Sliding mode control of a doubly fed induction

generator for wind energy conversion systems

M. Allam * Y. Djeriri
IRECOM-laboratory (Interaction REseau ICEPS- Laboratory (Intelligent Control & Electrical Power
COnvertisseur Machines) Systems)
University Djillali Liabes of Sidi Bel-Abbes, Algeria University Djillali Liabes of Sidi Bel-Abbes, Algeria
* allam2010@live.fr ydjeriri@yahoo.fr

H. Mesai Ahmed I. Dehiba

ICEPS- Laboratory (Intelligent Control & Electrical Power IRECOM-laboratory (Interaction REseau
Systems) COnvertisseur Machines)
University Djillali Liabes of Sidi Bel-Abbes, Algeria University Djillali Liabes of Sidi Bel-Abbes, Algeria
hamzamesai2@gmail.com de_imene@yahoo.fr

Abstract— This work presents a sliding mode control for The control of DFIG has been well developed, one of the
wind energy conversion system based on a Doubly Fed Induction most popular control techniques is the vector control (VC) or
Generator (DFIG). The power transfer between the stator and flux orientation control (FOC) associated with classical (PI)
the network is achieved by acting on the rotor signals via a regulators. The active and reactive power can be regulated by
bidirectional converter. Our objective is to apply this control to separately controlling the d-axis and q-axis rotor current
command independently the active and the reactive power
components. Actually the main drawback of FOC is the
generated by the DFIG. Simulation results on a 10 kW DFIG
system are provided to demonstrate the effectiveness and unstable system and sensitivity to machine parameters [4], [5].
robustness of the proposed control strategy during variations of In the area of the control of the electric machines, the
active and reactive power and machine parameters, based on research works are oriented more and more towards the
software MATLAB / SIMULINK.
application of the modern control techniques. These
techniques are the fuzzy control, the adaptive control, the
Keywords— Doubly Fed Induction Generator, flux orientation variable structure control etc. The aim of this paper is to apply
control, sliding mode control , wind energy, active and reactive a Variable Structure Control (VSC), called also the sliding
power. mode control (SMC) to a DFIG [13].
In recent years, the sliding mode control (SMC)
I. INTRODUCTION
methodology has been widely used for robust control of
Wind energy is one of the most growing uses of the nonlinear systems. Sliding mode control, based on the theory
renewable energy sources since it is clean and available, of variable structure systems (VSS), has attracted a lot of
moreover, because of its reduced cost and improved research on control systems for the last two decades [11]. It
techniques [6]. achieves robust control by adding a discontinuous control
signal across the sliding surface, satisfying the sliding
Now most wind turbines are equipped with a double-fed
condition. [2]
induction generator (DFIG) due to noticeable advantages: the
variable speed generation (_30% around the synchronous The use of this control mode can be justified by the high
speed), the decoupled control of active and reactive powers, performances required by DFIG and the robustness of the
the improvement of the power quality, and the low cost [1]. controller [15].
The stator of the DFIG is usually connected directly to the
In this work, first, a wind turbine system is presented and
three-phase grid; the rotor is also connected to the grid but via
its characteristics are depicted to estimate its dynamics and
a transformer and two back-to-back converters [3] ―Fig.1‖.
performances in different operating conditions. Then, a SMC
Usually, the rotor-side converter controls the active and
of the DFIG used to control independently the active and
reactive power and the grid side converter controls the DC-
reactive powers between the stator of and the grid is proposed
link voltage and ensures operation of the converter at a unity
and tested on a wind turbine equipped with a DFIG of 10 kW
power factor [14].
based on software MATLAB / SIMULINKTM.
 Wind P, Q stator Transformer 12000 Pt -optimal

Power of the turbine Pt (W)

Γturb 10000
11m/s
Grid 10m/s
Gearbox 3~ 8000
09m/s
Inverter Rectifier
Ωturb DFIG 6000
08m/s
07m/s
Wind Turbine 4000
06m/s

