INTERNATIONAL JOURNAL of RENEWABLE ENERGY RESEARCH

M. Nadour et al., Vol.7, No.3, 2017

Comparative Analysis between PI & Backstepping

Control Strategies of DFIG Driven by Wind Turbine

Mohamed Nadour*, Ahmed Essadki*, Tamou Nasser**

* Laboratory of Research in Electrical Engineering, Higher Normal School of Technical Education (ENSET), Mohammed V
University, 10100, Rabat, Morocco
** Laboratory of Research in Electrical Engineering, Higher National School of Computer Science and Systems Analysis
(ENSIAS), Mohammed V University, BP 713, Rabat, Morocco
(mohamed.nadour@um5s.net.ma, ahmed.essadki1@gmail.com, tnasser@ensias.ma)

‡
Corresponding Author; Mohamed Nadour, Morocco, Tel: +212 666 508 241,
, mohamed.nadour@um5s.net.ma

Received: 14.02.2017 Accepted: 08.04.2017

Abstract- This paper presents the modelling and control designs of Double-Fed Induction Generator (DFIG) integrated into a
wind energy conversion system (WECS). The aim is to design and compare two distinct strategies of controlling independently
the active and the reactive power generated by a DFIG, in order to conclude on their performances. In first place, a modelling of
wind turbine and DFIG is presented. Then, the PI controller, the Backstepping approach and a maximum power point tracking
(MPPT) strategy used to extract the maximum of power during the conversion are developed. Simulations results show
significantly improved performances of the proposed backstepping approach over the Proportional-Integral controller, in terms
of dynamic response, disturbance rejection, and robustness against parametric variations. The systems performances were tested
and compared using Matlab/Simulink Software.
Keywords Wind energy conversion system, DFIG, PI, Backstepping, Maximum power point tracking, MPPT.

1. Introduction mechanical complexity, separately controllable active and

reactive power, and a smaller power converter compared the
In addition to other renewable energy sources, the wind same class of other generators (reduced losses and costs),
energy is proving to be one of the preferred choices to produce about one third of the rated power for a variable speed
electricity. Unlike fossil fuels, wind energy is clean, pollution- generator operating at  30% of synchronous speed [14].
free from greenhouse gases emissions (CO2) and
Over the past few decades, the classical control strategies
inexhaustible. However, it is not powerful enough to replace
that are normally based on linear PI controller presents
the existing sources, this fact have attracted a lot of interests
suitable performances in many control applications. However,
on the methods of exploitation and development of the wind
these techniques suffer many limitations, such as machine
energy conversion systems (WECS).
parameters variations. Therefore, various studies and new
Over the years, several technologies have been proposed. strategies of control have been invading the electrical
In this paper, we focus on the grid connected, variable speed engineering laboratories, in attempt to achieve high
wind turbines equipped with double-fed induction generator performances. During the pasts few years, there has been a
(DFIG). Which is a wound rotor induction generator, where massive amount of activity on a new special control schemes
the stator windings are directly connected to the grid and the known as "Backstepping algorithms". With these control
rotor winding are connected through back-to-back converter approaches, the feedback control laws that can ensure at the
consists of two converters. i.e., rotor side and grid side same time a good tracking response and stability of the overall
converter, separated with a dc-link capacitor placed as energy system can be easily constructed. In this paper, it has been
storage, in order to keep the voltage variations small to the shown that replacing the PI controller by a backstepping
inverter [2,8]. DFIG is the most popular system among all control algorithm can significantly improve the tracking
WECS technologies, it due to its high-energy efficiency, less response and give excellent performances.
INTERNATIONAL JOURNAL of RENEWABLE ENERGY RESEARCH
M. Nadour et al., Vol.7, No.3, 2017

The paper is organized as follows. Section II presents a

description and a detailed aerodynamic modelling of the
variable speed wind turbine, the mechanical transmission
system and the DFIG. Section III deals with both PI and
backstepping control design, and a Maximum power point
tracking “MPPT” strategy using backstepping controller to
extract the maximum power from the wind velocity. Section
IV simulation results performed by means of Matlab/Simulink
software are presented and discussed, and then we finish by a
conclusion.

