Sliding Mode Control Strategy for Variable Speed Wind Turbine

Oscar Barambones and Jose Maria Gonzalez de Durana

Automatic Control and System Engineering Departament
University of the Basque Country
EUI de Vitoria.
Nieves cano 12. 1006 Vitoria. (SPAIN)
[email protected]

Abstract maximum wind power capture, maintaining the DFIG sta-

tor output voltage frequency constant and controlling the
The efficiency of the wind power conversions systems DFIG reactive power.
can be greatly improved using an appropriate control al- One of the main task of the the controller is to carry
gorithm. In this work, a robust control for variable speed the turbine rotor speed into the desired optimum speed, in
wind power generation that incorporates a doubly feed spite of system uncertainties, in order to extract the maxi-
induction generator is described. The electrical systems mum active power from the wind. This paper investigates
incorporates a wound rotor induction machine with back- a new robust speed control method for variable speed wind
to-back three phase power converter bridges between its turbines, in order to obtain the maximum wind power cap-
rotor and the grid. In the presented design it is applied ture in spite of system uncertainties [4], [7].
the so called vector control theory. The proposed con-
Basically, the proposed robust design uses the slid-
trol scheme uses stator flux-oriented control for the rotor
ing mode control algorithm to regulate both the rotor-side
side converter bridge control and grid voltage vector con-
converter (RSC). In the design it is used a vector oriented
trol. The proposed robust control law is based on a sliding
control theory in order to decouple the torque and the flux
mode control theory, that, as it is well known, presents a
of the induction machine. This control scheme leads to
good performance under system uncertainties.
obtain the maximum power extraction from the different
The stability analysis of the proposed controller under
wind speeds that appear along time.
disturbances and parameter uncertainties is provided us-
Finally, test of the proposed method based on a two-
ing the Lyapunov stability theory. Finally simulated re-
bladed horizontal axis wind turbine is conducted using the
sults show, on the one hand that the proposed controller
Matlab/Simulink software. In this test, several operating
provides high-performance dynamic characteristics, and
conditions are simulated and satisfactory results are ob-
on the other hand that this scheme is robust with respect to
tained.
the uncertainties that usually appear in the real systems.

2. System modelling
1. Introduction
The power extraction of wind turbine is a function of
Wind power has experienced a rapid global growth three main factors: the wind power available, the power
since the late 1990s. In the year 1997, there was only curve of the machine and the ability of the machine to
7480 MW of installed capacity worldwide, this capacity respond to wind fluctuation. The expression for power
was increased by about 2000 MW per year. However this produced by the wind is given by [1]:
anual growth rate has been increased until reach a rate of
1
27261 MW that were added in 2008. In this sense the Pm (v) = Cp (λ, β)ρπR2 v 3 (1)
worldwide wind power installed capacity reaches 121188 2
MW in 2008, and it is expected an annual growth rate of where ρ is air density, R is radius of rotor, v is wind speed,
31000 MW in 2009 and 38000 MW in 2010 [11]. Cp denotes power coefficient of wind turbine, λ is the tip-
Doubly Feed Induction Generator (DFIG), with vec- speed ratio and β represents pitch angle. The tip-speed
tor control applied, is widely used in variable speed wind ratio is defined as:
turbine system owing to their ability to maximize wind Rw
λ= (2)
power extraction. In these DFIG wind turbines the con- v
trol system should be designed in order to to achieve the where w is the turbine rotor speed. Therefore, if the rotor
following objectives: regulating the DFIG rotor speed for speed is kept constant, then any change in the wind speed