P, Q rotor 2000 05m/s

DC-Bus
04m/s
DFIG-Side Grid-Side 0
0 500 1000 1500 2000
Control Control
Mechanical speed of the DFIG Ωm (rpm)
Fig.1. Schematic diagram of the system with a DFIG.
Fig.3. Power-Speed characteristics of the wind turbine.
II. MODEL OF WIND TURBINE
The relation between the wind speed and mechanic power, III. MODELING OF THE DFIG
delivered by the wind turbine, can be described by the The dynamic model of doubly fed induction machine in d-
following equation: [14] q frame can be represented by the following equations: [7], [6]
R v
2 3
A. Electrical equations:
Pturb  C p
 ,   (1)  d
 RsIs    s
ds
2 V
 ds qs
dt
With: 
 d (4)
  s
qs
V qs  R s I qs 
   turb R / v (2) 
 dt
ds


V d
 R r I dr       qr
dr


dr s r
where: Cp – power coefficient; λ– relative speed; β– pitch dt

angle (deg); R– radius of turbine; v – wind speed (m/s); Ωturb– d
V      dr
qr
 R r I qr 

turbine speed (rd/s); ρ– air density (1.225 kg.m-3).
qr s r
 dt
According to Betz, the power coefficient Cp cannot be
B. Magnetic equations:
greater than 16/27 or 0.59 [7], [8].
 ds
 LsI ds
 MI dr
The torque produced by the turbine is expressed in the  (5)
 qs
 LsI qs
 MI qr
following way: 
  Lr I  MI
 turb  Pturb  turb
dr dr ds
(3)   Lr I  MI
 qr qr qs
A typical relationship between Cp and λ is shown in
―Fig.2‖. It is clear from this figure that there is a value of λ for C. Electromagnetic equation:
which Cp is maximized thus maximizing the power for a given Where the electromagnetic torque Γem: can be written as a
wind speed. Because of the relationship between Cp and λ, as function of stator fluxes and rotor currents:
the turbine speed that gives a maximum output power.
I 
M
 em  P qr
 ds  I dr  qs (6)
0.5 Ls
where: s / r are stator/rotor subscript; V / I –voltage/current;
Power coefficient Cp

β=0
0.4
2°
–flux; R – resistance; M – mutual inductance; Lr / Ls –
0.3
8°
inductance; r /s rotor/stator pulsation; P – number of the
4°
pairs of poles.
0.2 10°
6°
IV. Field Oriented Control of the DFIG
0.1
In order to easily control the production of electricity by
0
the wind turbine, we will carry out an independent control of
0 5 10 15 20 active and reactive powers by orientation of the stator flux [5],
Tip speed ratio λ
[8] and [4].
Fig. 2. Power coefficient Cp as a function of β and λ.
By choosing a reference frame linked to the stator flux,
―Fig.3‖ shows the wind speed effect on the mechanical rotor currents will be related directly to the stator active and
power for the 10kW turbine used in this work. reactive power. An adapted control of these currents will thus

2
permit to control the power exchanged between the stator and The resulting bloc diagram of the DFIG is presented in
the grid. If the stator flux is linked to the d-axis of the frame ―Fig.5‖.
―Fig.4‖ we have: [9], [15] gMVs /Ls

 ds   s ;  qs  0 (7) +
_ 1 Iqr Ps
 
2

 p L 
M
 - MVs/Ls
Vqr _ R r  r
L

 s 
Rotor axis
Stator axis gws (Lr−M 2∕Ls)
Reference frame Vs2/Lsws
2
gws (Lr−M ∕Ls)
B d
q s Vdr + 1 +
Qs
 
2

I qs I ds M -MVs/Ls
R  p L  
+ r  r
L
 +
  Idr
r
s

b s
Is a Fig. 5. Block diagram of the DFIG.
 sl
V. APPLICATION SLIDING MODE CONTROL
r A The sliding mode control knows a big success these last
years. It is due to the implementation simplicity and the
Fig.4. Diagram of the stator flux oriented with d-axis. robustness with regard to the system uncertainties and the
external disturbances.
The electromagnetic torque ―(6)‖ is then written: The SMC consists to return the state trajectory towards the
M sliding surface and to develop it above, with a certain
 em  p I qr  ds
(8) dynamics up to the equilibrium .Its design consists mainly to
Ls determine three stages. [10], [11], [13]
By substituting ―(7)‖ in ―(5)‖ the following rotor flux
equations are obtained:  The Switching Surface Choice
  ds  L s I ds  MI dr For a non-linear system represented by the following
 (9) equation:
0  L s I qs  MI
 qr

