L.

Sbita

J. Electrical Systems 3-2 (2007): 73-87

M. Ben Hamed

Regular paper
Internal Model Controller for Scalar
Controlled Induction Motor Drive:
Design and Experiments

JES
Journal of
Electrical
Systems

This paper deals with the performance analysis and implementation of a robust speed controller.
The robustness is guaranteed by the use of the internal model controller (IMC). Verification of the
proposed robust controller is provided by experimental realistic tests on scalar controlled
induction motor drives. Robust speed control at low speeds and in field weakening region is
studied in order to show the robustness of the speed controller with a wide range of load
Keywords: Robust speed control, induction motor, scalar control, internal model control.

1. INTRODUCTION
Induction motors are relatively inexpensive and rugged machines. Consequently much
attention has been given to induction motor control for starting, standstill, speed reversal,
position controletc. The induction motor torque is dependent both on the flux and the
speed. But neither relationship is linear, a fact that complicates the design of the control
system for induction machines. Thanks to scalar and Field orientation controls, induction
machines are used for wide field of industrial applications [21].
Not only the coupling behavior of the induction motor consists a great problem but also
the variation of their parameters mainly according to temperature [1]. Researchers have
used various types of closed loop controllers for the rotor speed of the induction motor [18].
Among these controllers, the proportional integral derivative controllers are widely used in
the outer speed loop. However, the use of this type of controller is more sensitive to
parameter variation. To overcome this problem adaptive and robust control were used. The
adaptive control imposes a very computation burden while H robust control requires the
knowledge of the limit of the disturbance [7], [8] and [10]. The sliding mode control is also
used in the speed loop but the most disadvantage of this type of controller is the high
switching frequency [22]. Recently, speed controllers based on artificial intelligence
techniques such as fuzzy logic and neural network based controllers have been proposed
[5], [9], [13] and [20]. Since these approaches do not require the knowledge of a
mathematical machine model, the algorithm would remains robust despite of parameter
deviations and noises measurements. However, the computation expenses and the
requirement of expert knowledge for the system setup have seriously restricted their
applications in practice.
In this paper, an internal model controller is developed for the feedback of the scalar
controlled induction motor. This IMC controller has the advantage of robustness, ease of
design, good responses in transient state and in field weakening region. Details of the
proposed IMC controller will be given in the following sections. Performance of the
developed speed controller and its accuracy are verified by performing a series of
experiments. These consist of step, acceleration, deceleration in low speeds and in field
weakening region under both no load and loaded conditions.

Research Unity of Modeling, Analyze and Control Systems National Engineering School of Gabs (ENIG)Tunisia. [email protected], [email protected]
Copyright JES 2007 on-line : journal.esrgroups.org

L. Sbita & M. Ben Hamed: Internal Model Controller for Scalar Controlled Induction Motor Drive

2. PROPOSED SCHEME OF THE DRIVE SYSTEM

Target

Internal
Model

Voltage
inverter
Scalar control

IM

Load

The block diagram of the proposed induction motor drive system is shown in Figure 1.
The closed loop control scheme of the induction motor is based on the internal model
controller for the rotor speed [15]. The strategy to assume the decoupling of the induction
motor is the scalar control using voltage source inverter [16] and [17].

IM
Model
Figure 1: Block diagram of induction motor drive.

2.1 Scalar control induction motor

Scalar control strategy is designed based on the steady state operation. Based on the
mathematical equations governing electrical dynamic of an induction motor in a
synchronous rotating frame in the steady state, we obtain:
vds = rs ids s qs

(1)

vqs = rs iqs + s ds

(2)

vdr = rr idr (s )qr

(3)

vqr = rr iqr + (s )dr

(4)

The d and q axis can be referred in a space vector if they are respectively placed at real
and imaginary axis. Hence,
vs = vds + jvqs

(5)

is = ids + jiqs

(6)

ir = idr + jiqr

(7)

s = ds + j qs

(8)

r = dr + j qr

(9)