978-1-4244-2728-4/09/$25.00 ©2009 IEEE

will change the tip-speed ratio, leading to the change of generator (DFIG). This induction machine is feed from
power coefficient Cp , as well as the generated power out- both stator and rotor sides. The stator is directly con-
put of the wind turbine. However, if the rotor speed is ad- nected to the grid while the rotor is fed through a vari-
justed according to the wind speed variation, then the tip- able frequency converter (VFC). In order to produce elec-
speed ratio can be maintained at an optimal point, which trical active power at constant voltage and frequency to
could yield maximum power output from the system. the utility grid, over a wide operation range (from subsyn-
For a typical wind power generation system, the fol- chronous to supersynchronous speed), the active power
lowing simplified elements are used to illustrate the fun- flow between the rotor circuit and the grid must be con-
damental work principle. The system primarily consists trolled both in magnitude and in direction. Therefore,
of an aeroturbine, which converts wind energy into me- the VFC consists of two four-quadrant IGBT PWM con-
chanical energy, a gearbox, which serves to increase the verters (rotor-side converter (RSC) and grid-side converter
speed and decrease the torque and a generator to convert (GSC)) connected back-to-back by a dc-link capacitor [6],
mechanical energy into electrical energy. [7].
Driving by the input wind torque Tm , the rotor of the
wind turbine runs at the speed w. The transmission out- 3 Rotor Side Converter Control
put torque Tt is then fed to the generator, which produces
a shaft torque of Te at generator angular velocity of we . When the wind turbine generator (WTG) operates in
Note that the rotor speed and generator speed are not the the variable-speed mode, in order to extract the maxi-
same in general, due to the use of the gearbox. mum active power from the wind, the shaft speed of the
The mechanical equations of the system can be charac- WTG must be adjusted to achieve an optimal tip-speed
terized by [8]: ratio λopt , which yields the maximum power coefficient
Cpmax , and consequently the maximum power [2]. The
Jm ẇ + Bm w = Tm + T (3) value of the λopt can be calculated from the maximum of
Je ẇe + Be we = Tt + Te (4) the power coefficient curves versus tip-speed ratio, which
Tt we = −T w (5) depends of the modeling turbine characteristics. Then,
the corresponding optimal generator speed command for
where Jm and Je are the moment of inertia of the tur- maximum wind power tracking is determined by:
bine and the generator, Bm and Be are the viscous friction
coefficient of the the turbine and the generator, Tm is the λopt · v
w∗ = (12)
wind generated torque in the turbine, T is the torque in the R
transmission shaft before gear box, Tf is the torque in the
Therefore, the RSC control scheme should be designed
transmission shaft after gear box, and Te is the the gener-
in order to regulate the wind turbine speed in order to
ator torque, w is the angular velocity of the turbine shaft
track the optimal wind turbine speed reference w∗ for
and we is the angular velocity of the generator rotor.
maximum wind power extraction. In the DFIG-based
The relation between the angular velocity of the turbine
wind generation system, these objective are commonly
w and the angular velocity of the generator we is given by
achieved by electrical generator rotor current regulation
the gear ratio γ:
we on the stator-flux oriented reference frame.
γ= (6)
w In the stator-flux oriented reference frame, the d-axis
Then, using equations 3, 4, 5 and 6 it is obtained: is aligned with the stator flux linkage vector ψs , and then,
ψds =ψs and ψqs =0 . This yields the following relation-
J ẇ + Bw = Tm − γTe (7) ships [3]:
with Lm iqr
iqs = (13)
J = Jm + γ 2 Je (8) Ls
Lm (ims − idr )
B = Bm + γ 2 Be (9) ids = (14)
Ls
From equations 1 and 2 it is deduced that the input −Lm ims iqr
Te = (15)
wind torque is: Ls
3 ws L2m ims (ims − idr )
Pm (v) Pm (v) Qs = (16)
Tm (v) = = λv = kv · v 2 (10) 2 Ls
w R diqr
vdr = rr idr + σLr − sws σLr iqr (17)
where dt
3
1 R diqr
kv = Cp ρ π (11) vqr = rr iqr + σLr (18)
2 λ dt
 
Now we are going to consider the system electrical σLr idr + L2m ims
equations. In this work it is used a double feed induction +sws (19)
Ls

