Upload
giuseppe-marsico
View
213
Download
0
Embed Size (px)
Citation preview
7/29/2019 Quaderni di applicazione tecnica 10 - impianti fotovoltaici
1/10
IJCSNS International Journal of Computer Science and Network Security, VOL.11 No.6, June 2011 105
Manuscript received June 5, 2011Manuscript revised June 20, 2011
Modelling and non linear control of a photovoltaic systemwith storage batteries: A bond graph approach
Zitouni.NP*P, Khiari.B*, Andoulsi.RP*P, Sellami.AP*P, Mami AP***PT,Hssen.A TP****P
* Laboratory of Photovoltaic and the Semiconductors in CRTEn, BP95 Hammam lif 2050 Tunisia,
** Research Unit: Networks and Machines Electric in INSAT Tunisia***Laboratory of automatic and control systems
**** Laboratory of Water and the Environment in CRTE
Abstract.This paper presents a bond graph modelling ofa photovoltaic source (PV). The insolation variation during
the day poses the problem of energy storage. For that, weused electrochemical batteries which present the best solution
by their good adaptation to photovoltaic source. A maximum
power point tracking (MPPT) device is located between PV
array and batteries to optimise the power transfer from the PV
array and batteries. A non linea control approach of a
photovoltaic system is introduced for the DC-DC converter.
The control strategy is based on state feedback input output
linearization, and the control law is determined from the
established bond-graph model.
Key words:bond graph, photovoltaic source, nonlinear control, storage
battery, DC/DC converter.
1. Introduction
In rural zones, conventional sources of energies arelimited, and photovoltaic energy (PV) becomes a
promising solution. In fact, many advantages arepresented by this energy like the availability on a largescale of planet area. A stand alone photovoltaic systemgenerally uses batteries to maintain supply when the solarenergy is not available [5]. In order to overcome theundesired effect of the PV output voltage, and to assureits maximum power point operating, its possible toinsert a DC-DC converter (MPPT) between the PV
generator and the batteries [3]. There is a thorough studyfound a MPPT control , In order to guide the generatortowards the optimum power point and to
exploit the maximum energy delivered by thephotovoltaic generator. Where [21] develop a PWMcontrol strategies, [20] use genetic assisted, multi-layer
perceptron neural network control,[18]and[19] use fuzzylogic control. In this paper, considering the nonlinearcharacteristic I-V of the PV system, a non linear (N.L)
control law is developed for this system . To exploit themodel bond graph of the photovoltaic system, Thecontrol law is determined from the established bond-
graph model [8][9]. In the first part of this paper, theB.G model for different elements of the system is
presented. In the second part, a theoretical study of thenonlinear control is presented, and the non linearapproach for the control based on the B.G model for thesystem and state feedback linearization is developed,Finally, different simulations are presented.
2. Bond graph modelling for the photovoltaic
source
2.1 The PV array modelThe PV array is basically formed by solar cells. Theequivalent circuit model of a solar cell consists of a
current generator ( phI ) and a diode (D ), serial ( sR )
and parallel ( shR ) resistances [2]. The equation (1) refers
to the PV array current. The equivalent circuit is given byfigure.1[2] [10].
sh
sp
T
psp
sphpR
IpRV
V
IRVIII
+
+= ]1)[exp( (1)
In literature, a model simplification is often done. This
simplification consists in neglecting resistances sR and
shR which facilitates considerably the exploitation of the
PV generator model by avoiding the implicit equations.The simplified equation of the photovoltaic courantintroduced by [2] is given as follows:
7/29/2019 Quaderni di applicazione tecnica 10 - impianti fotovoltaici
2/10
IJCSNS International Journal of Computer Science and Network Security, VOL.11 No.6, June 2011106
Figure. 1. Equivalent schema of a real photovoltaic cell
]1)[exp( = TPSPhP VVIII (2)
Whereq
TKnV BT = R R is the thermodynamic potential,
PhI is the photocurrent of the PV cell proportional to
illumination, SI is the diode reverse saturation current, q
is the electron charge (1,6.10 P-19 Pc), bK is the Boltzman
constant given as flow: (1.38.10 P-23 Pj/k), T is thejunction temperature in Kelvin degree, n is the idealfactor of the photovoltaic cell. The term
]1)[exp( TPs VVI is the darkness diode current
noted DI .
