2019 1st International Conference on Sustainable Renewable Energy Systems and Applications (ICSRESA), December 04-05, 2019, Tebessa,

Algeria

Doubly Fed Induction Generator Control in a Wind

Energy System Using a Fractional order MRAC-
Based Adaptive Control Approach

Sihem Djebbri Samir Ladaci Hanene Balaska

Departmentof Electrical Engineering Signal Processing Laboratory, UMC-1 Science and Technology Department,
20th August 1955University Constantine, 25000, Algeria. Larbi Ben Mhidi University
Skikda 21000,Algeria National Polytechnic School Constantine Oum El-Bouaghi 4000, Algeria.
Email: [email protected] Constantine, 25100, Algeria. Email: [email protected]
Email: [email protected]

Abstract - This work study proposes the implementation of a application in a medical drug administration [13], velocity
fractional order model reference adaptive control (FOMRAC) control for an electric vehicle using [14] and hybrid system
design for a doubly fed induction generator (DFIG) control of using FO Integrals [15]...etc.
active and reactive power in wind energy systems. In a first step, This work describes the implementation of MRAC of the
we introduce classical adaptive MRAC controller design to
reactive and active power regulation in wind energy system.
regulate the active and reactive power provided by DFIG. Then,
a fractional order adaptive FO-MRAC controller is designed in The classical controllers like PI are able to deliver good
order to improve the energy system performance. Through a dynamic responses in nominal operating conditions, but their
comparative study between the classical MRAC and the proposed performance level may be lost during grid disturbances [16].
FOMRAC regulators, we aim to demonstrate the efficacy of the To overcome these drawbacks, a design of MRAC for the
proposed fractional adaptive control comparatively to classical DFIG and its implementation are presented. The reactive and
integer order control. Numerical simulations are given to active power value is controlled, the objective is to track
validate this performance enhancement. permanently the maximum aerodynamic power with a
classical controller compared with FOMRAC control strategy.
Keywords—Doubly fed induction generator (DFIG), Renewable
Finally, we finish the study with a comparison between
energy, Fractional adaptive control, FOMRAC, Robustness. MRAC and FOMRAC using the quadratic error criterion, in
each case for two powers mentioned above.
The objective is to prove the superiority of the fractional order
I. INTRODUCTION
adaptive control in terms of robustness and performance.
Recently, renewable electrical resources gather a great interest
and research activity in order to replace the conventional fuel II. DFIG SYSTEM MODELLING
fired based [1]. The installation of wind turbines has When modeling the DFIG machine, it is important to give the
developed rapidly in the last years [2]. In the last years, wind general structure of the system represented in the figure below
energy became one of the most important sources for proper [17]:
and durable energy. Generally, it is coupled with doubly fed
induction generators [3-8].
Meany controls are based on traditional regulators containing
PI and PID of a whole nature or at base of the regulators of a
fractional nature by using the adaptive control.
Since the first works of Ladaci and al. [9-10], FOMRAC
present the approach of control engineering systems.
FOMRAC are the best performances and robustness indices in
regard to ancient adaptive control structures [11].
Different applications of FOMRAC control are also cited in
literature in various science and engineering fields, from
which one can relate: a study of voltage control of DC/DC Fig.1 DFIG for wind turbine.
converter in multi-source renewable energy systems [12],

978-1-7281-5356-8/19/$31.00 ©2019 IEEE

Authorized licensed use limited to: Consortium - Algeria (CERIST). Downloaded on August 10,2022 at 16:10:26 UTC from IEEE Xplore. Restrictions apply.
 The mathematical model can be expressed as follows [18]:
The voltage of stator is given by:
d
V ssd = R ss I ssd + ( phi ssd ) − ω ss phi ssq
dt (1)
V ssq = Rss I ssq +
d
( phi ssq ) + ω ss phi ssd
dt

The rotor voltage equation:

d
V rrd = R rr I rrd + ( phi rrd ) − (ω ss − ω rr ) phi rrq
dt
Fig 2. PARK model of DFIG.
(2)
d
V rrq = R rr I rrq + ( phi rrq ) + (ω ss − ω rr ) phi rrd B. Strategy of Control
dt
After the calculations we find the stator voltages as [20]:
The stator flux linkage equation:
phi ssd = L ss I ssd + L M I rrd Vdss = 0 , Vqss = Vss = ω ss phi ss
(3) The relations of the two powers become:
phi ssq = L ss I ssq + L M I rrq
L
P = −V ss M i qrr
The rotor flux linkage equation: L ss
phi rrd = L rr I rrd + L M I ssd (9)
phi rrq = L rr I rrq + L M I ssq (4) V phi ss V ss L M
q = ss − idrr
L ss L ss
The equation of the mechanical part:  L  di
2
 L 
2
V rrd = R rr I rrd +  L rr − M  drr − g ω ss  L rr − M iqrr (10)
 L ss  dt L ss 
 