Using ―(9)‖ a relation between the stator and rotor currents X  f X ,t   g X ,t  u X ,t ; X  R ,u  R
n
(14)
can be established:
 M s Where: f(X, t ) , g (X, t ) are two continuous and uncertain
 I ds   I dr  non-linear functions, supposed limited.
 Ls Ls
 (11)
We take the general equation to determine the sliding
I   M I
surface, proposed by J.J. Slotine [12], given by:

qs qr
Ls n 1
 d 
S X  e ; e  X  X
d
Or from the rotor currents by:   a (15)
 M  dt 
P  V s I qr where: e – error on the signal to be adjusted; a – positive
 s
 Ls coefficient; n – system order; Xd – desired signal; X – state
 (12) variable of the control signal.
 Q  V M s
I dr  V s

s s
Ls Ls  Convergence Condition
Equations showing the relationship between the rotor
currents and voltages are established and will be applied to The convergence condition is defined by the Lyapunov
control the generator equation; it makes the surface attractive and invariant:

   dI dr   S X S X  
2 2
M M 0 (16)
 V dr  R r I dr   L r  

 g s  Lr 

 I qr

  Ls  dt  Ls   Control Calculation

(13)
  dI qr  
2 2
 M M MV s The control algorithm is defined by the relation:
V qr  R r I qr   L r    g s  Lr   I dr  g
    
  Ls  dt  Ls  Ls u  u
eq
u
n
(17)

3
  u Sat  S ( X ) /   To control the power, the surface expression of the reactive
n m ax
with: u (18)
power control has the form:
S (Q )  (Q s  Qs)
ref
(27)
 sign  S  if S 
Sat  S ( X ) /     (19) Taking its derivative and replacing it in the active power Qs
S  ―(11)‖ we get:
 S / if
M 
S ( Q )  ( Q s  (  V s
ref
I dr ) (28)
where: u – control signal; ueq – equivalent control signal; un – Ls
switching control term; sat (S(X)/φ) – saturation function; φ –
Replacing Vdr by V dr  V dr  V dr , the control appears clearly
eq n
threshold width of the saturation function.
in the following equation:
A. Active power control . ref
M
To control the power, the surface expression of the active S ( Q )  (Q s
Vs ( V dr  R r I dr ) (29)
power control has the form: Ls Lr

S (P )  ( Ps
ref
 Ps ) (20) During the sliding mode and in steady state, we have:
S (Q )  0 , S ( Q )  0 , et V dr  0
n
Taking its derivative and replacing it in the active power Ps donc
―(11)‖ we get: The equivalent control amount V dr is found from the previous
eq

M  equations and written as:

S ( P )  ( Ps
ref
 Vs I qr ) (21)
Ls Ls Lr
 Q s
eq ref
V dr   R r I dr (30)
Replacing Vqr by V qr  V qr  V qr , the control appears
eq n
VsM

clearly in the following equation: During the convergence mode, so that the condition
M S ( Q ) S ( Q )  0 is verified, we set:
S ( P )  ( P s  V s
ref eq n
(( V qr  V qr )  R r I qr ) (22)
Ls Lr
To illustrate the control performance of SMC, sliding
With:   1  M / L s L r ,  – Leakage coefficient.
2
mode controller applied to DFIG, a block diagram of the
During the sliding mode and in steady state, we have: system proposed in ―Fig.6‖.
S ( P )  0, S ( P )  0 , V qr  0
n
donc et
eq
The equivalent control amount V qr is found from the Grid
3~
previous equations and written as: P I Rectifier

Ls Lr Q V
 P s
eq ref
V qr   R r I qr (23)
Qs
Vdr
VsM Ps
Direct
dq
PWM
Qs-ref DFIG
control Technique
During the convergence mode, so that the condition abc Wind
Ps-ref Vqr
S ( P ) S ( P )  0 is verified, we set:
Inverter

Gearbox

M
S ( P )   V s
n Anemometer
V qr (24) β
Ls Lr Pitch
control
Wind speed
And consequently, the switching term is given:
 R v
2 3