2.2. Bond-graph modelling of the photovoltaicsource
The PV generator has a non-linear current-voltage (I-V) characteristic. This one is characterized by an optimaloperating point corresponding to the maximum powerdelivered by the PV generator. To simulate (I-V)characteristic, we use the bond-graph model representedin figure.2.
Figure. 2. Equivalent scheme of simplified PV array model (a) and itsBond graph model (b)
In order to test the theoretical model, we hadcompared the experimental (I-V) characteristic of the PV
array to the corresponded of BG simulated one. The twocurves are shown in figure.3.
0 5 10 15 20 25 30 35 400
2
4
6
8
10
12
14
16
18
Panel voltage Up(V)
PanelcurrentIp(A)
0 5 10 15 20 25 30 35 40
0
2
4
6
8
10
12
14
16
Ca ract? istique? perimentale I=f(V)
Ip(A)
Vp(V)
Figure. 3. Simulated (a) and experimental (b) (I-V) characteristic of thePV array.
2.3. Bond graph modelling of the battery and chargeregulator
The storage device (load acid batteries) is difficult tocharacterise and several models are used in literature[2][3][5]. Nevertheless, some of these models require
(a)
(b)
(a)
(b)
7/29/2019 Quaderni di applicazione tecnica 10 - impianti fotovoltaici
3/10
IJCSNS International Journal of Computer Science and Network Security, VOL.11 No.6, June 2011 107
0 time 16.5
A
A
A
30
10
Ub (V)
B
B
B
50
-20
Ip (A)
C
C
C
70
-30
Ib (A)
D
D
D
50
0
Up(V)
E
E
E
1
0
reg-cont
na59 - exp13
D
D
C
C
B
B
AA
information on battery parameters, which are not easilyobtained. The photovoltaic generators present a limitedand fluctuating energy source. For that, batteries
allowing the storage of energy at illumination period areused. The accumulators are characterized by the state of
charge Q , the f.e.m )(QVa and internal resistance aR .There are several alternatives of battery modelling. Weretain that presents an electric model relating to the lead-acid batteries. For an ideal model, the battery is
represented by a simple voltage source bE in series with
a resistance bR (with bR is the internal resistance of the
battery) and a storage capacity bC to model the effect of
charge and discharge of the battery [3].
The model parameters are deduced from the current-voltage characteristics during the battery charge anddischarge. figure.4 depicts the equivalent electric schemaof the energy components. The corresponding bond-graph model is represented in figure.5. The bond-graphmodel simulations are illustrated by figure.6 [10][11].
Where bU is the battery voltage, bI represents the
battery current, PV is PV panel voltage, PI is the panel
current, and chI is the load current.
In this work we modelled a charge regulator, which isassimilated to two switches in the PV generator circuitdisposed on both sides of the battery. The charge
programme establishes operates the switches accordingto the battery charge state and the energy availability(solar illumination).
Figure. 4. Complete electric diagram of the energy components
The order of the switchers must hold in account thesense of battery current (charge or discharge phase). Forthe correct operation of the battery, the regulatordeveloped maintains the battery terminal voltage betweentwo limits. The switch model is developed in [10] [22].
Figure. 5. Bond-graph model of the energy components
In the bond-graph model, the battery is modelled by
an effort source bESe = in series with a resistance
brR : and a capacity bCC: , supplied with aphotovoltaic source (already presented). The charge ofthe battery from PV generator and the discharge towardsa receptor is controlled by a charge regulator. The last isassimilated for two switches, put in series on both sidesof the battery, allowing to limit the over charge and thedischarge of the battery. The switches are modelled by
modulated transformerMTF in serial with resistive
element onRR : . These elements are controlled by aprogram pre-established depending on the battery stateand the PV source. The simulation of the BG model is
presented by figure.6.