2. Modelling of the Wind-Energy Conversion System

Fig. 2 : Power coefficient Cp
The wind-energy conversion system consists of the wind Expression of the aerodynamic torque is given by:
turbine, gearbox and DFIG Fig.1. The turbine captures the
Pt SV 31
kinetic energy of the wind and converts it to a torque that Ct 
 C p (  , ) (5)
rotates the rotor blades. Thereafter, the DFIG converts the t 2 t
mechanical power into an electrical power. The gearbox is used as a connection, to adapt the speed
of the turbine to that of the generator. The friction, elasticity
and energy losses in the gearbox are neglected so:
 méc   t .G ; C t  Cg .G (6)
The fundamental equation of dynamics can be written:
d méc
C méc  J  C g  C em  f . méc (7)
dt

Fig. 1: Wind-energy conversion system 2.2. Modelling of DFIG

2.1. Modelling of wind turbine The modelling of the DFIG is similar to the induction
generator, the only difference is that the rotor windings are not
The wind power that pass through a surface S is defined short-circuited. The DFIG is represented in the park frame by
the following equations [3, 15]:
as follow ( is the density of air) [14]:
 d sd
SV13 v sd  R s i sd   w s  sq
Pv  (1) dt
2  d sq
The aerodynamic power captured by the rotor is given by: v  R s i sq   w s  sd
 sq dt
SV13  d
(8)
Pt  C p (, )Pv  C p (, ) (2)  v rd  R r i rd  rd  w s  rq
2  dt
The power coefficient Cp represents the aerodynamic  d rq
efficiency of the wind turbine. It depends on the blades  v rq  R r i rq   w s  rd
 dt
orientation angle β and the tip speed ratio:
R. t sd  L si sd  Mi rd
 (3)   L i  Mi
Va  sq s sq rq
 (9)
The expression of this power coefficient has been 
 rd   L i
s rd  Mi sd
approached for this type of turbine, by the following equation  rd  L si rq  Mi sq

[6]:
C p (, )  C1 (C 2 (
1 0.0035
 2 )  C3  C 4 ) V being the voltage (V), i is the current (A), s and r
  0.08   1 are angular speed of the rotating field referred respectively to
0.0035 the stator and the rotor (rad/s), Rs and Rr are respectively
. exp( C 5 (  C 6 ) (4) stator and rotor resistance (), sand r are the stator and the
  0.08).( 2  1) rotor flux. The electromagnetic torque can be expressed as a
C1=0.5109; C2=116; C3=0.4; C4=5; C5=21; C6=0.0068. function of the stator flux and current:
M
Fig.2 shows the power coefficient Cp curves for multiple C em  p
(sqi rd  sdi rq ) (10)
values of β. These curves are characterized by an optimum Ls
point which is the point corresponding to maximum power The stator active and reactive power are given by the
coefficient, e.g. for β=0 (Cpmax=0.48; opt=8.15). following expression:

1308
INTERNATIONAL JOURNAL of RENEWABLE ENERGY RESEARCH
M. Nadour et al., Vol.7, No.3, 2017

 Ps  Rev si s *  ( v sdi sd  v sqi sq )  d s

 (11) v sd  0
Qs  Imv si s *  ( v sqi sd  v sdi sq )  dt (15)
v sq  w s  s  v s

3. Control Strategies of DFIG The stator currents can be expressed in terms of rotor
currents as follows:
Fig.3 shows that in order to ensure the highest energy  s M
conversion efficiency the DFIG must operates at a variable i sd   L  L i rd
 s s
speed following the optimum power characteristics  (16)
i   M i
Pmax=f(opt) (with opt corresponding to Cpmax). Therefore, at  sq Ls
rq
a given wind speed, the turbine must have been exposed to a
Replacing stator currents expression in (9), rotor flux
resistive mechanical torque Cem-ref, or in other words an active
expressions can be rewritten as:
power Ps-ref delivered by the DFIG.
  M 2  Mv s
 rd   L r  i rd  w L
  L s  s s
 (17)
  M 2 
 
 rq   L r  L i rq
  s 
The expressions of rotor flux equation (19) is integrated
to (9). As a result, we can establish the relations between rotor
currents and voltages we obtain:
  M 2  di rd  M 2 
v rd  R r i rd   L r    gw s  L r  i rq
  L s  dt  L s  (18)