Employing (1) and (2) in (5) yields:

vs = rs is + js s

(10)

Similarly, the rotor voltage is defined as:

vr = rr ir + jr r

(11)

The stator and rotor flux are:

ds = Ls ids + M idr

74

(12)

J. Electrical Systems 3-2 (2007): 73-87

qs = Ls iqs + M iqr

(13)

dr = Lr idr + M ids

(14)

qr = Lr iqr + M iqs

(15)

Based on the equation system and equations (5)-(8), the stator and rotor flux in vector space
are defined by:
s = Ls is + M ir

(16)

r = Lr ir + M is

(17)

By replacing the flux with their expressions, (10) and (11) become:
vs = rs is + j s [Ls is + M ir ]

(18)

0 = rr ir + j r [Lr ir + M is ]

(19)

In what follows, we use the well known equivalent single phase model transformed to the
stator where we consider the magnetic leakages are totalized and grouped in the rotor and
designed by N e s [12]. Figure 2 shows this model.
is

Vs

rs

N es

V's

Rr/g

Figure 2: Single phase equivalent circuit of the induction machine.

where Rr ' and N e represent respectively the rotor resistance and total leakage inductance
located in the rotor.
The evolution of the stator synchronous field round the rotor produces an electromagnetic
power which will be transmitted to the rotor at synchronous speed. The electromagnetic
power is defined as [19]:
Pem = Tem s = 3

R 'r
Vs2
2
g R 'r
2

+ (N e s )
g

(20)

By assumption, we neglect the stator resistance and we obtain:

Tem = 6s 2

R 'r
fr
2

R 'r
2

+ (N e 2 )
f

(21)

where fr is the slip frequency.

75

L. Sbita & M. Ben Hamed: Internal Model Controller for Scalar Controlled Induction Motor Drive

The mechanical dynamics of the induction motor is expressed as:

J

d
= p(Tem TL )
dt

(22)

If we give the zero value to the load torque (TL ) , the transfer function which defines the
relation between the rotor speed and s is given as:
3p s

k
=
=
s
R 'r Js + (R 'r f + 3p s )
1 + s

(23)

Here p is the number of poles pair, f is the viscous coefficient, is the time constant and
k is the static gain of the system.
Hence, the scalar induction motor model is represented by a first order system. To identify
the characteristics of this system model, a standard recursive least square (RLS) method is
used.
2. 2 Identification with RLS algorithm
The recursive least square identification algorithm is described by the following steps
[14]:

1. Initial conditions: the initial value of the estimated parameter vector is set equal to
zero. The initial covariance matrix is assumed to be diagonal matrix with large positive
numbers.

2. Compute estimate y (k )


y (k ) = T (k 1)x(k )
(24)
3. Compute the estimation error of y(k )

(k ) = y(k ) T (k 1)x(k )

(25)

4. Compute the estimation covariance matrix at instant k

P (k ) = P (k 1)

P (k )x(k )x T (k )P(k 1)
1+x T (k )P(k 1)x(k )

5. Compute the estimation vector at instant k



(k ) = (k 1) + P (k )x(k )(k )

(26)

(27)

6. Repeat 2-5 until the iteration number (measurement number) or the target error value is
reached.
2.3 Induction motor model identification
The discreet induction motor model is defined as:


y (k )
bz 1
=
G (z ) =

1 + az 1
u(k )
Based on (28), the recurrent equation is given as:
76

(28)

J. Electrical Systems 3-2 (2007): 73-87




y (k ) = bu(k 1) ay(k 1)

(29)

This can be written as:


y(k ) = T (k )x(k )
(30)

is the vector of estimated parameters and x is the regression vector containing old
inputs and outputs of the system to be identified.