Figure. 6. Simulation of the Bond-graph model of the energycomponents
From simulations of figure.6, we have distinguishedthe charge and discharge phases of the battery between
two limits ( bdU and bcU ). The battery voltage is the
indicator of state of charge. According to this state theregulator commutates between two phases, where the
RchRRDIR h VRp
IRch
CRp
rRb
CRb
ERb
Chargeregulator
IR
Sf : IR h 0 : URp 1 : IR 0 : URb 1: IRch R : RRch
1 : IR R : rR
Se : ERb
C : CR
C : CR
R: RonR1 R: RonR2
R:RRD
Switchcontrol
7/29/2019 Quaderni di applicazione tecnica 10 - impianti fotovoltaici
4/10
IJCSNS International Journal of Computer Science and Network Security, VOL.11 No.6, June 2011108
battery current change the sign panel current joints zero
and the panel voltage )( pV towards the panel open
circuit value)( po
V .
2.4 The bond graph modeling of DC/DC converter
The DC/DC converter model
When the PV generator is directly connected to theload, the system will operate at the intersection of the
VI curve and load line, which can be far from themaximum power point MPP. The converter consists of
power components and control circuit. A control
algorithm permits to assure maximum output power. TheMPPT (Maximum power point tracking) controller usedin our case is based on a boost converter. The equivalentschema of DC/DC converter is presented by figure.7.
Figure. 7 Schema of the buck converter
3.1 The bond graph modeling of DC/DC converter
This method supposes the ideal switches and givesindirectly the average models appropriate to the variousstatic inverters which permit to deduce easily the bond-graph model. For the case of the boost converter, theaverage values of electric parameter are given byequation (3) and (4) [2][6][22].
=+=T
TIdtti
Tdtti
TI
210 11).(
1).(
1 (3)
10 222).(1).(1 Vdttv
Tdttv
TV
T
T
=+= (4)
The BG model corresponding to the DC/DC converterdeveloped by [2][22] is illustrated in the figure.7.The complete BG model of the photovoltaic source isobtained by associating the models developed previously.The global model and the bond-graph model arerespectively represented by the figure.9 and figure.10.
Figure. 8. The bond-graph model of boost converter
Figure. 9. Complete electric schema of the solar energy supply
Figure. 10. Global bond-graph model of photovoltaic system
3. Non linear control law
The main objective is to charge batteries, it consist tomaximize the power absorbed by the batteries [7][8]. Thecontrol law can be determined by an analytical method ordirectly from the B.G model. In this paper we will exploitthe B.G model to determine the control law, this methodincludes five stage [8][9]. In what follows we will
present a study of the input output linearization controllaw
7/29/2019 Quaderni di applicazione tecnica 10 - impianti fotovoltaici
5/10
IJCSNS International Journal of Computer Science and Network Security, VOL.11 No.6, June 2011 109
The control of a broad class of non-linear systems canhave a linear input-output behaviour by choosing anonlinear control law per state return. We point out the
theory of this control for the monovariable case [12] [5].We consider the monovariable model with single output:
( ) ( )( )xhy
uxgxfx
=
+=&(5)
WherenRx , mRu , (.)f et (.)g are two fields
of vectors ofn
R
By deriving y we obtains:
=y& LRfR h+ LRg Rh u uhLhLy gf +=& (6)
Where hLf et LRg Rh hLg respectively indicate the Lie
derivation of h in the direction of vectors fields f
and g .