  M 2  di rq  M 2  M
 
v rq  R r i rq   L r  L  dt  gw s  L r  L i rd  gw s L  s
  s   s  s
Fig. 3: Optimum power characteristic of the wind turbine
M2 Is constant called the dispersion coefficient.
  1
Ls L r
3.1. Reference choice for the two-phase DFIG model
  1 
i rd    Vrd  (gw s L r i rq )
In order to achieve independent control of active and 
reactive power, a two-phase d-q rotating reference frame is   R r  pL r  s
 (19)
chosen related to the stator field Fig.4 [3, 15]. Such as:   1   M 
   
v rq   R  pL   Vrq  (gw s L r )i rq  gw s L  s 
sd = s & sq = 0 (12)   r r  s  s 
Finally, according to (13), (17) and (18):
 M
Ps   v s L i rq

s (20)
s M
Qs   v s  vs i rd
 Ls Ls
From the above equations, we can develop a block diagram of
the DFIG to regulate Fig.5, the DFIG model become
approximately decomposed in two decoupled subsystems.
Fig. 4 : Orientation of Stator flux on d axis
The electromagnetic torque equation is then:
M
Cem  p sdi rq (13)
Ls
The equations of the stator flux can be rewritten as follows:
 s  L s i sd  Mi rd
 (14)
0  L s i sq  Mi rq
For medium and high power generators used in wind
turbine, the stator resistance Rs can be neglected [5, 10, 14].
In addition, assuming a stable grid which provides a constant
stator flux Фs, allows us to simplify the stator voltages
expressions, thus: Fig. 5 : Block diagram of DFIG to regulate

1309
INTERNATIONAL JOURNAL of RENEWABLE ENERGY RESEARCH
M. Nadour et al., Vol.7, No.3, 2017

3.2. Proportional Integral (PI) controller design With: A  L s R r  pL s L r  ; B  U s M

The PI controller is the most commonly used and the The open loop transfer function is given by:
easiest to synthetize for DFIG control purposes, and in many  
 p  k i  vs M
  k p  L s L r 
industrial control schemes. This type of regulator is a
 k  vs M
combination of both proportional and integral action. It has the G (p)   k p  i   (24)
effect of improving simultaneously the steady and transient  p  L s R r  pL s L r    p   Rr 
    p 
states.  k p   Lr 
 
Working Assumption [5]: To eliminate the zero of the transfer function (the
 According to the block diagram of DFIG Fig.5. compensation method) we choose:
Before proceeding with the regulator designs, a ki R
 r (25)
compensation block given by the following k p Lr
equations must be added after the PI controller:
The open loop transfer function becomes:
 vs M
e d  g L kp s
v M
 s
Ls L r  1 LL
 2
(21) G ( p)   ; r  s r (26)
e  v s Ls
R r  pL r  p r p k p vs M
 q L w Mv
 s s s
The closed loop transfer function can be expressed as:
In steady state, we can write: 1
F(p)  (27)
 vs M 1  r p
e d  g L Finally, we get:
 s
 2
(22)  1 Ls L r 
e  v s Ls
Rr k p  3 v M
 q L w Mv  10.10 s
 s s s  (28)
k  1 L R
s r
 The influence of coupling terms between the two  i 10.10 3 v s M
axes Fig. 5 is minimal which allows us to control
each axis independently [9,8]. In this study, the time constant is set to 10ms, which
corresponds to a fast enough value for a medium and high
 The dynamic of the inverter is very quick compared power energy conversion system [5].
the machine. In this case, the inverter model can be
reduced to its static gain K=1 which reduces the 3.3. Backstepping controller design
system order.
The arrival of '' Backstepping '' provides a systematic
A typical structure of a PI control system is shown in technique to decompose a complex nonlinear control design
Fig.6. The error signal e(t) is used to generate the proportional problem into smaller and simpler subsystems, by the use of
and integral actions, with the resulting signals weighted and so-called “virtual control”. Backstepping technique was
summed to form the control signal u(t) applied to the plant developed in the early 90 by Kanellakopoulos (1991) and
model. inspired by the work of Feurer& Morse (1978) on one hand
and Tsinias (1989) and Kokotović & Sussmann (1989) on the
other. This method of control design is divided into various
steps. In each step, we essentially deal with a simpler and
easier single input single output design problem, and each step
provides a reference for the next design step. The overall
system’s stability and performance are achieved by a
Fig. 6 : A typical PI controller structure Lyapunov function.

