEJournal.Vol.49,pp.426437,2003.
41. Venkat, A, et al. "Distributed Output Feedback MPC for Power System Control".
ProceedingsoftheIFACSymposiumonPowerPlantsandPowerSystemsControl.2006.
42.Talukdar,S,Dong,J,Hines,P,andKrogh,B."DistributedModelPredictiveControlforthe
Mitigation of Cascading Failures". IEEE Conference on Decision and Control, and the
EuropeanControlConference.Sevilla,Spain,pp.44404445,December2005.
43. Hines, Paul, Jia, Dong and Talukdar, Sarosh. "Distributed Model Predictive Control for
Electric Grids". Proc. of the Carnegie Mellon Transmission Conference. (December 2004).
[Online]http://www.andrew.cmu.edu/user/phines/pdfs/Hines_2004_CMU.pdf..
44.Capehart,B,Turner,WandKennedy,W."GuidetoEnergyManagement".5ed.s.l.:CRC,
pp.16,2006.
45. Shen, John, et al. "Automotive Electric Power and Energy Management A System
Approach".[Online]http://www.touchbriefings.com/pdf/11/auto031_r_shen.pdf.
46. Winkler, G, et al. "Intelligent Energy Management of Electrical Power Systems with
distributed feeding on the Basis of Forecasts of Demand and Generation". IEE Conference
PublicationCIRED.June2001.
47.West,M.J,Bingham,C.MandSchofield,N."PredictiveControlforEnergyManagementin
all/more Electric Vehicles with Multiple Energy Storage Units". IEEE Electric Machines and
DirvesConference.Vol.1,pp.222228,June2003.
48. Koot, Michiel, et al. "Energy Management Strategies for vehicular Electric Power
Systems".IEEETransactionsonvehiculartechnology.Vol.54,No3.May2005.
49. The Mathworks. "MATLAB Simulink Documentation". (October 2006). [Online]
http://www.mathworks.com/access/helpdesk/help/toolbox/simulink/.
127
50. OrtizRivera, E and Peng, F.Z. "Analytical Model for a Photovoltaic Module using the
Electrical Characteristics provided by the Manufacturer Data Sheet". IEEE 36th Power
ElectronicsSpecialistsConference.pp.20872091,1114Sept2005.
51.Knauff,M.C,etal."SimulinkModelforHybridPowerSystemTestbed".IEEEElectricShip
TechnologiesSymposium(ESTS).ArlingtonVA.:s.n.,pp.421427,May2123,2007.
52. Doerry, N. USN Capt. "Zonal Ship Design". Naval Engineering Journal. s.l.: American
SocietyofNavalEngineers,Vol.118,pp.3953,Winter2006.
53. Chen, Min. and RincnMora, Gabriel A. "Accurate Electrical Battery Model CApable of
PredictingRuntimeandIVPerformance".IEEETransactiononEnergyConversion.Vol.21No
2.,pp.504511,June2006.
54. Valenciaga, F., Puleston, P.F and Battaiotto, P.E. "Variable Structure System Control
Design Method based on a Differential Geometric Approach: application to a Wind Energy
Conversion Subsystem". IEE Procedings on Control Theory Applications. Vol. 15. N1, pp. 6
12.January2004.

128
Appendices
A. PVarrayDataSheet
129

00hybrid Controller

Uploaded by

Document Information

Original Description:

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

00hybrid Controller

Uploaded by

Copyright:

Available Formats

HIERARCHICALCONTROLOFHYBRIDPOWERSYSTEMS

The main contribution of this work is the development and validation of a

rces and load

The electric characteristic for a PV cell is presented on Figure 3, Points A and B

Before discussing SMC techniques it is necessary to understand Variable Structure

x(k +1|k) = Ax(k +]|k) +Bu(k +]|k)

The Shell Multivariable Optimizing Controller (SMOC) proposed by (Marquis &

due to the design based on passivity

Here, the reference bank current I

5. Control for a HPS using Model Predictive

u(t +k -i) +(t +k)

38. De Prada, Cesar. "Instrumentacion y control de procesos: Fundamentos de control

You might also like