dΩ  J dωrr  3 L
J =  = − p M ( phirrd I ssq − phirrd I ssd ) − Cr − f r Ω (5)  L  diqrr
2
 L 
2
L V
dt  p dt  2 Lss Vrrq = Rrr I rrq +  Lrr − M  + gωss  Lrr − M iqrr + gωss M ss
 Lss  dt  Lss  ω ss Lss

The current derivative is defined as:

d [i ] The latter terms may be overlooked because of their weak
= A [i ] + B [u ]
dt (6) influence.
where:
[i ] = [idss iqss idrr ]
iqrr is the current vector,
III. FRACTIONAL ORDER MRAC CONTROL

[u ] = [u dss
u qss u drr u qrr the vector of applied voltage. ] Using adaptive control strategies allows more flexibility for
the controller as it may adapt itself online and in real time to
Matrices A et B are defined as flows: modifications in the process in order to maintain a certain level
 k  (7) of performance.
 − γ ω ω k
1
ss
T
rr 1 
 r
k 1
 In principle, an adaptive control system measures a certain
 − ω − γ − ω k 
A =
 ss rr 1
T rr  index of performance (IP) of the system to be ordered from the
 L − 1 
 M
0 ω sl  difference between the desired performance index and the
 T rr T rr 

0
L M
− ω
− 1  measured one.
 T
sl
T rr 
 rr 
L ss L 1 M
T rr = ; T ss = ss ; γ = ;k = A. Theory of fractional order systems
R rr R ss δ Tr 1 δ .L s .L r
2
LM
ω sl = ω ss − ω rr ; δ = 1 − Non-integer order systems may be presented in the Laplace
L ss . L rr domain as:
 1 
0 k k  (11)
 δ L (8) κκ
 ss
 F(p) =
 1 p β
B = 0 0 k  (1 + )
 δ L ss
 PP
 0 0 1 0 
 
 0 0 0 1 
where : β : is the real exponent,
A. Vectoriel control
PP : Cut frequency, fractional order pole,
With reference to the stator flux, we obtain [19]:
p : Laplace operator.
phidss = phiqss , phiqss = 0

Authorized licensed use limited to: Consortium - Algeria (CERIST). Downloaded on August 10,2022 at 16:10:26 UTC from IEEE Xplore. Restrictions apply.
 B. Fractional order integration
Reference
Let σ ∈ C , R (σ ) > 0 , k ∈ R and f a locally integrable Model ym
uc
function defined on [k,+∞[. The σ order integral of f , of Adjustment
mechanism
lower bound k is defined as [10]:
Δ t (t − τ ) σ −1 u
I kσ f (t ) =  f (τ ) d τ (12) Controller Process y
k
ℑ(σ )
With t ≥ k , and ℑ the Euler’s gamma function.
Fig. 3. Direct MRAC control
C. Linear approximation of fractional order TF

It is necessary to approximate the model chosen as E. M.I.T Law

reference in FOMRAC control. For this aim we can use the
singularity function frequency method. The second order-like Consider the quadratic error criterion,
fractional order model of the form (13) with β a positive real
1 2 (17)
number ( 0 < β < 1 ), J (θ ) = e
2
In order to reduce J we change parameters in the sense of
1 (13) negative gradient of J, so:
H ( p) = β
 p² p  dθ ∂J ∂e (18)
 + 2ε + 1 = −γ = − γe
dt ∂θ ∂θ
 ωn ² ωn 
following bloc scheme of fig-4, with m = 1,
is approximated as:
δ
 p  p 
 + 1  Model
 ωn  ω n + 1  (14)
ym
H ( p) =
 p² p  GM
 + 2ϑ + 1
 ωn ² ωn  -
θ
With ϑ = ε m
and δ = 1 − 2 β . This can lead to the γ
− Π Σ
following function [9] : sm
uc u y +
 p  n −1
p

ω
 n
+ 1

∏i =1
(1 + )
zi (15)
Π
G
H ( p) =
 p² p  n
p Plant
 + 2ϑ + 1 ∏ (1 + )
pi
 ωn ² ωn  i =1
Fig. 4. Adaptation algorithm

D. MRAC algorithm
F. Application of MRAC to DFIG active and reactive
The adaptive regulator parameters’ tuning is performed by power control
taking into account the error between the plant output and the
model reference one. This could be depicted as in fig. 3.
The control is given by, The block diagram for MRAC control of DFIG is shown in
υ = φ .θ i
T
(16) fig. 5, with Femd and Femq representing the compensation terms:
where φ is the regression vector .

