TRANSACTION ON CONTROL AND MECHANICAL SYSTEMS, VOL. 2, NO. 7, PP. 321-326, JUL., 2013.

RECEIVED: 3, FEB., 2013; REVISED: 16, MAY, 2013; ACCEPTED: 22, MAY, 2013; PUBLISHED: 4, JUN., 2013. ISSN: 2345-234X

Abstract: In this paper, a sensorless Direct Torque control
(DTC) for Permanent Magnet Synchronous Motor (PMSM)
drive based on Adaptive Model Reference (MRAS) algorithm
and PI with adapted gains fuzzy logic (FL) are presented. The
MRAS is utilized to estimate speed and stator resistance and
compensate the effects of parameter variation on (stator
resistance variation introduce errors flux and torque
estimation and affect the performance of DTC). In other hand,
speed regulation by PI with two adaptive FLC is investigated
and compared with classical PI controller. Simulation results
are presented and show the effectiveness of the proposed
method.
Keywords: MRAS, PMSM, Speed and Resistance Estimation,
Sensorless Drive, FLC, DTC.
1

1. INTRODUCTION
High dynamic performance of servo motor drives is
indispensable in many applications of todays automatically
controlled machines. AC motor control has attracted much
attention recently in the power electronics field
[1]
.
Progress in the field of power electronics and
micro-electronics, software engineering, and materials,
enables the application the PMSM, based on modern rare
earth variety, becomes serious competitor to the induction
motor and conventional wound rotor synchronous motor
and DC motor. PMSM drives are used in many applications.
They are used more in the variable speed applications due to
some advantages like: more simplicity, low maintenance,
low dependency on the motor parameters, good dynamic
torque response, high rate torque/inertia and simplicity of
design
[2]
.
Direct Torque control (DTC) for induction motor
synchronous machines was introduced about more than
twenty years by Takahashi and Depenbrock
[3-4]
. The main
advantages of DTC are the simple control scheme, a very
good torque dynamic response, as well as the fact that it
does not need the rotor speed or position to realize the
torque and flux control, moreover DTC is not sensitive to
parameters variations (except stator resistor)
[1, 5]
.
However, it still has some disadvantages: high torque
ripples and variable switching frequency, which is varying

1,3
A. Ameur and B. Mokhtari are in LeDMaSD Laboratory, Electrical
Engineering Department, Laghouat University, Email: amaissa1@yahoo.fr
2
K. Ameur are in La LACoSERE Laboratory, Electrical Engineering
Department, Laghouat University, Email: ameur088@gmail.com.
Please refer to the Nomenclature at the end of the paper!
with speed, load torque, selected hysteresis bands and
difficulty to control torque and flux at very low speed
[5-6]
.
The high performance speed or position control requires
an accurate knowledge of rotor shaft position and velocity
in order to synchronize the phase excitation pulses to the
rotor position. This implies the need for speed or a shaft
position sensor such as an optical encoder or a resolver
[6]
.
However, the speed controller design of such systems plays
a crucial role on these performances
[7]
. The speed controller
PI compensator based torque-load angle and speed control
scheme have a simple structure and it can offer a
satisfactory performance over a wide range of operation but
fixed gain PI controllers does not provide satisfactory
control performance in the presence of parameter variations
and disturbances under a wide range of driving conditions
[8]
.
The FLC is seemed an ultimate solution in terms of high
dynamic response and best disturbance rejection. The
special merit of this controller is that it does not require the
knowledge of a mathematical model of the plant for the
adaptation mechanism, and it can be easily implemented
[7-9]
.
This technique proposes the tuning online adaptive gains of
classical PI controller based on Self tuning FL.
In other hand, the presence of this sensor (expensive and
fragile and require special treatment of captured signals),
causes several disadvantages from the standpoint of drive
cost, encumbrance, reliability and noise problem
[10-12]
.
A variety of different solutions for sensorless ac drives
have been proposed in the past few years. Their merits and
limits are reviewed based on a survey of the available
literature. These techniques are generally based on sliding
mode observer
[6]
or Luenberger observer, MRAS method
[10-11]
and extended Kalman filters
[12]
. The PMSM speed
sensorless DTC schemes are sensitive to motor parameters
variation, especially to the stator resistances that change
with temperature. In addition, the error in the PMSM stator
flux estimation is mainly because of stator resistance
variation. This error influences the stator flux as well as the
electromagnetic torque, which means significant error on
the position which causes a bad selection of state of the
switches and generates instability of the machine operation
[6]
. So far many methods such as MRAS-based and observer
based have been proposed for stator resistance estimation.