Fig. 7: Block Diagram of the proposed PI control design

1310
INTERNATIONAL JOURNAL of RENEWABLE ENERGY RESEARCH
M. Nadour et al., Vol.7, No.3, 2017

In this section, we present a backstepping control approach the synthesis of rotor current references (virtual control
applied to DFIG. The control scheme is designed in a way to variables), such as:
keep the same general structure of the field oriented control.
The synthesis of this control can be achieved in two steps. The   v M Mv s 
Ps  ref  s  Vrq  R r I rq  gs I rd  g   k1e1
DFIG model Fig.5 can be presented as differential equations  Ls   Ls  (34)
 
of active and reactive power, and rotor currents in the park Q vs M
 

frame, as follows:  s  ref  L  Vrd  R r I rd  gs I rq   k 3e3
 s 
 M Mv s This allows finding the desired current control state, either:
P s   vs L  (Vrq  R r I rq  gs I rd  g L ) 
 Ls    1  vs M 
I rq  ref   Ps  ref  k1e1   R  Vrq  gs I rd  g L 
s s
I rq   1 (V  R I  g I  g Mv s ) (29)  vs MR r   r  s  (35)
 rq r rq s rd 


Ls I 
Ls 
 rd  ref vs MR r
  1
 Qs  ref  k 3e3    Vrd  gs I rq 
 y1  Ps    Rr
Moreover: Therefore, the control variables are asymptotically stable.
 M
Qs   vs (Vrd  R r I rd  gs I rq )  Step 2: Computation of the control voltages
 L 
 s
The currents determined previously, are the desired
I   1 (V  R I  g I ) (30)
 rd 
rq r rd s rq variable for this step. We define the errors e2 and e4:
   

 y 2  Qs e 2  I rq ref  I rq  e 2  I rq ref  I rq
M2   
With:   L r   (L r ) e 4  I rd  ref  I rd  e 4  I rd ref  I rd (36)
Ls
Vrq and Vrd are now the real control signals. The new
 Step 1: Computation of the currents reference Lyapunov function are chosen as:
We select the first two subsystems from (30) and (31). The 1 1
output of the first subsystem Ps must tend to Ps_ref and the V2  (e12  e 22 ) ; V4  (e 32  e 24 ) (37)
2 2
output of the second subsystem Qs must tend to Qs_ref. We Using the equation (36) the derivative of functions (37):
define (e1, e3) representing the errors variables:
      v M Mv s 
e1  Ps  ref  Ps  e1  Ps  ref  Ps V 2  e1 P s  ref  s  Vrq  R r (I rq  ref  e 2 )  gs I rd  g 

     Ls  Ls 
e3  Qs  ref  Qs  e3  Qs  ref  Qs (31) 
  1 Mv s 
Virtual control variables are Ird and Irq . The first Lyapunov  e 2 I rq  ref    Vrq  R r (I rq  ref  e 2 )  g s I rd  g L 
  s   (38)
functions are chosen such that: 
 
 
 
 v M
V 4  e 3 Q s  ref  L  Vrd  R r (I rd  ref  e 2 )  gs I rq 
s
1 1
V1  e12 ; V3  e 32 (32)  
2 2  s
  
 
 vM
Using the equation (32):  e 4 I rd  ref  s Vrd  R r (I rd  ref  e 2 )  g s I rq 
  L s  
    v M Mv s 
V1  e1 e 1  e1 P s  ref  s  Vrq  R r I rq  gs I rd  g  In order to guarantee a stable tracking the control voltages Vrd
  Ls   Ls  (33) and Vrq are deduced as follow:

V  e e  e Q  vs M
      v MR r   v M
 3 3 3 3  s  ref  Vrd  R r I rd  gs I rq 
   k 2 e 2  I rq  ref  s e1    R r I rq  gs I rd  g s 
 Ls   Vrq C
   Ls   L s  (39)
To guarantee a stable tracking the derivative of V1 and 
  v s MR r 
V2 must be negative (Lyapunov second theory). This allows
V    k 4 e 4  I rd  ref  e 3   R r I rd  gs I rd 
 rd C  Ls 


Fig. 8: Block Diagram of the proposed Backstepping approach control design

1311
INTERNATIONAL JOURNAL of RENEWABLE ENERGY RESEARCH
M. Nadour et al., Vol.7, No.3, 2017

The derivative of equation is given by: The derivative of (44) is given by:
 
V2  k1e12  k 2 e 22 0 ; V4  k 3e32  k 4 e 24 0     
V1  e() e()  e()  ref    (45)
This allows concluding that the system is globally  
asymptotically stable.k1, k2, k3and k4 are positive constants, By setting (42) in (45) we obtain:
the good choice of these parameters guarantee the stability,

 

 1
and a fast dynamic response. V1  e() ref  Cg  Cem  f .méc  (46)
 J 
The block diagram of the proposed control scheme is
In order to guarantee a stable tracking Cem is chosen such
illustrated in Fig.8. The PI controller is replaced by four
that:
blocks that represent the backstepping approach. The “Irq-c
computation” and “Ird-c computation” blocks provides 

1


  ref  C g  C em  f . méc   ke (47)
respectively the rotor currents references components Irq-ref, Ird-  J 
ref (virtual control variables) via the active and the reactive k is a positive constant, finaly we get :
power feedback controls From Eq. (35). Moreover, the
 
voltages commands Vrq-c, Vrd-c are generated by the “Vrq-c C em  J  ref  ke()   C g  f . méc (48)
Computation” and “Vrd-c Computation” blocks via the rotor  
currents feedback control of according to (39). 
We have V1  ke 2 0 . Therefore, according to
3.4. Maximum power point tracking: Lyapunov the system is globally asymptotically stable.
Wind energy, even though abundant, varies continuously
due to the fluctuating nature of the wind speed. The power
captured by the turbine depends mainly on the power
coefficient Cp, which is a function of the speed of the
generator (or TSR ). Maintaining Cp at its maximum value
allows the wind-energy conversion systems to produce all the
power they are capable of.
Several maximum power extraction algorithms exist in
the literature. This paper, focus on the TSR control, using a
backstepping controller, we regulate the rotational speed of
the DFIG, in order to maintain the TSR to its optimum value
(Cp-max) at which power extracted is maximum. This method
requires the measurement or the estimation of both the wind Fig. 9: Maximum power point tracking using backstepping
speed and the turbine speed, and the turbine optimum TSR
Fig.2 (corresponding to Cpmax) must be known. 4. Simulations Results
The angular rotation speed t-opt that corresponds to the
In this section, we will test and compare the performance
optimum value of TSR (Cp−max ) is given by:
of the different controllers presented in the previous section.
 Cp max We have made three simulations for each, to conclude on the
 t opt  R (40)
V1 convergence properties and stability of the overall closed loop
And according to (6) we can write: system, disturbance rejection and robustness.
The proposed DFIG control has been simulated using
 méc ref   t opt .G (41)
Matlab/Simulink. DFIG parameters are those in appendix. In
All recent models of digital MPPT controllers available order to satisfy the convergence and stability conditions, the
are microprocessors controllers. In this work, we used a
backstepping algorithm to develop the control law allowing adjustment parameters of our backstepping approach are
the system to follow the desired trajectory. We have: chosen as follows: k1= 80000, k2=5000, k3=90000 and
k4=6000 to satisfy convergence condition.
dméc
J  Cg  Cem  f .méc (42)
dt 4.1. Dynamic performances
 Step 1: Computation of the control variable
The modelling of the wind profile requires climatic and
We define e that represents the error variable: geographical data of the concerned site, as well as the period
   of the year concerned by the study. Therefore, several
e()   ref    e()   ref   (43) researches have been carried out. In this work, the wind profile
We select the quadratic Lyapunov function: is modelled in deterministic form by a sum of several
harmonics, around an average speed Fig.10:
1 2
V1 
 b k w v t 
e (44) i
2 V( t )  A  a . sin (49)
k 1 k