Here and x are defined as:

(k ) =

T
b a

(31)

x= u(k - 1) -y(k - 1)

(32)

To identify the estimated parameters, an experimental test was done on a 1kw induction
motor scalar drive system. The drive is excited with a pseudo random binary sequence
(PRBS) in order to obtain a data file input output measurements. An off line identification
based on RLS is performed. The evolution of the output and input measurements are shown
in Figure 3.

Figure 3: Evolution of the input and the output signals.

The identification results are illustrated in Figure 4.

Figure 4: Evolution of the identification parameters.

77

L. Sbita & M. Ben Hamed: Internal Model Controller for Scalar Controlled Induction Motor Drive

Based on these results, the final discreet function can be written as:
Gm(z ) =

(k )
bz 1
=
1 + az 1
s (k )

(33)

Where b = 0.0611 and a = -0.8254.

2.3 Validity of the established discreet induction motor model
The principle of this test is highlighted by these steps:
1. Compute the quantity defined as:
N

RN (i ) =

(k )(k i)
k =1

2 (k )

, if i ; 1.

(34)

k =1

i = sup(na , nb + d ) with na is the order of the denumerator and nb is the order of the
numerator.

2. Verify the condition validity given as:

RN (i )

2,17
N

(35)

In our application N = 716 . Then, we get RN (i ) 0, 0811 .

The theoretical computation leads to:
RN (0) = 1,
RN (1) = -0.0733,
RN (2) = -0.0544,

(36)

RN (3) = -0.0466.

As shown by these results, the condition (35) is already verified. Therefore, we conclude
that the established induction motor model is valid.
2.4 Internal model controller
The speed controller is based on the principle of internal model control which makes it
robust and immune to disturbances (parameter variations and external load torque
disturbance) [2] and [3]. The internal model controller is consisting of mainly the controller
and the discreet system model. The block diagram of the structure of internal model
controller is given in Figure 5.
When the process is linear and so as, it can be described by a transfer function G (z ) ,
based on the bloc diagram of the internal model controller (Figure 5), we establish:

78

J. Electrical Systems 3-2 (2007): 73-87

Yc(z)

D(z)

Internal Model
Controller
Q(z)

Process
G(z)

u(z)

Process
Model
Gm(z)

Y(z)
Ym(z)

Figure 5: Block diagram of internal model control.

y(z ) = G (z ) u(z ) + D(z )

(36)

ym(z ) = Gm(z ) u(z )

(37)

The command u(z ) is expressed as:

u(z ) = Q(z ) yc(z ) (G (z ) Gm(z ))u(z ) D(z )

(38)

It can be simplified as:

1

u(z ) = 1 + Q(z )(G (z ) Gm(z )) Q(z )[yc(z ) D(z )]

(39)

where yc(z ) denotes the target value of the output value y(z ) .
By replacing u(z ) with its value in (36) yields
1

y(z ) = G (z ) 1 + Q(z )(G (z ) Gm(z )) Q(z )[yc(z ) D(z )] + D(z )

(40)

Here, we distinguish two circumstances:

We suppose that the modeling system is perfect. Therefore G (z ) = Gm(z ) . Using this
condition, we obtain:

y(z ) = G (z )Q(z )[yc(z ) D(z )] + D(z )

(41)

The aim is to assume that y(z ) = yc (z ) in steady state in tracking and regulation. It is clear
that if we choose Q(1) = G (1)1 , then y() = yc () if k converges to . The writing
Q(1) = G (1)1 is possible only if the process is stable.

Now, the modeling process is not perfect. To eliminate the effect of the modeling error, a
low pass filter F (z ) is added. In this case:
1

y(z ) = G (z ) 1 + Q(z ) F (z )(G (z ) Gm(z ) Q(z )[yc(z ) F (z )D(z )] + D(z )

(42)

From this equation, it is clear that to eliminate the disturbance and to assume
y() = yc () if k converges to , we must choose Q(1) F (1)Gm(1) = 1 . The easy way
is to choose a low pass filter with a unit static gain.
In practice, the low pass filter is used not only to eliminate the modeling error but also to
make the controller physically realized. It has an adjustable parameter. This adjustable
parameter can be adjusted on line and off line in order to make the internal model controller
79

L. Sbita & M. Ben Hamed: Internal Model Controller for Scalar Controlled Induction Motor Drive

more robust about modeling errors. Hence, the final form of the internal model controller is
given as:
Q(z ) = Q0 (z )F (z )

(43)

The model internal controller is applied for only stable system.