2
where To compensate for the above described uncertainties
vqs − rs iqs that are present in the system, it is proposed a sliding con-
ims = (20) trol scheme. In the sliding control theory, the switching
ws Lm
gain must be constructed so as to attain the sliding condi-
L2
σ = 1− m (21) tion [10]. In order to meet this condition a suitable choice
Ls Lr of the sliding gain should be made to compensate for the
Since the stator is connected to the grid, and the influ- uncertainties.
ence of the stator resistance is small, the stator magnetiz- Now, we are going to define the sliding variable S(t)
ing current (ims ) can be considered constant [6]. There- with an integral component as:
fore, the electromagnetic torque can be defined as follows:  t
Te = −KT iqr (22) S(t) = e(t) + (k + a)e(τ ) dτ (32)
0

where KT is a torque constant, and is defined as follows: where k is a constant gain.

Lm ims Then the sliding surface is defined as:
KT = (23)
Ls  t
Then, from equations 7 and 22 it is deduced that the S(t) = e(t) + (a + k)e(τ ) dτ = 0 (33)
0
wind turbine speed can be controlled by regulating the
q-axis rotor current components (iqr ) while equation 16 Now, we are going to design a variable structure speed
indicates that the stator reactive power (Qs ) can be con- controller in order to control the wind turbine speed.
trolled by regulating the d-axis rotor current components,
(ids ). Consequently, the reference values of iqr and idr u(t) = −k e(t) − β sgn(S) (34)
can be determined directly from wr and Qs references.
where the k is the constant gain defined previously, β is
From equations 7 and 22 it is obtained the following
the switching gain, S is the sliding variable defined in eqn.
dynamic equation for the system speed:
(32) and sgn(·) is the signum function.
1 In order to obtain the speed trajectory tracking, the fol-
ẇ = (Tm − γKT iqr − Bw) (24)
J lowing assumptions should be formulated:
= −aw + f − biqr (25)
(A 1) The gain k must be chosen so that the term (k + a)
where the parameters are defined as: is strictly positive, therefore the constant k should be
B γKT Tm k > −a.
a= , b= , f= ; (26)
J J J (A 2) The gain β must be chosen so that β > |d(t)| for
Now, we are going to consider the previous dynamic all time.
equation (25) with uncertainties as follows: Note that this condition only implies that the uncer-
ẇ = −(a + a)w + (f + f ) − (b + b)iqr (27) tainties of the system are bounded magnitudes.

where the terms a, b and f represents the uncertain-

ties of the terms a, b and f respectively. Theorem 1 Consider the induction motor given by equa-
tion (27). Then, if assumptions (A 1) and (A 2) are ver-
Let us define the speed tracking error as follows:
ified, the control law (34) leads the wind turbine speed
e(t) = w(t) − w∗ (t) (28) w(t), so that the speed tracking error e(t) = w(t)−w∗ (t)
tends to zero as the time tends to infinity.
where w∗ is the rotor speed command.
Taking the derivative of the previous equation with re- The proof of this theorem will be carried out using the
spect to time yields: Lyapunov stability theory.
Proof : Define the Lyapunov function candidate:
ė(t) = ẇ − ẇ∗ = −a e(t) + u(t) + d(t) (29)
1
where the following terms have been collected in the sig- V (t) = S(t)S(t) (35)
2
nal u(t),
Its time derivative is calculated as:
u(t) = f (t) − b iqr (t) − a w∗ (t) − ẇ∗ (t) (30)
V̇ (t) = S(t)Ṡ(t)
and the uncertainty terms have been collected in the signal = S · [ė + (k + a)e]
d(t),
= S · [(−a e + u + d) + (k e + a e)]
d(t) = −a w(t) + f (t) − b iqr (t) (31) = S · [u + d + k e]