1312
INTERNATIONAL JOURNAL of RENEWABLE ENERGY RESEARCH
M. Nadour et al., Vol.7, No.3, 2017

Where A is the mean value of wind speed, 𝑎𝑘 and 𝑏𝑘 . 𝑤𝑣 are In brief, for the first simulation, we can observe that with
respectively the amplitude and the pulsation of the harmonic both backstepping and PI controller, the output of our systems
of order k. converges perfectly to their reference values of the active and
reactive power (Ps-ref, Qs-ref) coming from the control in
MPPT operation conditions of the wind turbine, all tracking
errors converge to zero asymptotically. The wind captured
active power has the same shape as the wind profile applied
(it is considered negative because it is a generated power)
However, the obtained results shows the superiority of the
backstepping approach in terms of the response time
(TPI=0.8sec, TBS=0.005sec). On the other hand, in the context
of this test, the performance of these regulators can be
considered equivalent.

Fig. 10: Wind speed (m/s)

The wind conversion system model includes: wind
turbine, Double-fed induction generator (DFIG), two power
converters (the rotor side converter and the grid side
converter) controlled by the space vector PWM, used to
connect the rotor of DFIG to the grid Fig.1.
The blades orientation angle is fixed β=2°, a
characteristics resulting from a real measurements on a real
wind turbines show the this value is valid for power variation
ranging from 0 to approximately 1.5Mw (rated power) [14].
Beyond this limit, this angle must be controlled in order to Fig. 12 :Active Power
maintain the power generated by the DFIG constant.
The maximum power point tracking “MPPT” technique
described in section 3, has as shown in Fig.11 kept the power
coefficient Cp at its maximum value which corresponds for
β=2° to Cp-max=0.437.

Fig. 13: Reactive Power

4.2. Disturbance rejection

A good disturbance rejection is characterized by a quick
return of the error to zero. Therefore, in this case, we use a
Fig. 11: Power coefficient β=2° Cp-max=0.44; opt= 10.2 high fluctuations wind speed profile to observe the capability
The active power Ps-ref is a function of the wind speed, of our controllers to act against this kind of disturbance.
based on the open loop MPPT test. The reactive power Qs-ref The differences between PI regulator and the proposed
is maintained at zero to keep a unitary power factor [8]. backstepping approach appears clearly in this simulation
Simulation results of the overall model of wind-energy Fig.14 and Fig.15. Unlike the backstepping the PI regulators
conversion system are presented in Fig.12 and Fig.3. By these are very sensitive to the perturbation that we creates, we can
figures, we illustrate the active and reactive power variations. remark a significant deviation from the reference values
especially in reactive power control figure Fig.15. Now that,

1313
INTERNATIONAL JOURNAL of RENEWABLE ENERGY RESEARCH
M. Nadour et al., Vol.7, No.3, 2017

the synthesis of this type of controllers is based on transfer Fig. 16 and Fig.17 shows that the influence of parameter
functions without expecting any disturbance. On the other variations is higher with the PI controller than with
hand, backstepping controller shows excellent rejection of backstepping controller. The response time of the PI controller
disturbance and a satisfactory tracking performance has been significantly increased. Whereas the backstepping
respectively of active and reactive power. control scheme provide a good behaviour in the case of a large
parametric variation.