To obtain the final form of the internal model controller, five steps are to be followed:
Step 1: the zeros of Q0 (z ) are the poles of the model system.
Step 2: the poles of Q0 (z ) are chosen as
the stable zeros of the model system,
the reverse of the instable zeros of the model system,
a pole zero is added at each zero with a negative real part,
Step 3: a pole zero is added to Q0 (z ) ,
Step 4: to obtain a zero error in the steady state, the controller gain is chosen so that
Q0 (1)G (1) = 1 ,
Step 5: a low pass filter is added in order to overcome the problems caused by the modeling
error. In general [4] and [6]:
F (z ) =

(1 )z
z

(44)

where 0 < < 1 .

The robustness of this controller is achieved by a good choice of the coefficient .
The final form of the IMC controller can be written as:
Q(z ) = Q0 (z )F (z )

(45)

Now, in order to determinate u(k ) , we calculate the product between F (z ) and u 0 (k ) .

Therefore, we obtain:
u(k ) = F (z )u 0 (k )

(46)

Replacing F (z ) with its value, (46) yields

u(k ) = u(k 1) + (1 )u 0 (k )

(47)

If is nearby 1 ( 1 converges to zero), then, u(k ) remains nearby u(k 1) although

the difference process model occurs which prove the robustness of this controller. Hence
the process dynamic in closed loop is fixed by this filter parameter. A several experimental
tests were carried to tune the filter parameter with respect to the compromise between
rapidity and stability of the overall system. The final value used is then fixed to 0.9.
The condition of the internal model control is justified by the Nyquist locality shown in
Figure 6. Based on the Nyquist criterion which announce that a necessary and sufficient
stability condition is that the round number of Nyquist diagram round the critical point
(1, 0) in direct direction have to be equaled to instable zero number of his transfer
function in open loop. In our application, the transfer function in open loop has no instable
80

J. Electrical Systems 3-2 (2007): 73-87

zero and as given in Figure 6, its Nyquist diagram doesnt turn round the critical point (-1,
0) in direct direction. Therefore, based on this criterion, the induction motor model is sable.
Hence, the IMC can be used in the control loop.

Figure 6: Nyquist locality of the induction motor open loop transfer function.
Based on these five steps, the internal model controller is obtained as
u(k ) = b0e(k 1) + b1e(k 2) + a 0u(k 1)

where e is the error between real output and model output, a 0 = , b0 =

b1 =

(47)
1
and
a

(1 )b
.
a

3. EXPERIMENTAL RESULTS
The experimental setup is shown in figure 7. It consists of appropriate hardware and its
software implemented through a processor based digital controller. The major parts of the
drive system are:
IMC Speed controller,
Inverter drive,
A power inverter based on the IGBT transistors and L, C filter,
Scalar control set around a micro-controller,
Pc with an acquisition and control card (10v ) ,
Induction motor (1kw-4.6A) coupled to DC generator (1kw-6.2A) supplying a
variable resistive load.
Some of them are implemented through software and are namely IMC speed controller. The
remaining parts are implemented using the developed hardware.
Many experiments were carried out on the various operating conditions to verify the
performance of the proposed internal model speed controller (low speed operation, reverse
operation, field weakening regionetc.) under no load and load applied. Some selected
results are presented in this section to highlight the significant operating conditions for the
proposed system.

81

L. Sbita & M. Ben Hamed: Internal Model Controller for Scalar Controlled Induction Motor Drive

Figure 7: Experimental setup.

