4th International Conference on Electrical Engineering (ICEE 2015)

IGEE, Boumerdes, December 13th -15th, 2015

Modeling and Hybrid Intelligent control double-fed

asynchronous machine for supply of power
to the electrical network
Abdelkarim GUEDIRI

Djilani BENATTOUS

Department of Electrical Engineering, University Med

Khaider
Biskra, Algeria
Karim_elect@yahoo.fr

Department of Electrical Engineering, University Echahid

Hamma Lakhder
El-Oued Algeria
dbenattous@yahoo.com

Abstract this paper presents an A Hybrid Intelligent control

of a doubly fed asynchronous machine (DFAM) is proposed.
First, a mathematical model of DFAM written in an appropriate
d-q reference frame is established to investigate simulations
results. In order to control the rotor currents of DFAM, a torque
tracking control law is synthesized using PI controllers; the
stator side power factor is controlled at unity level. Specifically
neuro-fuzzy systems are created to overcome the disadvantages
of neural networks and fuzzy systems. Results obtained in
Matlab/Simulink environment show that the neuro-fuzzy control
is more robust, have superior dynamic performance and hence
found to be a suitable replacement of the conventional PI
controller for the high performance drive applications.

Keywords Turbine;Doubly fed asynchronous machine (DFAM)

transformation of the currentsinto a synchronously rotating

dq reference frame that is aligned with one of the fluxes.

II.

WIND TURBINE MODEL

Several models for power production capability of wind

turbines have been developed. The mechanical power,
captured pmech by a wind turbine [4], depends on its power
coefficient Cp given for a wind velocity and can be
represented by [4]

(1)

pmech = C p R 2 V 3

; neuro-fuzzy control.

0.5
BITA=0.9
0.45

I.

INTRODUCTION

Doubly-Fed Asynchronous is an electrical asynchronous

three-phase machine with open rotor windings which can be
fed by external voltages. The stator windings are directly
connected to the line grid, while the rotor windings are
controlled by means of an inverter. This solution is very
attractive for all the applications where limited speed
variations around the synchronous velocity are present, since
the power handled by the converter at rotor side will be a
small fraction (depending on the slip) of the overall system
power,[1][2].
wound rotor induction machine, used as a Doubly Fed
Induction Generator (DFIG) wind turbines are nowadays
becoming more widely used in wind power generation. The
DFIG connected with back to back converter at the rotor
terminals provide a very economic solution for variable speed
application. Three-phase alternative supply is fed directly to
the stator in order to reduce the cost instead of feeding through
converter and inverter. For the control of these converters
different techniques will be adopted [3].
The network side converter control has been achieved
using Field Oriented Control (FOC). This method involves the

BITA=05

0.4

BITA=10

0.35

BITA=15
BITA=20

0.3

BITA=25
0.25
0.2
0.15
0.1
0.05
0

10

12

14

Fig. 1. Typical Power Coefficient versus Tip-Speed-Ratio Curve

III. DESIGN OF THE NEURO-FUZZY CONTROLLER

Fig. 2 shows the block diagram of the neuro-fuzzy

controller (NFC) system. The NFC controller is composed of
an on-line learning algorithm with a neuro-fuzzy network [5].
The neuro-fuzzy network is trained using an on-line learning
algorithm. The NFC has two inputs, the rotor current error eidr
and the derivative of rotor current error eidr & eiqr . The
output is rotor voltage Vdr. For the NFC of rotor current Irq is
similar with Ird controller [6].

2015 IEEE

16

A. Description of NFC:
For the NFC, a four layer NN as shown in fig. 5 is used.
Layers IIV represents the inputs of the network, the
membership functions, the fuzzy rule base and the outputs of
the network, respectively.[7].
1. Ayer I: input layer
Inputs and outputs of nodes in this layer are represented as:

net

I
1

= e idr (t ), y

net

I
2

= e idr (t ), y 2I = f 2I net

I
1

= f 1 net

) = net
) = net

I
1

I
1

I
2

= e i dr (t )

I
2

= e i dr (t )

(3 )
(4 )

Layer II: Membership layer

In this layer, each node performs a fuzzy set and the
Gaussian function is adopted as a membership function
net 1II. J =
net 2II. k =

(x

(x

II
1. j

m 1II. j

( )

m 2II. j

( )

Fig. 2.

II
2.k

, y 1II. j = f 1 II. j net 1II. J = exp net 1II. j

II
1. J

II
2. j

w IV
jk =

eidr (t ) eidr (t ) net 0IV a

1
= 0IV y III
jk
net 0IV
a w IVj k b

Since the weights in the rule layer are unified, only the
approximated error term needs to be calculated and
propagated by the following equation [9]:
e (t )eidr (t ) net 0IV y1III. J
1
IV
0IV = idr
= 0IV w IV
jk y 0
IV
III
III
net 0
y1. J net JK b

(11)

) (12 )

The error received from Layer III is computed as:

e (t )e (t ) net III
y 1II. j
jk
idr
idr

(13 )
1II. K =

III
II
y
net jk
net 1II. j
k

1. j

The updated laws of also can be obtained by the gradient

decent search algorithm [10].

) (5 )

, y 2II. k = f 2II.5 net 2II. k = exp net 21II . k

) (6 )

Block diagram of the neuro-fuzzy controller

1. Layer III: Ruel layer

This layer includes the rule base used in the fuzzy logic
control (FLC). Each node in this layer multiplies the input
signals and outputs the result of product [8]

III
III
III
net1III.k = x1III. j x1III. K , y III
j k = f jk net J .k = net j .k (7 )

2. Layer IV: Output layer

This layer represents the inference and defuzzification
used in the FLC. For defuzzification, the centre of area
method is used; therefore the following form can be obtained:
a =

net 0IV =
The
0IV

IV
jk

III
jk

,b =

III
jk

a IV
a
, y 0 = f 0IV net 0IV =
b
b
on-line

Learning

e idr (t )e idr (t ) y 0IV

=
= 5 e idr (t )
y 0IV
net 0IV

(8 )

(9 )
algorithme

(10 )

Is a gradient descente search algorithm in the space of

network parameters.

Fig. 3.

: Schematic diagram of the neuro-fuzzy network

e idr (t ) e idr (t ) net 1II. j

2 x1II. j m 1II. j

m 1II. j =
= 4 1II. j
II
III
2

net jk
1II. J

m 1. j

e idr (t )e idr (t ) net 2II. K

m 2II. K =
m II
net 2III. k
2. K

II

e
t
e
t

net
(
)
(
)
idr
1. J

1II. J = idr
m II
net 1II. J
1. J

(14 )

(15 )

(16 )

IV. SIMULATION RESULTS

Fig. 4 shows the block diagram of the DFIG neuro-fuzzy.

The results. Fig.5 shows the results of rotor currents responses
using neuro-fuzzy, respectively. The stator power Ps follows
the current Irq. These results in unity power factor on the grid
as the stator reactive power Qs is zero are shown in Fig. 6.
Rotor current errors are controlled at zero level. The reactive
component of the stator current Isd is almost equal to zero

during all the timeas shown in

fig. 7. As result, the stator
phase current, reported in fig. 8, has an opposite phase angle
to the line voltage indicating that the stator power is injected
to the line grid. Reported in fig. 9 and fig.12, shows the
results of stator and rotor three Phase currents responses using
neuro-fuzzy.

100

Is q

Is d

0
-100
-200

0.1

Fig. 7.

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

Stator currents response With neuro-fuzzy

400

Is a

Vs a

300
200
100
0
-100
-200
-300
-400
0.4

0.41

Fig. 8.

0.42

0.43

0.44

0.45

0.46

0.47

0.48

0.49

0.5

Stator voltage and current

Is a b c

200

-200

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

Fig. 9. Stator Three Phase Currents (Is_ abc).

150
Is a
100

Is a

50

-50

Fig. 4.

Block diagram of the DFIG neuro-fuzzy

-100

-150

200

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

i rq

I rd

Fig. 10. Stator per phase Currents (Is_ a)

150

100
150
I A

50

100

-100

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

IA

50

-50

-50

-100

Fig. 5.

Rotor currents responses

-150

0. 1

0.2

0.3

0. 4

0.5

0.6

0.7

0.8

0. 9

Fig. 11. : phase Rotor Currents (Ir_ a).

x 10

Qs

Ps
150
Ir A

Ir B

100

Ir C

-1

IA
B
C

-2

50

-3

-50
-4

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

-100

Fig. 6.

Active and reactive powers

-150

0.1

0.2

0.3

0.4

Fig. 12. Rotor Three Phase Currents (Ir_ abc).

0.5

0.6

0.7

0.8

0.9

VII. CONCLUSION
In this paper a hybrid intelligent control based torque
tacking approach for doubly fed asynchronous drive have
been proposed. Torque tracking control strategy has been
achieved by adjusting rotor currents and using stator voltage
vector oriented reference frame. The performances of neurofuzzy controller which is based on the torque tracking control
algorithm has been investigated and compared to those
obtained from the PI controller. Simulations results have
shown that the NFC is more robust, efficient and more robust
under parameters variations of the DFIG.
ANNEXURE
Nominal Power

Pn = 1.5 MW

Stator Voltage

vs = 200V

Stator Frequency

fs = 50 Hz

Stator Resistance

Rs = 0.12

Stator Inductance

Ls = 0.0205 H

Rotor Resistance

Rr = 0.021

Rotor Inductance

Lr = 0.0204H

Mutual Inductance

Lm = 0.0169 H

Inertia Constant

J = 1000 Kg-m2

References
[1]

Brahim Nait-kaci, Mamadou L. Doumbia, Active and Reactive power

control of a doubly fed induction generator for wind applications,
IEEE 2009.
[2] Arantxa Tapia, Gerardo Tapia, J. Xabier Ostolaza, Modeling and
Control of a Wind Turbine Driven doubly fed Induction Generator,
IEEE 2003.
[3] C. Eisenhut, F. Krug, C. Schram and B. Klockl, Wind Turbine Model
for System Simulation Near Cut- in Wind Speed IEEE Trans, on
Energy Conversion, June 2007,
vol. 22, 414-420.
[4] Lee. G et J. S. Wang, Efficient Neuro-Fuzzy Control Systems for
Autonomous Underwater Vehicle
Control, IEEE International
Conference on Robotics and Automation, Seoul, Core, 2001.
[5] Chaiba. A, R. Abdessemed, M. L. Bendaas and A. Dendouga,
Performances of Torque Tracking
Control for Doubly Fed
Asynchronous Motor using PI and Fuzzy Logic Controllers, Journal of
Electrical Engineering, JEE, Vol.5, N2, pp. 25-30, Romania, 2005..
[6] B. Shen, B. Mwinyiwiwa, Y. Zhang, and B. T. Ooi, Sensorless
Maximum Power Point Tracking of Wind by DFIG Using Rotor
Position Phase Lock Loop (PLL),IEEE Transactions on Power
Electronics, Vol. 24, no. 4, pp. 942 951, Apr. 2009.
[7] A.D. Hanseen, Generator and power electronics for wind turbine
Chapter in Wind Power in Power Systems, John Wiley and sons Ltd.,
2004
[8] C. Elmas, , O. Ustun, H. H. Sayan, A neuro-fuzzy controller for speed
control of a permanent magnet synchronous motor drive.. Expert
Systems with Applications, Vol. 34. n1, 2008, pp. 657- 66
[9] H, Closure properties of the compositional rule of inference, Fuzzy
Sets and Systems,35, 163183, 1990.
[10] Abdel Karim Guediri1, D. Ben Attous2, "Modeling and fuzzy control of
a wind energy system based on double-fed asynchronous machine for
supply of power to the electrical network, The Society for Reliability

Engineering, Quality and Operations Management (SREQOM), India

and The Division of Operation and Maintenance, Lulea University of
Technology, Sweden 2015