Fig. 14: Active Power

Fig. 16: Active Power

Fig. 15: Reactive Power

Fig. 17: Reactive Power
4.3. Robustness
5. Conclusion
In the previous tests, both PI and backstepping regulators
were synthesized assuming that the machine parameters are This work has been devoted to the modelling and control
fixed. In fact, several physical phenomena cause variations of design of DFIG integrated into a wind energy conversion
these parameters e.g. (a rise in temperature increases the system operating at variable speed. In the first phase, we
resistance values, inductors saturation…). Moreover, the present the modelling of the wind turbine and the DFIG based
identification of these parameters is exposed to inaccuracies on physical equations. Next, we introduced the field oriented
due to the methodology adopted and the measuring devices. control scheme of DFIG, in order to ensure a flexible and
Hence, it is particularly interesting to compare both systems independent control of its active and reactive power
performances against such a phenomena. exchanged (collected and injected) between the stator and the
grid. Then, a MPPT technique is presented using a
To demonstrate the performance of each regulator against
backstepping algorithm to control the mechanical speed, so
the model uncertainties that affect stability of the closed-loop
that the power coefficient Cp can be kept at its maximum
systems. This test consists of varying the model parameters
value. Thus, the maximum wind power is extracted. In the
used in the DFIG:
second phase, we have designed and compared two control
Rr’= 2Rr ; Ls’=1.2Ls ; Lr’=1.2Lr; strategies applied to DFIG. First, a PI regulator is developed.
Secondly, a backstepping control scheme is proposed.

1314
INTERNATIONAL JOURNAL of RENEWABLE ENERGY RESEARCH
M. Nadour et al., Vol.7, No.3, 2017

The open loop MPPT block diagram provide the active power Speed” Journal of Theoretical and Applied Information
reference for our control schemes, while the reactive power is Technology. Vol.81, No.2 2015. (Article)
maintained equal to zero, so that we keep a unitary power
[5] F. Poitiers, T. Bouaouiche, M. Machmoum., “Advanced
factor for the grid [8]. The simulation results allows us to show
control of a doubly-fed induction generator for wind
the proposed algorithms capabilities to react, in terms of
energy conversion”, Electric Power Systems Research,
tracking performances, disturbance rejection, and robustness
Vol.79, pp.1085–1096, 2009. (Conference paper)
against parametric variations. We have made three
simulations for each control scheme. All in all, from a [6] Iman. Nouira, Adel Khedher. “A Contribution to the
conceptual point of view, we noticed that the proposed Design and the Installation of a Universal Platform of a
backstepping approach provides more meaningful results Wind Emulator using a DC Motor”. International Journal
compared to the PI control. Therefore, this strategy is offers a of Renewable Energy Research, IJRER. Imen Nouira et
good candidate for controlling DFIG integrated into a wind al., Vol.2, No.4, 2012 (Article)
energy conversion system interconnected to the grid.
[7] J. Ben Alaya, A. Khedher and MF. Mimouni, “DTC and
Appendix Nonlinear Vector Control Strategies Applied to the DFIG
operated at Variable Speed”, WSEAS Transactions on
TABLE 1 environment and development, Vol. 6, No. 11, pp. 744-
Wind Turbine data 753, November 2010 (Article)
Blade radius R=35
[8] E. Mahersi, A. Khedher, MF. Mimouni, “The Wind energy
Power Coefficient Cp max=0.44
Conversion System Using PMSG Controlled by Vector
Optimal relative wind speed λ max=10.2
Control and SMC Strategies”, International Journal of
Mechanical speed multiplier G=60 Renewable Energy Research, IJRER, Vol. 2, No.4, pp.
Moment of inertia J=0.41 732-741, April 2012. (Article)
Damping coefficient F=0.017
[9] BEKAKRA, Youcef & ATTOUS, D. Ben. “Sliding mode
Density of air d=1.2
controls of active and reactive power of a DFIG with
MPPT for variable speed wind energy conversion”.
TABLE 2
Australian Journal of Basic and Applied Sciences, 2011,
Double Fed Induction Generator data vol. 5, no 12, p. 2274-2286. (Article)
Rated Power 1.5MW
Rated stator voltage vs=690V
[10] Khriss, Ali, Nasser, Tamou, et Essadki, Ahmed. “A Linear
Active Disturbance Rejection Control applied for DFIG
Nominal frequency fs=50Hz based Wind Energy Conversion System”. International
Number of pole pair P=2 Journal of Computer Science Issues (IJCSI), 2013, vol. 10,
Rotor Resistance Rr=0.021 no 2, p. 391-399. (Article)
Stator Resistance Rs=0.012
[11] Chakib, R., Essadki, A., & Cherkaoui, M. “Active
Rotor Inductance Lr=0.01367 Disturbance Rejection Control for Wind System Based on
Stator Inductance Ls=0.0137 a DFIG”. World Academy of Science, Engineering and
Mutual Inductance M=0.0135 Technology, International Journal of Electrical, Computer,
Energetic, Electronic and Communication Engineering,
References 2014, vol. 8, no 8, p. 1306-1315. (Article)
[1] B. Bousoufi, M. Karim, A. Lagrioui, M. Taoussi, A. [12] F. Kendouli, K. Nabti, K. Abed, H. Benalla.
Derouich, “Adaptive Backstepping Control of DFIG "Modélisation, simulation et contrôle d’une turbine
Generators for Variable-speed Wind Turbines system”. éolienne à vitesse variable basée sur la génératrice
International Journal of Computers and Technology, IJCT. asynchrone à double alimentation". Revue des Energies
Vol.12, No.7. 2014. (Article) Renouvelables, Vol.14, No.1, 2011. (Article)
[2] N. Khemiri, A. Khedher, M. Faouzi. “Wind Energy [13] S.E. Ben Elghali. “Modélisation et Commande d’une
Conversion System using DFIG Controlled by Hydrolienne Equipée d’une Génératrice Asynchrone
Backstepping and Sliding Mode Strategies”. International Double Alimentation”. European Journal of Elecrical
Journal of Renewable Energy Research, IJRER. Nihel Engineering. 2008. (Standards and Reports)
Khemiri et al., Vol.2, No.3, 2012. (Article)
[14] Salma El AIMANI. “Modelisation de Differentes
[3] Intissar Moussa, Adel Bouallegue, Adel Khedher. “3kW Technologies D’Eolienne Intégrées Dans un Réseau
Wind Turbine Emulator Implementation on FPGA Using Moyenne Tension”. 2004. (Standards and Reports)
Matlab/Simulink”. International Journal of Renewable
Energy Research, IJRER. I. Moussa et al., Vol.5, No.4, [15] Yousfi MESSAOUD, Synthese des Controlleurs par
2015. (Article) Backstepping de la Machine Asynchrone 2014. (Standards
and Reports).
[4] Hassan Mahmoudi, Madiha El Ghamrasni, Ahmed
Largrioui, Badre Bossoufi. “Backstepping Adaptive [16] Handan Nak, Ali Fuat Ergenc “A New Controller for
Control of DFIG Generators for Wind Turbines Variable- Variable Speed Wind Turbine Generator”. International