Figures 8 and 9 show speed responses under constant load of 400w and 200zm
respectively. Figures 10-11 show, respectively, the step responses for 700rpm and 2000rpm
with different selected resistive applied load about 400W and 200W. These results justify
the robustness of the overall control system under parameter variation and load
disturbances.
In Figure 11, a wide range of selected resistive load is considered and applied in order to
test the robustness of the proposed control algorithm. In this test, both the field weakening
operation and load torque disturbance are considered. Hence, high speeds require rather
large input voltages. In practice, the voltage must be kept within the inverter ceiling limits
[11] so that the flux is decreased from the rated as the speed increases above rated one
(scalar law strategy at constant power). This method of reducing the flux at high speeds is
called "field weakening". It can be noted that at a high speed of 1800rpm, the control
system achieve a good disturbance rejection and stability. The system power limitation
explains the speed droop occurring at 19s.

(a)

(c)

(b)

(d)

Figure 8: Speeds, control voltage, load current and voltage responses under constant load of 400w

82

J. Electrical Systems 3-2 (2007): 73-87

(a)

(c)

(b)

(d)

Figure 9: Speeds, control voltage, load current and voltage responses under constant load of 200 w

(a)

(c)

(b)

(d)

Figure 10: Speeds, control voltage, current and load responses under variable load of 350 w at t=5s,
450 w at t=6s no load at t=8s, 450w at t=18s and 550 w at t=19s (variable target speed as acceleration
and deceleration).

83

L. Sbita & M. Ben Hamed: Internal Model Controller for Scalar Controlled Induction Motor Drive

(a)

(b)

(c)

(d)

Figure 11: Speed, control voltage, load current and load voltage responses of speed IMC controller
under constant load of 200 w at t=3.5s and 0.5 kw at t=6s (variable target speed as acceleration and
deceleration).

The tests of the algorithm under no load are given in figures 12 to 17. Low speed, field
weakening region and reverse operation are all of studied then tested experimentally. From
these figures, it is seen that the proposed control algorithm has good performance and
stability.

Figure 12: Speed and control voltage under IMC controller (700rpm).

Figure 13: Speed and control voltage under IMC controller (2000rpm: field weakening region).

84

J. Electrical Systems 3-2 (2007): 73-87

Figure 14: Speed and control voltage under IMC controller (variable target speed).

Figure 15: Speed and control voltage under IMC controller (variable low target speed: permanent
target speed of 80rpm).

Figure 16: Speed and control voltage under IMC controller (variable target speed 1000rpm).

Figure 17: Speed and control voltage under IMC controller (variable target speed: field weakening
region, permanent target speed of 2800rpm).

4. CONCLUSION
In this paper, an internal model speed controller for induction motor is designed and
implemented for a scalar induction motor drives system. An identification methodology

85

L. Sbita & M. Ben Hamed: Internal Model Controller for Scalar Controlled Induction Motor Drive

based on the recursive least square algorithm is successfully applied in this work to identify
the parameters of the established mathematical model of the drive system. From the
experimental results using a 1kw induction motor drive system, it is shown that the
proposed speed controller over the entirely speed range and under a wide range of load has
a good performance and stability. The most important advantage of the proposed algorithm
is to provide a robust structure and a simple and easy design compared to other methods.
Due to the drawbacks of the speed sensorless, a simple speed estimator based on the
generation of the controller signal in one hand and an artificial neural networks observer on
the other hand and the torque optimization could be the subject of future follow up research
work.
REFERENCES
[1]

C. Mastorocostas, I. Kioskeridis and N. Margaris, Thermal and slip effects on rotor time
constant in vector controlled induction motor drives, IEEE Trans. Power Electron., vol. 21, no.
2, pp. 495 504, Mar. 2006.