P s  ref  C  ,  
V qr  KV
n p

qr
sign ( S ( P )) (25) 2

To verify the system stability condition, the parameter Fig.6. Schematic diagram of SMC strategy for DFIG.
KVqr must be positive.
To reduce any possible overshoot of the reference voltage VI. Validation of the Proposed Control
Vqr, it is often useful to add a Voltage limiter which is In this part, simulations are investigated with a 10 kW
expressed by: DFIG connected to a 400V/50Hz grid (appendix), by using the
lim max
Matlab/Simulink. The both control strategies FOC and SMC
V qr  V qr sat(P) (26) are simulated, tested and compared in terms of power reference
tracking, robustness against machine parameter variations.
B. Réactive power control

4
 According to the ―Fig.7- (a) and (b)‖ and ―Fig.8- (a) and 0

(b)‖ we can see that the power references are followed by the P
s-réf

Active power Ps (W)

-2000
generator both for active and reactive power. Ps
-4000
The active power of the stator side is negative, which (a)
means that the network in this case is a receiver of the energy -6000

supplied by the DFIG, reactive power is adjusted according to

the network requirements. -8000

Ps-réf -10000

-2000

P -12000
s 0 1 2 3 4 5
Active power Ps (W)

-4000 Time (s)

(a) 5000
Q

Réactive power Qs (VAR)

-6000
s-ref

-8000
Qs (b)
-10000

-12000
0 1 2 3 4 5

Time (s)
5000

Qs-réf
Réactive power Qs (VAR)

-5000
0 1 2 3 4 5
Qs Time (s)
(b) 150

0
Rotor current (Zoom) Ir (A)

100
(c)

50

-5000
0 1 2 3 4 5
-50
Time (s)
150
-100

(c)
Rotor current (Zoom) Ir (A)

100
-150
0 0.2 0.4 0.6 0.8 1 1.2

50
Time (s)
50

0
40
Stator current (Zoom) Is (A)

(d)
30
-50

20

-100 10

0
-150
0 0.2 0.4 0.6 0.8 1 1.2
-10

Time (s) -20

50
-30
Stator current (Zoom) Is (A)

40
(d) -40
30
-50
20 0 0.2 0.4 0.6 0.8 1 1.2

10
Time (s)
0
Fig.8. System responses with sliding mode controller (SMC).
-10

-20 The simulation results in ―Fig.7-(c) and (d)‖ and ―Fig.8-(c)

-30 and (d)‖ show the response of rotor and stator currents of PI
-40 controller and SMC respectively. The changeover from
-50
0 0.2 0.4 0.6 0.8 1 1.2
subsynchronous to supersynchronous speed crossing the
Time (s) synchronous speed in 0.12s is observed to be smooth without
Fig. 7. System responses with conventional controller (PI). any transients in the rotor current waveform.

5
 P P (SMC) P (PI) APPENDIX
s-réf s s
0
Table1. Doubly fed induction generator parameters
-9000
Rated power, Pn 10 kW
-2000
Stator rated voltage, Vs 230/400 V
Active power Ps (W)

-10000

-4000 Rated current, In 20/30A

1.95 2 2.05
Stator rated frequency, f 50 Hz
-6000

(a) Stator resistance, Rs 0.455 Ω

-8000
Rotor resistance, Rr 0.19 Ω
-10000 Rotor inductance, Ls 0.07 H
Inductance rotorique ,Lr 0.0213 H
-12000
0 1 2 3 4 5
Mutual inductance, M 0.034 H
Time (s)
Number of pair of poles, p 2
Q Q (SMC) Q (PI)
s-réf s s Moment of inertia, J 0.031 kg.m2
5000

-3500 viscous friction, fr 0.00114 kg.m2/s

Réactive power Qs (VAR)