)(,)(1
xfx
hfdhxhL j
n
j j
f =
>====<
Ifn
g RxxhL ,0)( , The control low has thefollowing form:
vxxu )()( += (8)
Where (.) and (.) are two function of nR This law is given by:
)(1
vhLhL
u fg
+=
(9)
Thus leading to the linear system:
vy =& If 0)( xhLg , we derives (5) to have:
uhLLhLy fgf )(2 +=&& (10)
With )(2 hLLL fff = and )( hLLhLL fgfg =
As previously, If 0hLL fg ,n
Rx the control
law:
( )vLhLL
u ffg
+= 21
(11)
In a way more general, if r is the smallest entirety such
as 02 hLL fg for 2,...,0 = ri
Andnr
fg RxxhLL ,0)(1 ,
then the control law is written:
( )vhLhLL
ur
fr
fg
+=1
1
(12)
Giving:
vyr =)(
The number of derivation of the output necessary toreveal the input is known as relative degree of the model,extension of the definition of the relative degree intolinear[14].The diagram block of a control system by linearization
input-output by return of state is as follows:
Figure. 11 : diagram block of a controller process by linearizationinput-output by return of state
Once the linearization is achieved, several controlmethod can be carried out such as the placement of pole,the reference model .... etcBy using the input-output linearization by state return,the dynamic of a nonlinear system are broken up partlyexternal (input-output) and an internal part(unobservable). Since the external part makes up of alinear relation between y and v (or per equivalence the
canonical form of commandability between y and u ), it
is easy to conceive input v , where the output has the
Input-output
linearizati
Externallinear
control
Processus Output
R
vx
u y
7/29/2019 Quaderni di applicazione tecnica 10 - impianti fotovoltaici
6/10
IJCSNS International Journal of Computer Science and Network Security, VOL.11 No.6, June 2011110
desired behaviour. It remains to know if the internsdynamic well will behave (if the internal states willremain limited). The interns stability for nonlinear
systems was treated in various work [15][16]. Theessential definitions and results are as follows:
Dfinition 1 :
System (5) have a relative degree rif:
)()()( 2 xhLLxhLLxhL rfgfgg== K = 0 and
)(2 xhLL rfg 0 xRPnP.
The model (5) have a relative degree r if eachn
Rx the output requires r derivations before revealing the
control in (6).
If a system has a relative degree r, there is easy to check
that eachn
Rx 0
there is a vicinity 0u of 0x suchas the change of co-ordinates for the state defined by:
)(
)(
122
111
xhLzT
xhzT
f==
==(13)
M
)(11
xhLzTr
frr
==
With 0)()(1 =xgxdT for nri ....1+= is adiffeomorphism
If we considerT
nr TTZ )....( 12 += ,the equation (5) canbe written in the following normal canonical form:
1211 zz =&
M
rr zz 111&&
= (14)
),(
),(),(
212
2112111
zzz
zzgzzfz r
=
+=
&
&
11zy = (15)
In (14) ),( 211 zzf represent )(xhLr
f and ),( 211 zzg
represent )(1 xhLrf . Now if 0=x is a balance point of
the system below ( 0)0(0)( == handof ), then the
dynamics (16) are the zeros dynamics.
),0( 22 zz =& (16)
These dynamics are the dynamic ones madeunobservable by the state return. Indeed the lawlinearization by state return is the nonlinear equivalentof the poles zero placement in loop closed of the model,in order to make them unobservable [15].It is noted that the subspace:
{ } { }0:)0()(: 1010 ==== zUxhLxhLUxL rff LCan be invariant while choosing:
)),((),(
1211
211
vzzfzzg
u += (17)
The dynamics (16) are the dynamics on this space.The system (5) is known as minimal of phase if thedynamics of the zeros are asymptotic stable as indicates itthe following definition:
Dfinition 2 : [16]
A nonlinear system is known as asymptotically phaseminimal if the dynamics of the zeros is asymptoticallystable.The dynamic interns associated with the linearizationinput-output, corresponding to the last equations: normalform, generally depend on the output. For that wedefines an intrinsic property of the nonlinear systemconsider the system internal dynamics when the entrymakes it possible to maintain the output in zeros. Tosuppose the null output implies that all its derivatives are
null. For that the dynamic interns correspond to themodel where the zeros dynamic describe the movements
on a surface 0M of rn dimension defined by
01 =Z [14].