The adaptation scheme is represented by the block-scheme of

fig. 3.

Authorized licensed use limited to: Consortium - Algeria (CERIST). Downloaded on August 10,2022 at 16:10:26 UTC from IEEE Xplore. Restrictions apply.
 The RST control structure is characterized by:
A .R + B .S = A r = A0 . A m (22)
where R, S and T are polynomials.

Ao is a stable polynomial called the "observer" polynomial.

We have deg(Am) = 2 (reference model order), deg(A) = 1 and

deg(B) = 0 (Plant model order), thus we have deg(Ao) = 0 and
deg(R) = deg(S) = deg(T) = 1.

Fig. 5 MRAC of active and reactive power of DFIG in wind system

Therefore, the regulation vector is given by:
 
IV. RESULTS AND INTERPRETATION θ =  r0 s0 s1 t0 t1 
  (23)
The uncoupled control results from the two powers of DFIG of the regression vector is obtain,
the wind energy, by integer order MRAC, and fractional order
  (24)
FOMRAC are presented in fig.6-7. ϕ T
=
b0  U Py y − Puc − uc 
A robustness test is finally proposed by introducing A0 Am  
 
disturbances with output noises.
The recurrence equation of ϕT obtained after the
The parameters of the used DFIG system are presented in table
1. discretization, with a samplin time of T=10-5 s, is given by:
ϕ ( t + 1, :) = 0 . 0008007 ϕ (t , :) − 2 .505 e − 7ϕ ( t − 1, :) + α ( t , :) (25)
Table 1. Characteristic parameters of the DFIG
where α(t,: ) is the measures vector.
Parameter Value
Vss 230 Volt B. FOMRAC controllers
p 2 pairs of pôles.
Rss 0.455 H The reference model has a second order-like form (3) with
fractional order β = 0.4 . Using Charef's approximation
Rrr 0.19 Ω
Lss 0.07 H method we obtain the approximation below:
Lrr 0.0213 H
LM 0.034 H 6 . 4 . 10 07
Gm ( p) =
f 50 Hz ( p 2 + 15200 p + 6 . 4 . 10 07 ) 0 . 4 (26)

The approximation function is of order 5:

A. MRAC controller
0.4 7.308.10-09.p2 + 0.0001806p +1 (27)
Gap ( p) =
6.376.10-22 p5 + 7.412.10-17p4 + 2.909.10-12p3 + 4.893.10-8 p2 + 0.0003641p+1
The system is presentedby the relation below:

(19) Then deg(Am) = 5 and from the Diophantine equation (22) we

H(p) =
L M .V ss obtain: deg(S) = deg(R) = deg(S) = 4.
 LM
2
 Therefore, the regulation parameters vector is given as:
L ss . R rr + P . L ss  L rr − 

 L   
ss
θ = r1 r2 r3 r0 s0 s1 s2 s3 s4 t 0 t1 t 2 t 3 t 4 
  (28)
The numerical replacement gives:
y( p) 2 . 334 e 04 B( p) The regression vector,
H(p) = = = (20)
u( p) p + 39 . 7 A( p) b0  k −1 
ϕT =  P u .... u P l y .... y − P m u c .... − u c 
A0 Am   (29)
According to the characteristics of the system studies, the
reference model is chosen as: Figures Fig. 6 and 7 show comparative results for active and
6 . 4 . 10 07 reactive power control respectively using integer and
Gm ( p) = (21)
(p2 + 15200 . p + 6 . 4 . 10 07
)β fractional order MRAC.

Authorized licensed use limited to: Consortium - Algeria (CERIST). Downloaded on August 10,2022 at 16:10:26 UTC from IEEE Xplore. Restrictions apply.
 0
1000
-2000
0
-4000
-1000
-6000
P (w)

Q(VAR)
-8000
-2000

-10000 -3000

Pinteger
-12000 -4000
Pfractional
Pref
-14000 -5000
0 0.1 0.2 0.3 0.4 0.5 0 0.1 0.2 0.3 0.4 0.5
Time (s) Time(sec)