-4000

1 1.05 REFERENCES
0 [1] S.A.E. Ardjoun,M. Abid,―Fuzzy sliding mode control applied to a
doubly fed induction generator for wind turbines‖, Journal of Electrical
(b) Engineering & Computer Sciences,pp.1-14,2015.
[2] Z. Boudjema, A. Meroufel and Y. Djerriri, ―Nonlinear control of a
doubly fed induction generator for wind energy conversion‖, Journal of
Electronic and Computer Engineering, vol.1, no.6, pp.28-35, 2013.
-5000
0 1 2 3 4 5 [3] Y. djeriri, A. Meroufel, M. Allam,― Artificial neural network-based
Time (s) robust tracking control for doubly fed induction generator used in wind
energy conversion systems‖, JARST, vol.2, no.1, pp.173-181, 2015.
Fig.9. Response of system with 20% (Ls, Lr, M). [4] M. Arifujjaman, ―Vector control of a DFIG based wind turbine‖, Journal
of Electrical & Electronics Engineering, Vol.2, no.9, pp.1057-
From the results obtained by ―Fig.9- (a) and (b)‖ it can be 1065,2009.
seen that the control of active and reactive power by sliding [5] A. Boyette, ―Direct and indirect control of a Doubly Fed Induction
mode controller provides good performance and robustness Generator wind turbine including a storage unit‖, In: 32th edition of
Industrial Electronics Conference IECON’, IEEE, France, pp. 2517-
compared to PI controller. 2522, 2006.
[6] K. Mouilah, M. Abid, A.Naceri, M.Allam, ―Fuzzy control of a doubly
VII. CONCLUSION fed induction generator for wind turbines‖, Journal of Electrical
Engineering, JEE, Romania, , vol.14, no.4, pp.01-05, 2014.
The modeling, the control and the simulation of a wind [7] F. Poitiers, ― Advanced control of a doubly-fed induction generator for
energy conversion system based on a doubly fed induction wind energy conversion‖,Electric Power Systems Research,no.79, pp.
generator (DFIG) connected directly to the grid by the stator 1085-1096, 2009.
and fed by a converter on the rotor side has been presented in [8] A. Djoudi,― Exploitation robuste d’une éolienne à base d’une machine
this paper. Our objective was the implementation of a sliding asynchrone à double alimentation‖, vol.15,no.4, pp.20-29,2012.
mode control method of active and reactive powers generated [9] B. Hopfensperger,―Stator-flux-oriented control of a doubly fed induction
machine with and without position encoder‖, IEEE Proc.-Electr. Power
by the stator side of the DFIG, in order to ensure of the high Applications, vol.147, no. 4, pp. 241-250, 2000.
performance and a better execution of the DFIG, and to make [10] H. Buhler ,―Réglage par mode de glissement‖, Presses Polytechniques
the system insensible with the external disturbances and the romandes, Lausanne, 1986.
parametric variations. In the first step, we started with a study [11] A. Kerboua , M. Abid,― hybrid fuzzy sliding mode control of a doubly-
of modeling on the whole system. In second step, we adopted fed induction generator in wind turbines‖, Rev. Roum. Sci. Techn. –
Électrotechn. Et Énerg, Vol.4, pp. 412–421, 2012
a vector control strategy in order to control statoric active and
[12] Slotine, J.J., Li,E,― Applied nonlinear control‖, Prence Hall, USA, 1998.
reactive power exchanged between the DFIG and the grid. In
[13] M. Allam,―Contribution à la commande robuste du système d’énergie
third step, the description of the classical sliding mode renouvlable ‖, Thèse de doctorat, de l’université de Sidi Bel-Abbes,
controller (SMC) is presented in detail. Alger, 2017.
The simulation results show the ameliorated quality of the [14] Y. Djeriri,― Commande vectorielle par les réseaux de neurones artificiels
de l’énergie d’une MADA intégrée à un système éolien ‖, Revue des
control based on the controller by sliding mode. Through Energies Renouvelables, Centre de Développement des Energies
the response characteristics, we observe good performances in Renouvelables- CDER, Algérie, vol.13, no.4, pp. 669-682, 2010.
terms of response time and reference tracking without [15] M. Allam, B. Dehiba, M. Abid, Y. Djeriri, et R.Adjoudj,― Étude
overshoots. Furthermore, the strong point of this control comparative entre la commande vectorielle directe et indirecte de la
Machine Asynchrone à Double Alimentation (MADA) dédiée à une
technique is the simplicity of implementation and robustness application éolienne ‖,Journal of Advanced Research in Science and
with respect to parameters variation of the DFIG. Technology, ISSN: 2352-9989, Vol.1, No.2, pp. 88-100, 2014.