4. Determination of the control law directly on Bond
Graph
The control law can be determined directly from the B.Gmodel, this method includes five stage [8][9].
7/29/2019 Quaderni di applicazione tecnica 10 - impianti fotovoltaici
7/10
IJCSNS International Journal of Computer Science and Network Security, VOL.11 No.6, June 2011 111
The photovoltaic state equation can be put in thefollowing form (5).
(18)
Stage 1: Bond-graph model with standard causality
The minimal dynamic way or output-input way ( )pv, ,having only one integrator. The number of this integrator
gives the relative output-input degree: 1=r .
Figure.12. Bond-graph model with the minimal dynamic way
Stage 2: development of the reverses model
The method is based on the causal inversion input-outputfor the B.G of the system; it consists to deduce the staticand dynamics input-output control law directly on bond-graph model. For the design of the control system, it isoften useful to determine the input from the desiredoutput. This is obtained by the inversion of the systemdynamics where the input is replaced by the output,which corresponds for a control of the opposite system.To determine the opposite model, the bond graphtechnique was extended by the introduction of theconcept of bi causality [13].
Figure.13. The opposite bond-graph model
The currents3
f et6
f became the output of the
transformerWith:
6
31
f
fm ==
(19)
From this equation we can deduce the control law :U
6
3
f
fU ==
=
=
63
36
ff
ee
(20)
Stage 3: Deduction of the minimal dynamic equation
We deduce a mathematical model reverses calledminimal dynamic equation. The equation is obtainedwhile reading the bond graph model with oppositecausality.The cyclic ratio of the boost converter is the input of thecontrol law U .
The state vector isT
cbcpch qqpx ][=
The output is the voltage of the generator: pvy =
225
46 ,, fdt
dvcq
R
ef
l
pf
p
Dch
ch ==== & (21)
According to(20) and (21) we obtain:
+
=
ph
b
cb
cp
ch
ch
pDch
bpchb
cb
cp
ch
I
E
q
q
p
l
cRl
cclr
q
q
p
00
10
01
001
01
1
&
&
&
7/29/2019 Quaderni di applicazione tecnica 10 - impianti fotovoltaici
8/10
IJCSNS International Journal of Computer Science and Network Security, VOL.11 No.6, June 2011112
ch
chTpsph
ch
ch
pTtpsph
p
lycvvII
p
lU
vcvvIIf
&
&
=
=
)1)(exp[
]1)[exp(3(22)
Stage 4: development of control law
From the dynamic equation minimal, we replace
)1()( =ry r& by a new control input v ; the input-outputrelation system (with feed-back) becomes linear
according to the equation: vyy r == & The linearization static law is:
vp
lcIcvqI
p
lU
ch
chphTcs
ch
ch += ])1)exp([( (23)
The feedback linearization control law is inserted inPV voltage control loop in order to reach the optimalvoltage. This objective can be achieved by a simple
control law )(: yykvv d =
In this case, the control transformed the system intointegrating block, a proportional P regulator issuitable.We can make the measurable sizes:
vI
cIvvI
IU
b
phTps
b
+= ])1)exp([(1
(24)
The diagram of the system is given by the followingfigure.
Figure.14. Diagram of the system with control law
This control law made two states unobservable
( chch pq , ) from the new control v . According to thenonlinear control theory, this variety represents thedynamics of the zeros of the system whose study ofinternal stability is essential
Stage 5: study of stability
The dynamics of the zeros is obtained by
imposing 0=y , all bonds attached of 0 junction iscancelled, consequently the power is cancelled. We applythe second method of lyapunov to the bond-graph toanalyze the stability of dynamics of the zeros. We canchoose the sum of the energies stored in the elements of
storage Iet C like function candidate of lyapunov [9].