[2]

X. Shao, J. Zhang, Z. Zhao and X. Wen, Adaptive internal Model Control of Permanent
Synchronous Motor Drive System, in Proc. Electrical Machines and Systems, ICEMS 2005,
vol. 3, pp. 1843-1846, Sept. 2005.

[3]

A. M. Segayer and A. M. Jaroushi, Competitive co-evolutionary design of robust internal

model controllers to deal with prescribed large plant uncertainties. Presented at SSD2005.

[4]

P. Borne, G. D. Tongy, J. P. Richard, F. Rotella and I. Zambettakis, Commande et Optimisation

des Processus. Paris: Technip, 1998.

[5]

L. Sbita and M. Ben Hamed, Fuzzy controller and ANN speed estimation for induction motor
drives. Presented at SSD2007.

[6]

J. M. Flaus, Rgulation Industrielle: Rgulateurs PID, Prdictifs et Flous. Paris: Herms, 2000.

[7]

G. Duc and S. Font, Commande H et analyse. Paris: Herms, 1999.

[8]

A. Oustaloop, La Robustesse: Analyse et Synthse des Commandes Robustes. Paris: Herms,

1994.

[9]

O. Omidvar and D. L. Elliott, Neural systems for control. New York: Academic Press, 1997.

[10] Y. Ben Salem et L. Sbita, Synthse H Loop shaping applique la commande vectorielle dun
actionneur asynchrone sans capteur de vitesse. Presented at IEEE JTEA2006. [CD - ROM].
[11] Y. Ben Salem and L. Sbita, A Nonlinear state feedback control for induction motors, Presented
at IEEE. ICIT2006. [CD - ROM].
[12] G. Grellet and G. Clerc, Actionneurs Electriques, Paris: Eyrolles, 1996.
[13] M. A. Denai and S. A. Attia, Fuzzy and Neural Control of an Induction Motor, Int. J. Appl.
Math. Comput. Sci., Vol. 12, no. 2, pp. 221- 223, 2002.
[14] Y. Koubaa, Recursive identification of induction motor parameters, Int. J. Simulation Modeling
Practice and Theory, 12, pp. 363381, 2004.
[15] A. Jaime., A. Joaquin and R. J. Herrera, Internal model control for non linear systems:
application to an induction motor, Int. J. Control Engineering Practice, Vol. 2, no. 6, pp. 1063
1066, Dec. 1994.
[16]

86

F. Zidani, M. S. Nait Said, R. Abdessemed, D. Diallo and M. E. H. Benbouzid, A Fuzzy

Technique for Loss Minimization in Scalar-Controlled Induction Motor, Electric Power
Components and Systems, vol. 30, pp. 625-635, 2002.

J. Electrical Systems 3-2 (2007): 73-87

[17] A. M. Garcia, T. A. Liop and D. W. Novotny, A new induction motor V/f control method
capable of high-performance regulation at low speeds, IEEE Trans. Ind. Applicat., vol. 34, no.
4, pp. 813-821, Jul.-Aug. 1998.
[18] J. Holtz, Sensorless Speed and Position Control of Induction Motor Drives, In Proc. 27th Annu.
Conf. IEEE Industrial Electronics Society IECON'01, Denver, Colorado, 2001, pp. 1554-1562.
[19] I. Boldea and S. A. Nasar, Electric Drives, Taylor and Francis Group, 2005.
[20] J. Fonseca, J. L. Afonso, J. S. Martins and C. Couto, Fuzzy logic speed control of an induction
motor, Int. J. Microprocessors and Microsystems, 22, pp. 523534, 1999.
[21] H. M. B. Metwally, Proposed torque optimized behaviour for digital speed control of induction
motors, Int. J Energy Conversion & Management, 43, pp. 16751688, 2002.
[22] C. K. Lai and K. K. Shyu, A novel motor drive design for incremental motion system via
sliding mode control method, IEEE Trans. Power Electron., vol. 52, no. 2, pp. 7-30, Apr.
2005.

87