MRAS for Speed Sensorless Direct Torque Control of a
PMSM Drive Based on PI Fuzzy Logic and Stator
Resistance Estimator
Ameur Aissa
1,
, K. Ameur
2
, and B. Mokhtari
3

TRANSACTION ON CONTROL AND MECHANICAL SYSTEMS, VOL. 2, NO. 7, PP. 321-326, JUL., 2013.
TRANSACTION SERIES ON ENGINEERING SCIENCES AND TECHNOLOGIES (TSEST)

322
Adaptive Model Reference system (MRAS) algorithm
provide an effective way of improving the robustness of the
control system against parameter variation, load disturbance
due to its simple structure and fast response
[10]
.
This paper presents MRAS for a speed sensorless DTC
of PMSM based stator resistance estimator and Fuzzy Logic
PI controller. The improvements and the effectiveness of
the proposed even against DTC key parameter deviation is
demonstrated by some simulation results.
2. PMSM MODEL
The stator and rotor flux equation of PMSM can be
written in the reference frame of Park in the following form
[3]
:

(
(
(

+
(
(
(

(
(
(

=
(
(
(

0 0
0
e
sq
sd
q
d
q
d
I
I
L
L |
|
|
(1)

While the equations of the stator voltages are written in
this same reference frame in the following form:
0

0
0 0

0
d ds d ds
s
q qs d qs
d ds
r r
d qs e
v I L I
d
r
dt
v I L I
L I
p p
L I
( ( (
(
( ( (
(
= + +
( ( (
(
( ( (
(


( ( (
(
( (
O + O
(
( (
(
( (


(2)
In addition the electromagnetic torque can be expressed:
( ) ( ) e
3

2
d q sd sq e sq
T p L L I I I = +

(3)
The mechanical equation of the motor can be expressed
as flows:

.
e r
l
J T T f O= O
(4)
3. CONVENTIONAL DTC
The methods of direct torque control (DTC) as shown in
Fig. 1 consist of directly controlling the turn off or turn on
of the inverter switches on calculated values of stator flux
and torque from relation (6). The changes of state of the
switches are linked to the changes in electromagnetic state
motor. They are no longer controlled based on voltage and
frequency references given to the commutation control of a
pulse width voltage modulation inverter
[1, 3]
. The reference
frame related to the stator makes it possible to estimate flux
and the torque, and the position of flux stator. The aim of
the switches control is to give the vector representing the
stator flux the direction determined by the reference value
0
0
( )

( )
t
s s s s
t
s s s s
v r I dt
v r I dt
o o o
| | |

}
}
(5)
The DTC is deduced based on the two approximations
described by the formulas (6) and (7)
[4]
:
( 1) ( )
s s s E s s E
k k VT VT + ~ + A ~ (6)
( )
' '
( ) sin
e s r s r
T k k o = = (7)
More over:

2 2

s s s
s
s
s
arctg
o |
|
o

= +

Z
(8)
A two levels classical voltage inverter can achieve seven
separate positions in the phase corresponding to the eight
sequences of the voltage inverter.



Fig. 1. Diagram of DTC control applied for PMSM supplied with a
three-phase inverter with PMW.


Fig. 2. Different vectors of stator voltages provided by a two levels
inverter.

Where:
I(D)F : Increase (Decrease) of Flux amplitude.
I(D)T : Increase (Decrease) of Torque.
Tables 1 have the sequences corresponding to the
position of the stator flux vector in different sectors
(see Fig. 1).