))()
(2
1
)()()(
22
1 1
b
cb
ch
ch
n
i
n
j
jjnii
c
q
l
pV
qpVtEi j
i
+=
+== = =
+
(25)
The derivative of this energy represents the differencebetween the power delivered by the source and theconsumption by the dissipative elements.
0])([)()(
)()()(
22
1 1211
7/29/2019 Quaderni di applicazione tecnica 10 - impianti fotovoltaici
9/10
IJCSNS International Journal of Computer Science and Network Security, VOL.11 No.6, June 2011 113
and acV ). The battery voltage is the indicator of charge
state of the battery. According to this state the regulatorcommutates between two phases, where the battery
current change the sign and panel current joins zero. Thecontrol law stabilise the system on the desired output
power and the output follows the input.
Figure 15. The BG model with feedback linearization input outputcontrol
5. Conclusion
The pilot unit, the object of our study, consists of aphotovoltaic generator, storage battery with chargeregulator, Maximum power point tracking (MPPT) andthe loads. With these varied elements, the modelling ofthis pilot unit becomes complex. For that, we proposed a
unified approach of modelling based on graphictechnique known as bond-graph. This technique issystematic and has a sufficient flexibility to be able tointroduce various components into the system and toestablish a nonlinear control law.In this paper an electric model of the photovoltaic sourceis presented and the complete BG model for differentelements of the system is elaborated. This model wasused to design a state feedback linearization input outputcontrol law. A non linear approach has been proposed tocontrol the loading of a battery and exploit the maximumof energy delivered by the PV panel. In this paper wehave exploit the B.G model to determine the control law,
(this method includes five stages). Reliable simulationresults are presented to demonstrate the validity of the
proposed control approach.
0 2 4 6 8 10 12 14 16
28
30
32
Up
and
Upref(V)
0 2 4 6 8 10 12 14 160
0.5
1
controllaw
u
0 2 4 6 8 10 12 14 1610
20
30
Ub(V)
0 2 4 6 8 10 12 14 165
10
15
Times (s)
Ib(V)
Figure.16. Simulation of model BG of the system with state feedbacklinearization control
0 2 4 6 8 10 12 14 16
20
30
40
Up,
Ub,
Uo
p
(V)
0 2 4 6 8 10 12 14 160
0.5
1
charge
control
0 2 4 6 8 10 12 14 16
0
10
20
Ip(A)
0 2 4 6 8 10 12 14 16-20
-10
0
10
Times(s)
Ib(A)
Ub(V)
Up(V)
Upref(V)
Figure. 17. Simulation of model BG of the system with chargeregulator and state feedback linearization input-output control
law.
7/29/2019 Quaderni di applicazione tecnica 10 - impianti fotovoltaici
10/10
IJCSNS International Journal of Computer Science and Network Security, VOL.11 No.6, June 2011114
References[1] B.Allard, H Morel,Ph Lautier, and J.M.Retif, Bong-graphs
for averaged modelling of power electronic converters.
Proceeding of the SCS International Conference on Bond-graph Modeling (ICBGM 97). Phoenix, USA, Vol 29, N1, 201-206.
[2] R.Andoulsi, A. Mami, G. Dauphin-Tanguy, M. annabi,Modelling and simulation by Bond Graph technique of aDC motor fed from a photovoltaic source via MPPT boostconverter. Proceeding of CSSC99, pp 4181-4187.
[3] Bogdan S.Borouy, Zied M. Salameh: Dynamic reponse ofa stand-Alone wind energy Conversion System with
battery energy storage to a wind Gust. IEEE transaction onEnergy Conversion. Vol 12.N1,marsh1997.
[4] J.N. Ross, T. Markvart and WHE, Modeling batterycharge regulation for a stand-alone photovoltaic system,Solar Energy vol.69 N3, issue 2000,181-190.
[5] B.S.Borowy and Z.M.Salameh, Dynamic response of a
stand-Alone wind energy conversion system with batterystorage to a wind gurst, IEEE Transaction on Energyconversion vo1. 12, issue 1 Marsh 1997, 73-75.