Fig.6 Comparison between MRAC control and FOMRAC control with β=0.4 Fig.9 Reactive power with FOMRAC with random output noise of 30%
of active power.
Computing these criteria for different values of m and the
C. FOMRAC with random output noise integer case β=1, we obtain the results presented in tables 2
and 3:
Let us consider the proposed FOMRAC controller for
β=0,4 in presence of random output noises of 30% of the Table 2. Je and Ju criteria for active power with fractional order
reference signal amplitude. reference models FOMRAC (β = 0.4, 0.7, 0.9, 1).
For a comparative evaluation between the different output
performance of MRAC to FOMRAC, we define: β Je Ju
N max 0.4 14667.7 312.66
Je =  (u
n=0
c ( n ) − y ( n )) 2 (30)
0.7 16109.7 309.19
0.9 25115.1 349.02
N max
1 16088.0 337.80
Ju =  u (n)
n=0
2 (31)

Je: quadratic criteria for error Simulation results show that the control using FOMRAC is
Ju: input energy. able to optimize the transfers of energy.
4
x 10
2.6
1000
2.4
0
Je for active power

-1000 2.2

2
Q(VAR)

-2000
Je for integer
Je for Fractional reference m=1
-3000 1.8 reference
m=0.4
-4000
1.6
Q integer
-5000 Q fractional
1.4
Qref
-6000
0 0.1 0.2 0.3 0.4 0.5 1.2
Time(sec) 0.2 0.4 0.6 0.8 1 1.2
Fig.7 Comparison between MRAC control (blue) and FOMRAC control m
(green) with β=0.4 of reactive power. F Fig.10 Variation of error Je for different values of m for active power.

0
P
Pref
Table 3. Je and Ju criteria for reactive power with fractional order
-2000
reference models FOMRAC (β= 0.4, 0.7, 0.9, 1).
-4000

β Je Ju
P(w)

-6000
0.4 8995.1 122.58
-8000
0.7 10054.3 120.14
-10000 0.9 11895.9 135.48
1 10044.8 135.67
-12000
0 0.1 0.2 0.3 0.4 0.5
Time (sec)

Fig.8 Active power with FOMRAC with random output noise of 30% .

Authorized licensed use limited to: Consortium - Algeria (CERIST). Downloaded on August 10,2022 at 16:10:26 UTC from IEEE Xplore. Restrictions apply.
 x 10
4
[7] T. Surinkaew, and I. Ngamroo, “Coordinated Robust Control of DFIG
1.3
Wind Turbine and PSS for Stabilization of Power Oscillations
Considering System Uncertainties”, IEEE Transactions on Sustainable
1.2 Energy, vol. 5 , no. 3, pp. 823-833 , July 2014.
je for reactive power

Je for integer [8] M. Doumi, A. Aissaoui, A. Tahour, M.Abid, « Commande adaptative