1315
INTERNATIONAL JOURNAL of RENEWABLE ENERGY RESEARCH
M. Nadour et al., Vol.7, No.3, 2017

Conference on Renewable Energy Research and [19] Adinda Ihsani Putri, Minho Ahn, Jaeho Choi “Speed
Applications, ICRERA, 2013. (Conference paper). Sensorless Fuzzy MPPT Control of Grid Connected
PMSG for Wind Power Generation” International
[17] Abdel Ghani Aissaoui, Ahmed Tahour, Mohamed Abid,
Conference on Renewable Energy Research and
Najib Essounbouli, Frederic Nollet “ Using Neuro Fuzzy
Applications, ICRERA, 2012. (Conference paper).
PI Techniques in Wind Turbine Control” International
Conference on Renewable Energy Research and [20] E.Aydin, A. Polat, L.T.E Ergene “Vector Control Of DFIG
Applications, ICRERA, 2013. (Conference paper). n Wind Power Application” International Conference on
Renewable Energy Research and Applications, ICRERA,
[18] M.B.C Salles, A.J.S Filho and A.P.Grilo “A Study on the
2016. (Conference paper)
Rotor Side Control of DFIG-based Wind turbine during
Voltage Sags without Crowbar System” International
Conference on Renewable Energy Research and
Applications, ICRERA, 2013. (Conference paper).

1316