AISSA et al.: MRAS FOR SPEED SENSORLESS DIRECT TORQUE CONTROL OF A PMSM DRIVE BASED ON PI FUZZY LOGIC AND STATOR RESISTANCE ESTIMATOR.
TRANSACTION SERIES ON ENGINEERING SCIENCES AND TECHNOLOGIES (TSEST)

323
TABLE 1
VECTORS VOLTAGE LOCALIZATION


The flux and torque are controlled by two comparators
with hysteresis illustrated in Fig. 3. The dynamics torque
are generally faster than the flux then using a comparator
hysteresis of several levels, is then justified to adjust the
torque and minimize the switching frequency average
[4]
.


Fig. 3. Comparators with hysteresis used to regulate flux and torque.
4. FUZZY LOGIC SPEED CONTROLLER
A proportional-integral controller (PI controller) is a
generic control loop feedback mechanism (controller). A PI
controller calculates an "error" value as the difference
between a measured process variable (speed) and a desired
setpoint. The controller attempts to minimize the error by
adjusting the process control inputs
[9]
.
s
K
K controller PI
i
p
+ =
(9)



Fig. 4. Fuzzy controller input and output membership functions.
Two fuzzy logics searched the gains of the PI speed
controller. Each FL has two inputs and one output. The
inputs to the gains of the PI are the normalized error
between the reference and actual rotor speed, e(k) =w
r
*(k)-
w
r
(k), and the normalized change in Flux error
Ae(k)=e(k)-e(k-1). The selected inputs and output
memberships for the fuzzy estimator are given in Fig. 4, the
centroid defuzzification algorithm is used, in which the
output fuzzy variable value is calculated as the centre of
gravity of the membership function. In addition, the rule
base controlling the defuzzified output according to the
fuzzified input values is given in Table 2.

TABLE 2
LINGUISTIC RULE BASE FOR TWO PI- FUZZY LOGIC CONTROLLER
de e NB NM NS ZE PS PM PB
N P NS ZE PS PM PB PB
ZE PB P NS ZE PS PM PB
P PB PB PS NS ZE PS PM

5. MODEL REFERENCE ADAPTIVE
SYSTEM
Accurate and robust estimation of motor variables which
are not measured is crucial for high performance sensorless
drives. Multitude observers have been proposed, but only a
few
[6, 10-12]
are able to sustain persistent and accurate wide
speed range sensorless operation.
The model reference adaptive system (MRAS) is an
important adaptive controller
[10]
. The MRAS for estimating
rotor position angle and speed is based on a stator current
estimator using discontinuous control. Due to the fact that
only stator currents are directly measurable in a PMSM
drive. In this way, when the estimated currents, i.e., state,
reach the manifold. Comparing to other adaptation
techniques, this method is simple and needs a low
computation power and has a high speed adaptation even at
zero speeds. This method is more stable and robust because
it eliminates the error generated in the adjustment of the
speed
[10]
. The rotor speed and the stator resistance are
reconstructed using the model reference adaptive system
(MRAS). The MRAS principle is based on the comparaison
of the outputs of two estimators. The first is independent of
the observed variable named as model reference. The
second is the adjustable one. The error between the two
models feed an adaptive mechanism to turn out the
observed variable.
From (2) the state space axis stator currents of PMSM
designed as reference model is given by:

A|
s
AC
e
S
1
S
2
S
3
S
4
S
5
S
6

1
1 110

010 011 001 101 100
0 000 000 000 000 000 000
-1 101 100 110 010 011 001
0
1 010 011 001 101 100 110
0 000 000 000 000 000 000
-1 001 101 100 110 010 011

0
+1
0
+1
-1
TRANSACTION ON CONTROL AND MECHANICAL SYSTEMS, VOL. 2, NO. 7, PP. 321-326, JUL., 2013.
TRANSACTION SERIES ON ENGINEERING SCIENCES AND TECHNOLOGIES (TSEST)

324
C BU AX X + + =
-

(10)