3
 = S · [−k e − β sgn(S) + d + k e] In this example simulation it is used a variable speed
= S · [d − β sgn(S)] wind farm with a rated power of 9 MW. The farm consists
of six 1.5 MW wind turbines connected to a 575 V bus
≤ −(β − |d|)|S|
line. The wind turbines use a doubly-fed induction gener-
≤0 (36) ator (DFIG) consisting of a wound rotor induction gen-
erator and an AC/DC/AC IGBT-based PWM converter.
It should be noted that the eqns. (32), (29) and (34) and
The stator winding is connected directly to the 60 Hz grid
the assumption (A 2) have been used in the proof.
while the rotor is fed at variable frequency through the
Using the Lyapunov’s direct method, since V (t) is AC/DC/AC converter.
clearly positive-definite, V̇ (t) is negative definite and
V (t) tends to infinity as S(t) tends to infinity, then the The system has the following mechanical parameters.
equilibrium at the origin S(t) = 0 is globally asymptoti- The combined generator and turbine inertia constant is
cally stable. Therefore S(t) tends to zero as the time tends J = 5.04s expressed in seconds, the combined viscous
to infinity. Moreover, all trajectories starting off the slid- friction factor B = 0.01pu in pu based on the generator
ing surface S = 0 must reach it in finite time and then rating and there are three pole pairs [9].
will remain on this surface. This system’s behavior once In this simulation examples it is assumed that there is
on the sliding surface is usually called sliding mode [10]. an uncertainty around 20 % in the system parameters, that
When the sliding mode occurs on the sliding surface will be overcome by the proposed sliding control.
(33), then S(t) = Ṡ(t) = 0, and therefore the dynamic
Finally, the following values have been chosen for the
behavior of the tracking problem (29) is equivalently gov-
controller parameters, k = 100, β = 30, λ = 70 and
erned by the following equation:
γ = 25.
Ṡ(t) = 0 ⇒ ė(t) = −(k + a)e(t) (37) In this simulation it is used a variable wind speed, and
figure 1 shows the wind speed used in this simulation. As
Then, under assumption (A 1), the tracking error e(t) it can be seen in the figure, the wind speed varies between
converges to zero exponentially. 0m/s and 20m/s.
It should be noted that, a typical motion under sliding
mode control consists of a reaching phase during which
trajectories starting off the sliding surface S = 0 move
25

toward it and reach it in finite time, followed by sliding

phase during which the motion will be confined to this 20

surface and the system tracking error will be represented

by the reduced-order model (eqn. 37), where the tracking
wind speed (m/s)

15
error tends to zero.
Finally, the torque current command, i∗qr (t), can be ob- 10
tained from equations (34) and (30):
1
i∗qr (t) = [k e + β sgn(S) − a w∗ − ẇ∗ + f ]
5
(38)
b
0
Therefore, the proposed variable structure speed con- 0 10 20 30 40 50
Time (s)
trol resolves the wind turbine speed tracking problem for
variable speed wind turbines in the presence of uncertain- Figure 1. Wind speed
ties. This wind turbine speed tracking let us obtain the
maximum wind power extraction for all wind speeds.
Figure 2 show the reference (dashed line) and the real
4 Simulation Results rotor speed (solid line). As it may be observed, after a
transitory time in which the sliding mode is reached, the
In this section we will study the variable speed rotor speed tracks the desired speed in spite of system un-
wind turbine regulation performance using the proposed certainties. In this figure, the speed is expressed in the
sliding-mode field oriented control scheme. The objective per unit system (pu), that is based in the generator syn-
of this regulation is to maximize the wind power extrac- chronous speed ws = 125.60rad/s.
tion in order to obtain the maximum electrical power. In Figure 3 shows the generated active power, whose
this sense, the wind turbine speed must be adjusted con- value is maximized by our proposed sliding mode con-
tinuously against wind speed. trol scheme. As it can be observed in this figure, at time
The simulation are carried out using the Mat- 32.5s the mechanical power (and therefore the generated
lab/Simulink software and the turbine model is the one active power) should be limited by the pitch angle so as
provided in the SimPowerSystems library [9]. not to exceed the rated power of this system.

4
 1.25
At wind speeds less than the rated wind speed, the
speed controller seeks to maximize the power according
1.2
to the maximum coefficient curve. As result, the variation
Reference and Real Generator Speed (pu)

1.15
of the generator speed follows the slow variation in the
1.1 wind speed. At large wind speeds, the power limitation
1.05 controller sets the blade angle to maintain rated power.
1 Finally, by means of simulation examples, it has been
0.95
shown that the proposed control scheme performs reason-
ably well in practice, and that the speed tracking objective
0.9
w
is achieved in order maintain the maximum power extrac-
0.85 *
w tion under system uncertainties.
0.8

0.75
0 10 20 30 40 50
Acknowledgment
Time (s)

The authors are very grateful to the Basque Govern-

Figure 2. Reference and real rotor speed
ment by the support of this work through the project
S-PE07UN04, to the UPV/EHU by its support through
project GUI07/08.
9

8 References
7 [1] F ERNANDO D. B IANCHI , H RNAN D E BATTISTA AND
R ICARDO J. M ANTZ 2007 Wind Turbine Control Systems
Generated Power (MW)

6 Advances in Industrial Control Series. Springer.

[2] G.M. J OSELIN H ERBERT, S. I NIYAN , E. S REEVALSAN ,
5
S. R AJAPANDIAN 2007, A review of wind energy tech-
4 nologies, Renewable and Sustainable Energy Reviews.,
11, pp. 11171145.
3 [3] YAZHOU L EI , A LAN M ULLANE , G ORDON L IGHTBODY,
AND ROBERT YACAMINI 2006, Modeling of the Wind
2 Turbine With a Doubly Fed Induction Generator for Grid
Integration Studies, IEEE Trans. on Energy Conversion,
1 vol. 21, no. 1, pp. 257-264.
[4] M ULLER , S., D EICKE , M., DE D ONCKER , R.W., 2002,
0
0 10 20 30 40 50 Doubly fed induction generator system for wind turbines,
Time (s) IEEE Industry Application Magazine., 8, pp. 26-33.
[5] D. W. N OVOTNY AND T. A. L IPO, 2000 Vector Control
Figure 3. Generated active power and Dynamics of AC Drives. Oxford, U.K.: Oxford Univ.
Press, 2000.
[6] R. P ENA , J. C. C LARE , AND G. M. A SHER 1996 Doubly
5 Conclusion fed induction generator using back-to-back PWM con-
verters and its application to variablespeed wind-energy
In this paper a sliding mode vector control for a dou- generation Proc. Inst. Elect. Eng., vol. 143, no. 3, pp.
bly feed induction generator drive, used in variable speed 231241.
wind power generation is described. It is proposed a new [7] W EI Q IAO , W EI Z HOU , J OS M. A LLER , AND RONALD
variable structure control which has an integral sliding G. H ARLEY 2008, Wind Speed Estimation Based Sen-
surface to relax the requirement of the acceleration signal, sorless Output Maximization Control for a Wind Turbine
that is usual in conventional sliding mode speed control Driving a DFIG IEEE Trans. on Power Electronics, vol.
techniques. Due to the nature of the sliding control this 23, no. 3, pp. 1156-1169.
[8] Y.D. S ONG , B. D HINAKARAN , X.Y. BAO, 2000,
control scheme is robust under uncertainties that appear
Variable speed control of wind turbines using nonlinear
in the real systems. The closed loop stability of the pre-
and adaptive algorithms, Journal of Wind Engineering
sented design has been proved through Lyapunov stability and Industrial Aerodynamics., 85, pp. 293-308.
theory. [9] SimPowerSystems 5. Users Guide, The MathWorks
The implemented control method allows the wind tur- [10] U TKIN V.I., 1993, Sliding mode control design principles
bine to operate with the optimum power efficiency over a and applications to electric drives, IEEE Trans. Indus.
wide range of wind speed. Electro., 40, 26-36.
The simulations show that the control method success- [11] W ORLD W IND E NERGY A SSOCIATION 2009 World
fully controls the variable speed wind turbine efficiently, Wind Energy Report 2008 WWEA Head Office. Bonn,
within a range of normal operational conditions. Germany.