reference m=1
1.1 d’un système éolien », Rev. Roum. Sci. Techn. – Électrotechn. et Énerg.,
vol. 60, no. 1, p. 99–110, Bucarest, 2015.
1 Je for fractional [9] S. Ladaci, A. Charef, “Fractional order model reference adaptive control
reference m=0.4 of a robot arm (in French)”, Revue Communication Sciences &
Technologie. ENP Oran, Algeria. vol.1, pp. 50-52, 2002.
0.9
[10] S. Ladaci, A. Charef, “On Fractional Adaptive Control”. Nonlinear
0.2 0.4 0.6 0.8 1 1.2 Dynamics, vol.43, no. 4, pp. 365-378, 2006.
m
[11] S. Ladaci, A. Charef, “Fractional order adaptive control systems: A
Fig.11Variation of error Je for different values of β for active power. survey”. In: Mitchell, E.W., Murray, S.R. (eds.) Classification and
Application of Fractals, pp. 261-275. Nova Science Publishers, USA
2012.
Whereas fig. 9 and 10 prove that the FOMRAC control gives [12] S. Djebbri , S. Ladaci, A. Metatla, H. Balaska, “Fractional-order model
better performance than the MRAC because the responses of reference adaptive control of a multi-source renewable energy system
with coupled DC/DC converters power compensation”, Energy Systems,
the two powers follow perfectly the suggested references, pp. 1-41, 2018.
besides, Tables 2 and 3 confirm that the tracking error [13] S. Ladaci, “Postoperative Blood Pressure Control Using a Fractional
obtained for FOMRAC (β = 0.4) is smaller than that obtained order Adaptive Regulator”, Proc. the 13th International conference on
for the MRAC in two powers control confirming the Sciences and Techniques of Automatic control & computer engineering
– STA’2012, pp. 254-265, Monastir, Tunisia, December 17−19, 2012.
robustness of the proposed control scheme [21].
[14] H. Balaska, S. Ladaci, H. Schulte, A. Djouambi, “Adaptive Cruise
Control System for anElectric Vehicle Using a Fractional Order Model
V. CONCLUSION Reference Adaptive Strategy”,9th IFAC Conference on Manufacturing
Modelling, Management and Control, MIM’2019. Berlin, Germany;
In the present study, adaptive control (FOMRAC) with August 28-30, 2019.
fractional order model reference is designed to supervise the [15] S. Djebbri, S. Ladaci, A. Metatla and H. Balaska, “Robust MRAC
active and reactive powers of DFIG in wind energy Supervision of a Multi-source Renewable Energy System Using
conversion. Numerical simulation results based on Fractional-Order Integrals”, International Conference on Electrical
Sciences and Technologies in Maghreb (CISTEM), IEEE, Algeria, 2018.
experimental data; show that the proposed regulation using
[16] S. Zhanfeng, S. Tingna, X. Changliang, C. Wei,“A novel adaptive
FOMRAC is more powerful than classical MRAC controls in control scheme for dynamic performance improvement of DFIG-Based
regards of disturbance rejection, response time and overshoot. wind turbines”. Energy, vol. 38, pp. 104–17, 2012.
It is demonstrated that FOMRAC is able to improve the [17] K. Ghedamsi, D. Aouzellag, E.M. Berkouk, "Power Control of a
control of energy transfers. Further research will focus on Variable Speed Wind Turbine Driving an DFIG". Renevelable energy,
robustness methods for this control technique against plant’s vol. 33, pp. 2145-2156, 2008.
model parametric variations. [18] F. Poitiers, T. Bouaouiche, M. Machmoum,"Advanced control of a
doubly-fed induction generator for wind energy conversion". Electric
Power Systems Research, vol. 79, pp. 1085–1096, 2009.
[19] Y. Zhao, X.D. Zou, Y.N. Xu, Y. Kang and J.Chen, "Maximal Power
REFERENCES Point Tracking under Speed-Mode Control for Wind Energy Generation
System with Doubly Fed Introduction Generator". 2006 CES/IEEE 5th
[1] G. Tapia, A. Tapia and J. R. Séenz, "A new simple and robust control International Power Electronics and Motion Control Conference,
strategy for wind power farm reactive power regulation". Proc. of the Shanghai, China, 14-16 Aug. 2006.
International Conference on Control Applications, IEEE, Glasgow, UK, [20] S. Abdelmalek, A. Gencer, and S. Rezazi, “Multiple Models
10 December 2002. Representation for a Doubly-Fed Induction Generator based Wind
[2] L. Mihet-Popa, F. Blaabjerg, and I. Boldea,"Wind Turbine Generator Energy Conversion Systems”, 2019 1st Global Power, Energy and
Modeling and Simulation Where Rotational Speed is the Controlled Communication Conference (GPECOM), Nevsehir, Turkey, 12-15 June
Variable".IEEE transactions on industry applications, vol. 40, no. 1, pp. 2019.
3 – 10, january/february 2004. [21] S. Ladaci, J.J. Loiseau, and A. Charef, “Using Fractional order Filter in
[3] R. Melício, V.M.F. Mendes, J.P.S. Catalão,"Power converter topologies Adaptive Control of Noisy plants”, in Proc. of the third International
for wind energy conversion systems: Integrated modeling, control Conference on Advances in Mechanical Engineering And Mechanics,
strategy and performance simulation," Renewable Energy, vol. 35, no. ICAMEM 2006 Hammamet, Tunisia, 17-19 December 2006.
10, pp. 2165-2174, October 2010.
[4] M. Chakib, A. Essadki, T. Nasser, “Robust ADRC Control of a Doubly
Fed Induction Generator Based Wind Energy Conversion System”,
International Conference on Electronic Engineering and Renewable
Energy, ICEERE, 2018.
[5] F.E.V. Taveiros, L.S. Barros, F.B. Costa , “Back-to-back converter state-
feedback control of DFIG (doubly-fed induction generator)-based wind
turbines”, Energy, vol. 89, pp. 896-906, September 2015.
[6] M. Kalantar, S.M. Mousavi G. “Dynamic behavior of a stand-alone
hybrid power generation system of wind turbine, microturbine, solar
array and battery storage” , Applied Energy, 87 3051–3064, 2010.

Authorized licensed use limited to: Consortium - Algeria (CERIST). Downloaded on August 10,2022 at 16:10:26 UTC from IEEE Xplore. Restrictions apply.