Where
| | | |
T
qs ds
T
qs ds
v v U , I I X = =
,
1 0
0 1
1
(
(
(

=
(
(
(
(
(

O
O
=
d
d
r
r
d
l
B ,
l
r
p
p
l
r
A

(
(
(
(
(

=
r
d
e
p
l
C
|
0

In this work, the actual system is considered as the
model reference and the observer is used as the adjustable
one. The rotor speed is included in the (10) that present
current model are relevant to rotor speed. So the state space
axis stator current model is chosen as the state variable:
C BU X A X

+ + =
-

(11)


Where
| | | |
qs ds
T
qs ds
v v U , I I X = =


,
1 0
0 1
1

(
(
(

=
(
(
(
(
(

O
O
=
d
d
r
r
d
l
B ,
l
r
p
p
l
r
A

(
(
(
(
(

=
r
d
e
p
l
C

0
|

The adaptation mechanism is designed in a way to
generate the value of estimated speed used so as to
minimize the error between the estimated and reference axis
stator currents. By adjusting the estimated rotational speed,
the error between the reference and the estimated axis stator
currents from (11) is reduced
[13]
. The error between the
estimated and reference axis stator currents are defined as:
qs qs qs ds ds ds
I I e I I e

= =
(12)
The value of the rotor speed error and resistance stator
error are given as:
r r r r r r
r r r ;

= A O O = AO
(13)
The state error model of the PMSM expressed in the
synchronous reference frame (d_q) is given as follow:
W e A e + =
A
-

(14)
Where
| |
T
qs ds
e e e =
is the error state vector, A

is the state
matrix and is the output vector of the feedback block
defined as follow:
,

(
(
(
(
(

=
(
(
(
(
(

O
O
=
A
d
e
d
q
d
r
r
d
l
I
I
W ,
l
r
p
p
l
r
A
|

According to
[14]
, to ensure the hyperstability of the
system can be achieved, two criterions must be established.
Firstly, the linear time-invariant forward path transfer
matrix,
1
) ( ) (

A
= A pI p H must be strictly positive real
and secondly, the nonlinear feedback (which includes the
adaptation mechanism) must satisfies the following Popovs
criterion for stability.
2
0
0
>
}
Wdt e
t
T

(15)
Where t0 0, is a finite positive real constant, which is
independent of t0.
By using the Popov's theory, we obtain: lim
=e
T
()=0
and therefore, the system of the MRAS speed and resistance
estimations is asymptotically stable.
Finally, the stator resistance and the rotor speed are built
around the following adaptive mechanisms:

+ = O
+ =
}
}
t
i p r
qs qs
t
ds ds
d
s
edt k e k
p
dt I e I e
l
r

0
0
) (
1

)

(
1

(16)
qs
d
e
qs qs ds ds
e
l
q
I I I I e =
(17)
qs qs qs ds ds ds
I I e I I e

= =
(18)
Where k
p
and k
i
are the proportional and integral
constants respectively. The tracking performance of the
speed estimation and the sensivity to noise are depending on
proportional and integral coefficient gains. The integral gain
k
i
is chosen to be high for fast tracking of speed. While, a
low proportional k
p
gain is needed to attenuate high
frequency signals denoted as noises
[10]
.


Fig. 5. Basic MRAS structure.


+
-
v
ds
,v
qs,
I
ds
, I
qs
Reference
Model





Adaptation
Mechanism
Adaptive
Model


AISSA et al.: MRAS FOR SPEED SENSORLESS DIRECT TORQUE CONTROL OF A PMSM DRIVE BASED ON PI FUZZY LOGIC AND STATOR RESISTANCE ESTIMATOR.
TRANSACTION SERIES ON ENGINEERING SCIENCES AND TECHNOLOGIES (TSEST)

325
6. RESULTS OF SIMULATION
Table 3, summarizes the PMSM parameters used in this
simulation
[6]
.
We simulated the system drive for a reference speed of
100 (rd / s) load at startup. At t = 0.1(s), the PMSM is
tracking load equal to 5 (Nm), then from t = 0.02 (s), we
assumed a variation of the stator resistance (see Fig. 6).
The results are obtained using a PI speed controller and PI
controller based on Self tuning FL.
Figure 6 illustrates the evolution of stator resistance,
actual and estimated (delivered by the propose MRAS). The
two quantities are combined in practice, in steady state. As
in Fig. 7, it illustrates estimated speed (rad / s) issued by
MRAS, the speed response is achieved without dip and with
a shorter recovery time which is almost similar with the
actual speed motor.

TABLE 3
PMSM PARAMETERS
Pole pairs 3
Rated power KW (at 50 Hz) 1.5
Rated voltage (V) 220/380
Rated Flux (Wb) 0.30
Rated torque (Nm) 5
R
s
() 1.4
L
d
; L
q
(H) 0.0066; 0.0058
Flux magnet (Wb) 0.15
J(Kg.m) 0.00176
f
r
(Nm/(rad/s)) 0.0038

Figure 8 shows the stator flux estimation using the
MRAS. We notice that it is not affected by these changes.
Electromagnetic torque follows his rate as shown in Fig. 9.
It can be seen that a rapid and non-overshoot speed response
is obtained show the robustness of the proposed PI
controller based on Self tuning FL (Fig. 7).


Fig. 6. Stator Resistance.

7. CONCLUSIONS
In this paper, a model reference adaptive system has
been presented for estimating the stator resistance and speed
of a PMSM in order to achieve a robust speed sensorless
DTC.


Fig. 7. Rotor speed.


Fig. 8. Stator Flux.


Fig. 9. Electromagnetic Torque.

The effect of the variations of motor parameters such as
stator resistance on speed estimation has been studied. A
novel design of a PI controller with fuzzy adapted gains
applied to the speed regulation of DTC motor is fully
presented.

The obtained simulation results were satisfactory in
terms of estimation errors, robustness and global stability of
the drive electrical system for different operating
conditions.
0 0.05 0.1 0.15 0.2
1
1.5
2
2.5
3
Time (s)
S
t
a
t
o
r

R
e
s
i
s
t
a
n
c
e

(
O
h
m
)

Actual Resistance
Estimated Resistance
0 0.05 0.1 0.15 0.2
0
20
40
60
80
100
Time (s)
R
o
t
o
r

S
p
e
e
d

(
r
d
/
s
)

Actual Speed
Estmated Speed with FLC
Estmated Speed
0 0.05 0.1 0.15 0.2
0
0.05
0.1
0.15
0.2
0.25
0.3
Time (s)
S
t
a
t
o
r

F
l
u
x

(
W
b
)

0 0.05 0.1 0.15 0.2
-5
0
5
10
15
Time (s)
E
l
e
c
t
r
o
m
a
g
n
e
t
i
c

T
o
r
q
u
e

(
N
.
m
)

Torque with classical PI
Torque with FLC
TRANSACTION ON CONTROL AND MECHANICAL SYSTEMS, VOL. 2, NO. 7, PP. 321-326, JUL., 2013.
TRANSACTION SERIES ON ENGINEERING SCIENCES AND TECHNOLOGIES (TSEST)

326
NOMENCLATURE
dq Synchronously rotating frame.
,
sd sq
I I dq stator current components.
sd sq
, dq stator flux components.
e
| Flux magnet.
M Mutual inductance.
l
d
, l
q
Stator d and q axis inductances.
R
s
Stator resistance.
J, f
r
Rotor inertia, friction coefficient.
T
em
Electromagnetic torque.
T
l
Load Torque
p Number of pole pairs.
r
O Rotor angular
T
e
Sampling Period
k
i
, k
p
Integral and proportional (IP) speed controller.
I Identity matrix

REFERENCES
[1] I. Takahashi, Y.Ohmori, High-performance Direct Torque
Control of an Induction Motor , IEEE Transactions on
Industry Applications, Vol. 25, March/April 1989, pp.
257-264.
[2] Mohamed Boussak, Implementation and Experimental
Investigation of Sensorless Speed Control With Initial Rotor
Position Estimation for Interior Permanent Magnet
Synchronous Motor Drive, IEEE trans. on power
electronics, vol. 20, no. 6, nov. 2005.
[3] L. Zhong, M. F. Rahman, W.Y. Hu, K. W. Lim, M. A.
Rahman, A Direct Torque Controller for Permanent
Magnet Synchronous Motor Drives ., IEEE Transactions
on Energy Conversion, Vol. 14, September 1999, pp.
637-642.
[4] B. Mokhtari, A. Ameur, L. Mokrani, B. Azoui and
M.F.Benkhoris, DTC Applied to Optimize Solar Panel
Efficiency, the 35 th annual conference of the IEEE
industrial electronics society iecon'09 ,11. 2009, porto,
portugal.
[5] M.C. Paicu, I. Boldea, G.-D. Andreescu F. Blaabjerg, Very
low speed performance of active flux based sensorless
control: interior permanent magnet synchronous motor
vector control versus direct torque and flux control, IET
Electr. Power Appl., 2009, Vol. 3, Iss. 6, pp. 551561.
[6] A. Ameur, B. Mokhtari, L. Mokrani, B. Azoui, N.
Essounbouli, and A. Hamzaoui, An Improved Sliding
Mode Observer for Speed Sensorless Direct Torque Control
of PMSM Drive with a Three-Level NPC Inverter Based
Speed and Stator Resistance Estimator, In Journal of
Electric Engineering, Vol. 10, Edition-4, 2010, pp. 1-10.
[7] L. Mokrani, R. Abdessemed, A fuzzy self-tuning PI
controller for speed control of induction motor drive, in:
Proceedings of the IEEE Conference on Control
Applications 2003, Piscataway, NJ, USA, 2003, pp. 785
790.
[8] S.M. Gadoue, D. Giaouris, J.W. Finch, Tuning of PI speed
controller in DTC of induction motor based on genetic
algorithms and fuzzy logic schemes, in: Proceedings of the
5th International Conference on Technology and Automation
2005, Thessaloniki, Greece, 2005, pp. 8590.
[9] C. Pewmaikam1, J. Srisertpol, C. Khajorntraidet, Torque
Control with Adaptive Fuzzy Logic Compensator for
Permanent Magnet Synchronous Motor, 2012 International
Conference on System Modeling and Optimization (ICSMO
12), Singapore.
[10] S. Meziane, R. Toufouti and H. Benalla, Speed Sensorless
Direct Torque Control and Stator Resistance Estimator of
Induction Motor Based MRAS Method, International
Journal of Applied Engineering Research (IJAER ), Vol. 3,
N 6, 2008, pp. 733-747.
[11] F. Grouz and L. Sbita, Speed Sensorless IFOC of PMSM
Based On Adaptive Luenberger Observer, International
Journal of Electrical and Computer Engineering, 2010,
PP:149-155.
[12] G. Terorde, K. Hameyer and R. Belmans, Sensorless
Control of a Permanents Magnets Synchronous Motor For
PV-Powered Water Pump Using the extended Kalman
Filter, in Conference on Electrical Machines and Drives
(ICEMD-IEE), No 468, pp. 368-370. Belgium, 1999.
[13] A. Khlaief , M. Boussak , M. Gossa ; A Sensorless Initial
Rotor Position Estimation Scheme for a Vector Controlled
IPMSM Drive, International Journal of Sciences and
Techniques of Automatic control & computer engineering
IJ-STA, Volume 6, N 1, June 2012, pp. 17061723.
[14] Kojabadi, H. M., Ghribi, M. MRAS-based adaptive speed
estimator in PMSM drives, Proc. of 9th IEEE International
Workshop on Advanced Motion Control Conference., pp.
569572, May 2006.

Available online at:
http://tsest.org/index.php/TCMS/article/view/140
Download full text article at:
http://tsest.org/index.php/TCMS/article/download/140/106
Cite this work as:
Ameur Aissa, K. Ameur, and B. Mokhtari, "MRAS for
Speed Sensorless Direct Torque Control of a PMSM
Drive Based on PI Fuzzy Logic and Stator Resistance
Estimator," TSEST Transaction on Control and
Mechanical Systems, Vol. 2, No. 7, Pp. 321-326, Jul.,
2013.