[6] V. Mahout,On the robustness of the flatcontrol, ISC3,juin 2003.
[7] H. SiraRamirez and Silva Navarro G.,Regulation andtraking for the overage boost converter circuit: ageneralized proportional integral approach, Int. Journal ofcontrol, vol. 75, pp. 988-1001, 2002.
[8] Bertrand, J.C. Sueur, G. Dauphin-Tangy. Bond Graph forModelling and Control Structural Analysis Tools for thedesign input output Decoupling State Feedbacks.Proceding ICBGM1997/Simul.Series 29:1.
[9] Junco. Sergio, Linarisation exacte entre-sortie et stabilit
de la dynamique des zeros directement sur bon graph dessystmes non linaires. Actes de CIFA 2000, Premireconfrence Internationale Francophone dautomatique, 4-7
juillet 2000.Lille, France.[10]Junco. Sergio, Lyapunovs second method and feed-back
stabilization directly on Bond Graph, Proceeding ofICGBM01, 2001, pp 86-98.
[11]Zitouni.N, Andoulsi.R, Sellami.A, Mhiri.R, Bond graphmodelling of a UV water Disinfection pilot unit fed by a
photovoltaic source. JTEA'2006, 12-14 may 2006 Hammamet, Tunisie.
[12]P.J. Gawthrop, D.J. Wagg, S.A. Neild, Bond graph basedcontrol and substructuring, Simulation Modelling Practiceand Theory 17 (2009) pp 211227.
[13]N. Yadaiah, N. Venkata Ramana, Linearisation of multi-
machine power system: Modeling and control A survey,Electrical Power and Energy Systems 29 (2007) 297311
[14]Roger F. Ngwompo, Peter J. Gawthrop, Bond graph-basedsimulation of non-linear inverse systems using physical
performance specifications, Journal of the FranklinInstitute 336 (1999) 1225}1247
[15]J. J. E. Slotine W.Li, Applied nonlinear control, Printice-hall international editions. Englewood Cliffs, New Jersey1991.
[16] C Kravaris, Input/output linearization: A nonlinear analogof placing poles at process zeros. AICHE Journal, Vol 34,
N11, November 1988, pp 1803-1812.
[17] A. Isidori, A tool for semi-global stabilizationof uncertainnon-minimum-phase nonlinear systems via output feed-
back. IEEE trasaction on automatic control, Vol 45.N 10October 2000, pp 1817-1827.
[18]I.H. Atlas and A.M.Sharaf, Anouvel maximum powerfuzzy logic controller for photovoltaic solar energysystems, Renewable energy 2007.
[19]Nopporn Patcharaprakiti, Suttichai.P andYosanai.S,Maximum power point traking usingadaptative fuzzy logic control for grid-connected
photovoltaic system, Renewable energy 30 (2005), pp1771-1788.
[20]R.Akkaya, A.A.Kulaksiz and . Aydogdu, DSPimplementation of a PV system with GA-MLP-NN basedMPPT controller supplying BLDC motor drive, EnergyConversion and Management 48 (2007), pp 210-218.
[21]N.Hamrouni, M.jraidi, A.Cherif, A.Dhouib, Measurementand simulation of a PV pumping systems parameters using
MPPT and PWM control strategies, IEEE MELECOM2006.[22]D.Mezghanni, R. Andoulsi, A. Mami, G. Dauphin-Tanguy,
Bond graph modelling of a photovoltaic system feedingan induction motor-pump Simulation Modelling Practice andTheory 15 (2007) 12241238.
Naoufel,Zitouni is born in Tunis ; Tunisia 1978. He obtainedthe engineer diploma of engineer and master diploma inIndustrial Informatics and Automatics from (ESSTT) Tunisiarespectively in 2002 and 2004. He is a thesis student. His fieldof interest concerns the photovoltaic power, treatment water,
bond-graph modelling.Zitouni Naoufel, FST,Post address: campus universitaire, el Manar II, Tunis, email: