Professional Documents
Culture Documents
7, 2007 1
*
Mohammed CHENAFA, **Abdelkader MEROUFEL, *Abdellah MANSOURI, **Ahmed MASSOUM
*
Department of electrical engineering, E.NS.E.T. Oran, Laboratory of Automatics and Systems Analysis (L.A.A.S), B.P.
1523 El.M’naouer, Oran ,Algeria, Email : chenafa@enset-oran.dz
**
Department Intelligent Control & Electrical Power Systems Laboratory (ICEPS). Faculty of engineer sciences. Department
of electrical engineering, Djillali Liabes University, Sidi Bel Abbes, Algeria, Email : ameroufel@yahoo.fr
SUMMARY
The proposed system combines the input-output linearization technique and the advantages of fuzzy logic control
(intuitiveness, simplicity, easy implementation and minimal knowledge of system dynamics) to obtain a robust, fast and
precise control of induction motor speed. The input-output linearizing controller decouples the flux control from the speed
control and makes the synthesis of linear controllers possible. However, in practice, parameter variations limit the evolution
of the machine flux and the rotor speed. So, a speed fuzzy logic controller for induction motor via feedback linearization with
input-output decoupling is presented here in order to achieve better tracking performances. The signal rotor flux is estimated
with a nonlinear observer and controlled by classical regulator. Simulation results show the importance of the inclusion of
fuzzy logic in designing input-output linearizing control laws. Similarly, it validates the robustness and effectiveness of the
proposed fuzzy logic controller for high performance of input-output linearized induction motor drive.
Keywords: Induction Motor (IM), Input-Output Linearization (IOL), Fuzzy Logic Controller (FLC), and Nonlinear
Observer (NLO).
f ( x) gu f ( x) g1uDs g 2 uEs
expertise knowledge of the system for FLC
parameter setting, and the controller can be only as x (1)
ISSN 1335-8243 © 2007 Faculty of Electrical Engineering and Informatics, Technical University of Košice, Slovak Republic
2 Fuzzy Logic Control of Induction Motor with Input Output Feedback Linearization
Where: us (uDs uEs )T stator voltage, Thus the derivatives of the outputs are given by
Ir (IDr IEr ) : rotor flux,
L f h1 ( x) Lg1 h1 ( x)uDs Lg2 h1 ( x)uEs
T
h1 ( x)
is (iDs iEs )T : stator current, : : mechanical speed,
h2 ( x) L f h2 ( x) (4)
« »
« »
r
« r »
performed. The nominal part of the IM model is
«M »
T T r written in the form of higher derivatives of inputs.
« iEs p:IDr IEr »
1
¬« Tr Tr ¼» ª h1 ( x) º ª L f h1 ( x) º ª uD s º
« » « 2 » D( x) « »
¬« h2 ( x) ¼» ¬« L f h2 ( x) ¼» ¬uEs ¼
(5)
ª 1 º ª º
« 0 0 0» , g 2 «0 0 0»
T T
1
¬ VLs ¼ ¬ VLs ¼
g1
ª Lg1 h1 ( x ) Lg2 h1 ( x) º
« »
¬« Lg1 L f h2 ( x ) Lg1 L f h2 ( x) ¼»
Where D ( x)
V 1 ,Q
M2 Rs Rr M 2
VLs VLs L2r
,
Lr Ls
ª u Ds º ª L f h1 ( x) v1 º
equation is:
« » D 1 ( x ) « 2 »
ª M º ¬uEs ¼ «¬ L f h2 ( x) v2 »¼
« p L (IDr iEs IEr iDs ) »
(6)
ª h1 ( x) º
« h ( x) » « »
¬ 2 ¼ « »
>v1 v2 @ is the new system inputs. The
y ( x) (2)
I I
r
¬ Dr Er ¼
2 2 T
Note that v
resulting closed loop system is governed by the
Where Te : Electromagnetic torque, equation (7)
TL : Load torque
The parameters of the considered motor are given in h1 ( x) v1
(7)
appendix B. h2 ( x) v2
k p1 (T Tref ) Tref
of the system’s nonlinearities in order to obtain a
linear relationship between the control inputs and the v1
k p 2 (Ir2 Ir2 ref ) kd 2 (I r2 I r2 ref )
Ir2 ref
system outputs in the closed loop. The nonlinear (8)
v2
control law is reduced to a successive differentiation
of each output until at least one input appears in the
derivative. The following notation will be used for Where Tref and I2r ref are torque and flux references.
¦
wh( x)
n
their derivatives. Since we want to control speed we
wxi
L f h( x ) f i ( x) (3) implement a FLC which gives the reference torque
i 1 and compensate variations of the load torque.
ISSN 1335-8243 © 2007 Faculty of Electrical Engineering and Informatics, Technical University of Košice, Slovak Republic
Acta Electrotechnica et Informatica No. 2, Vol. 7, 2007 3
The fuzzy controller is used to perform nonlinear The inputs to the fuzzy controller are error ( e )
decoupling control in order to obtain better and variation error ( de ). The output of the fuzzy
responses under parameter variations [13].
controller is ( du ). The universe of ( e ), ( de ), and
( du ) are partitioned into three fuzzy sets. NB
4.1. Fundamentals of FLC (Negative Big), EZ (Equal Zero), and PB (Positive
Big). Each fuzzy set is represented by triangular
The schematic diagram of a FLC for IM drive is membership function or trapezoidal membership
shown in fig.1 function (see fig.2 (a)). The rule base of the FLC
contains nine rules based on the IF-THEN structure
which are tabulated in table1. The crisp output of the
FLC is obtained by using Max-Min inference
algorithm and the centre of gravity defuzzification
approach. It is known that the sensitivity and
performance of FLC system are significantly
affected by the weighting factors C1, C2, and C3.
However, we lack a general methodology for
choosing these factors. In this paper, a trial and error
approach is used to determine and adjust these
weighting factors [13]. The control surface fig.2 (b)
represented by a three dimensional graphic showing
the output variable corresponding to all
Fig. 1 Internal FLC structure combinations of values of the inputs can be used to
facilitate the FLC tuning.
databases. The functional blocks in FLC are 0.6 NB EZ PB 0.6 NB EZ PB 0.6 NB EZ PB (a)
0.4 0.4 0.4
0.2 0.2 0.2
0 0 0
1. Fuzzification -100 0 100 -0.5 0 0.5 -200 0 200
0
systems, non fuzzy input data is mapped to fuzzy (b)
-100
sets by treating them as triangular membership 0.5 100
0 50
functions, gaussian membership functions, ... de -0.5
-100
-50 0 e
3. Defuzzification
5. NONLINEAR OBSERVER
There are many methods which can be used for
converting the conclusions of the inference Variable speed AC drives based on IOL control
require an evaluation of the instantaneous magnetic
mechanism into the actual input for the plant. Centre
of gravity defuzzification method is often used [13]. flux of the rotor. Here, a NLO [12] is used to
Other defuzzification strategies can be found in construct the unknown states of the rotor flux. This
observer is given by
technique literatures.
ISSN 1335-8243 © 2007 Faculty of Electrical Engineering and Informatics, Technical University of Košice, Slovak Republic
4 Fuzzy Logic Control of Induction Motor with Input Output Feedback Linearization
ª º
«Q 0 T p:K »
Where
ªk1
K
0 º
« » « »
« K» k1 » ª fDi º
« 0 (iDs IDr iˆEs Iˆ Er )
t
«0 Q p:K » « » (9)
2M ˆ ˆ
« »
hˆ3
« p:k2 » uˆs « »
t »
xˆ gu « 2
T fEi
«M » « 0 »
xˆ
k Tt
0 p: » « Tr »
« « »
« » ' IDr p:Iˆ Er )iˆDs 2( 2 Iˆ Er p:Iˆ Dr )iˆEs
1
« r » ¬«0 ¼»
k2 ˆ k
« p:k2 2 »
2(
« 1 »
T T k
«
¬ Tr ¼»
t
p:
Tr Tr
«0 »
M
¬« Tr Tr ¼»
ĥ3 is an artificial auxiliary output [12].
K [1 ( p:Tr ) 2 ]
r
k2
x xˆ ª L hˆ L hˆ u L hˆ u º
ª hˆ º
This leads to the following error equation x
« 1» « fˆ 1 g1 1 Ds g 2 1 Es
»
« hˆ » « 2 »
ª º « 2» « hˆ2 hˆ3 ' »
«Q k1 p:K » « Tr »
K (15)
« ˆ »
« » « ˆ »
0
« M k2 » «0 »
(10)
« p:k2 p: » « »
1
« Tr Tt » ¬«0 ¼»
The Lie derivative estimates are given in appendix C
« 1 »
Tt
« p:k2 p: »
M k2
¬« Tr Tr Tr »¼ The error between the desired trajectory of the
outputs and the estimate outputs are
ª hˆ1 h1ref º
ªe1 º « »
The two controlled outputs of the system are
unknown, so, we define the estimated outputs « » « hˆ2 h2 ref »
«e2 » « »
(16)
«¬e3 »¼ « hˆ3 h3ref »
ª M º ¬ ¼
ª hˆ1 ( x) º «p L (Iˆ Dr iˆEs Iˆ Er iˆDr ) »
« » « r »
«¬ hˆ2 ( x) »¼
(11)
« Iˆ D2 r Iˆ E2r »
¬ ¼
Let us design the control inputs as:
« » (17)
« »
ˆ
« »
¬uEs ¼ «¬ Lg1 hˆ3 Lg2 hˆ3 »¼ «¬Lfˆ hˆ3 e2 k p3 e3 h3ref »¼
The derivatives of ĥ1 and ĥ2 are defined as: *
h2 h3 '
h2 h2 ref k p 2 e2
2 ˆ ˆ
L f hˆ2 (13) 2 ˆ
Tr h3ref (19)
Tr
ISSN 1335-8243 © 2007 Faculty of Electrical Engineering and Informatics, Technical University of Košice, Slovak Republic
Acta Electrotechnica et Informatica No. 2, Vol. 7, 2007 5
ª e1 º ª k p1 0 0 º ª e1 º
« »« »
The proposed FLC for controlling the speed of
« »
« e2 » « 0 k p2 1 » « e2 »
IM with IOL decoupling and flux NLO was
« » «e »
(20)
¬« e3 ¼» ¬« 0 1 k p3 ¼» ¬ 3 ¼
designed for 1.1kw drive is represented in fig.3.
The performance was verified by means of
simulations. Parameter values of the IM are given in
the appendix A. In order to valid the algorithm for a
A simple linear study would be sufficient to large operating domain, we use the reference
establish the stability of (20). profiles shown on figure.4
Fig.5 shows in increasing regime or decreasing linearization technique. So, the observed flux or
regime, the rotor speed tracks the reference speed rotor flux, after a short transient regime in start up
with smooth constant delay. In constant regime, the fellows only the reference flux and is independent of
rotor speed follows exactly the reference speed. At the speed and torque variations.
the abrupt variations of the speed or the load torque
a transient pick appears in the speed error which
200
shows the reaction of the corrector in the speed
ref erence speed (rad/s)
response.
0
Fig.6 shows that the electromagnetic torque
follows the load torque when the speed is constant. -200
0 1 2 3 4 5 6 7 8 [s] 9
During an increase or decrease of the speed the
electromagnetic torque tracks the load torque with
an augmented torque corresponding to accelerator or ref erence rotor f lux (Wb)
0.5
decelerator torque.
The performance of the drive is tested for 0
parameter variation, which is shown in fig.7 the 0 1 2 3 4 5 6 7 8
[s]
9
ISSN 1335-8243 © 2007 Faculty of Electrical Engineering and Informatics, Technical University of Košice, Slovak Republic
6 Fuzzy Logic Control of Induction Motor with Input Output Feedback Linearization
200 200
ref erence speed (rad/s) ref erence speed (rad/s)
rotor speed (rad/s) rotor speed (rad/s)
0 0
-200 -200
0 1 2 3 4 5 6 7 8 [s] 9 0 1 2 3 4 5 6 7 8 9
[s]
5 60
40 electromagnetic torque (Nm)
speed error tracking (rad/s)
20 load torque (Nm)
0
0
-20
-5 0 1 2 3 4 5 6 7 8
0 1 2 3 4 5 6 7 8 [s] 9
[s] 9
0.8
60
40 electromagnetic torque (Nm)
0.8 observ ed rotor f lux (Wb)
load torque (Nm) 0.6 0.6
20
0.4
0 0 0.05 0.1
0.4
-20 0 1 2 3 4 5 6 7 8 9
0 1 2 3 4 5 6 7 8 9 [s]
[s]
Fig. 5 Speed variations under nominal inertia Fig. 9 Speed variations with constant flux
and load variations
200 7. CONCLUSION
ref erence speed (rad/s)
0 rotor speed (rad/s)
-200
0 1 2 3 4 5 6 7 8
[s]
9
The different results obtained show that the
2
speed error tracking (rad/s)
developed FLC used with NLC method gives very
0 good performances for the induction motor control.
-2
The decoupling between the rotor flux and rotor
0 1 2 3 4 5 6 7 8 [s] 9
60
speed is maintained under load and speed variations
electromagnetic torque (Nm)
40
load torque (Nm)
( fig.9).The advantages of the proposed NLO may be
20
0 resumed as follows: high convergence rate of the
Fig. 6 Speed variations under double inertia and decoupling by nonlinear state feedback and the FLC
load variations theory based on linguistic expert control
interpretations permits to avoid the problem of flux
orientation and inexactitude of the representative
system model. This strategy of control gives a stable
50
electromagnetic torque (Nm) system with a satisfactory performance either with
40 or without load variations. The performance depends
30 on the operator expertise. The proposed speed FLC
20
load torque (Nm) has a good robustness and less sensitive to parameter
variations. The assumption made in the development
10
of this control law is the linearity of the magnetic
0
circuit of the machine. However, there is no
0 1 2 3 4 5 6 7 8
[s]
9 guarantee that the magnetic flux remains in the
linear magnetic region during machine transients.
Fig. 7 Electromagnetic torque and load torque Then, the problem of controlling the IM with
magnetic saturation will be studied in the
perspective.
1 rotor f lux (Wb)
ISSN 1335-8243 © 2007 Faculty of Electrical Engineering and Informatics, Technical University of Košice, Slovak Republic
Acta Electrotechnica et Informatica No. 2, Vol. 7, 2007 7
p IEr
M ˆ
VLsTr
Lg1 hˆ3
I Dr
M ˆ
VLsTr
Lg2 hˆ3 p
REFERENCES
«© Tr ¹ »
p:IDr iDs I Er i Es
input output decoupling. Int. J. Control,
M « »
Vol.51, no4, Mars1990, pp.863-883
Lr « »
L f h1 p
« »
adaptive speed controller for induction motor
¬ ¼
based on feedback linearization. IEEE, Trans.
Energy. Conversion., Vol.13, no1, 1998, pp.1-6.
Tr ¬
2 observer based composite adaptive speed
L f h2
controller for an induction motor. IEEE, Trans.
Energy. Conversion., Vol.12, no4, 1997,
6M 2Q
L2f h2 ( 2 2 M)(ID2r IE2r ) ( 2 M)(IDr iDs IEr iEs )
4 2K pp.323-329.
Tr Tr Tr Tr
2Mp:
[5] J.V. Raumer : Commande adaptative non
(IDr iEs IEr iDs ) 2 M2 (iD2s iE2s )
2 linéaire de la machine asynchrone. Thèse de
Tr doctorat, INP Grenoble, 1994
Tr
pK IEr 2 Rr K IDr
[6] A. Mansouri, M. Chenafa, A. Bouhena, and E.
Lg1 h1 Lg1 L f h2 Etien : Powerfull nonlinear observer associed
pK IDr 2 Rr K IEr
with the field oriented control induction motor.
Lg2 h1 Lg2 L f h2 Int. J. Applied Math. Comput.vol.38,no2, 2004,
pp.208-221.
IDr
M ˆ [9] Liu Mingji, Lai Zhongqin, Cheng Ximing, and
VLs Lr
Lg2 hˆ1 p Ouyang Minggao : Adaptive position servo
control of permanent magnet synchronous
QiˆDs IDr p:K Iˆ Er 2TiˆDs f Di
K ˆ motor. Proceeding of the 2004 American
fˆ1
Tr control conference Boston, Jun 30- July 2, 2004
ISSN 1335-8243 © 2007 Faculty of Electrical Engineering and Informatics, Technical University of Košice, Slovak Republic
8 Fuzzy Logic Control of Induction Motor with Input Output Feedback Linearization
[12] Chenafa Mohammed : Contribution à la in 1979, the MS degree from the same University
stabilité de la commande non linéaire avec (USTO) in 1990 and the PhD degree from The
observateur du moteur asynchrone. Thèse Electrical Engeneering Institute of University of Sidi
doctorat USTOran (Algérie), 2005. Bel-Abbes (Algeria) in 2004. He is currently
Professor of electrical engineering in this University.
[13] A. Meroufel , A. Massoum, B. Belabes : He is a member of ICEPS (Intelligent Control
Fuzzy logic control of permanent magnet Electrical Power System) Laboratory. His research
synchronous machine fed by voltage source interests are robust control of electrical machines.
inverter via input output linearization. Revue
roumaine des sciences techniques, serie Abdellah Mansouri : was born in Oran in Algeria,
Electrotechnique et energétique, tome 49, in 1953. He received his BS degree in electronic
juillet-septembre 2004, p.385-393 engineering from USTO (Algeria) in 1979, the MS
degree in engineering control from USTO (Algeria)
in 1991, and the PhD degree in engineering control
BIOGRAPHIES from USTO (Algeria) in 2004. He is currently
professor of automatic control at ENSET of Oran
Mohammed Chenafa : was born in Oran in Algeria, (Algeria). His research interests are non linear
in 1954. He received his BS degree in electronic control and observers applied in induction motor and
engineering from USTO (Algeria) in 1979, the MS manipulator robot.
degree in signal processing and robotic from USTO
(Algeria) in 1998, and the PhD degree in Ahmed Massoum: was born in 1959 in M’sirda
engineering control from USTO (Algeria) in 2005. Fouaga, Tlemcen, Algeria. He received his BS
He is currently professor of automatic control at degree in electrical engineering from the Electrical
ENSET of Oran (Algeria). His research interests are Engineering Institute (INELEC) of Boumerdes 1985
non linear control and observers applied in induction and the MS degree from the Electrical Engineering
motor and manipulator robot. Institute of Sidi Bel-Abbes University in 2004 where
he is currently Professor of electrical engineering.
Abdelkader Meroufel: was born in Sidi Bel-Abbes He is a member in Intelligent Control Electrical
(Algeria) in 1954. He received his BS degree in Power System Laboratory (ICEPS).
Electrical ingeneering from the USTOran (Algeria)
ISSN 1335-8243 © 2007 Faculty of Electrical Engineering and Informatics, Technical University of Košice, Slovak Republic