Adaptive Stator Resistance Compensator for High Performance Direct Torque

Controlled Induction Motor Drives

Byeong-Seok Lee and R. Krishnm
Motion Control Systems Research Group
The Bradley Department of Electrical and Computer Engineering
Virginia Tech, Blacksburg, VA 24061
Phone: (540)231-4311, Fax: (540)231-3362
E-mail lcramu@vt.edu

Abstract
Direct Torque Control @TC) has drawn the
attention of the motor drives designers because its
implementation requires no position sensor. Crucial to the
success of this scheme is the estimation of electromagnetic
torque and stator flux linkages using the measured stator
voltages and currents. The estimation is dependent only on one
machine parameter, stator resistance. The variation of the
stator resistance, which is usually in the range of 0.75-1.7 times
its nominal value, deteriorates the performance of the drive by
introducing errors in the estimated flux linkages magnitude
and its position and hence in the electromagnetic torque.
Resistance change also skews the torque linearity thus making
the motor drive a less than ideal torque amplifier. Parameter
compensation using stator current phasor error has been
proposed in literature. To obtain the stator current phasor
error, the stator current reference is required which is not
usually available in direct torque control schemes. An analytical
derivation of the stator current phasor reference is derived
systematically from the reference electromagnetic torque and
flux linkages. The error between the stator current phasor
reference and its measured value is a measure of the stator
resistance variation from its set value. For the first time, it is
demonstrated in this paper that DTC motor drive system can
become unstable when the set value of the stator resistance in
the controller is higher than the stator resistance in the machine.
Hence parameter adaptation is not only important for torque
linearity but also for stability of the system is shown in this
paper.

I. INTRODUCTION
High performance inverter-fed induction motor drives with
no mechanicalloptical position sensor are desirable for the
emerging market applications. A control scheme, which
achieves that for induction motor, is the direct torque
controller, also known as direct self-controller. The scheme
uses feedback control of electromagnetic torque and stator
flux linkages. The electromagnetic torque and stator flux
linkages are estimated in stator reference frames using the
measured stator voltages and currents. The machine model is
dependent on stator resistance only. The direct torque
controlled induction motor drives have been studied in
different forms starting with Depenbrock [l] utilizing the
continuous stator flux angle calculation and schemes using
the limited information of the stator flux angle being in one

0-7803-4943-1/98/$10.00 0 1998 IEEE

of the six sixty degrees sextants [2], [3]. A number of

implementations of the scheme have been made based on
how the currents and voltages are measured or estimated. The
stator currents may be obtained using only the dc link current
sensor [4] and the motor line voltages can be reconstructed
inexpensively utilizing the gate signals [5]. Studies using
space vector modulation [6], [7], [8] for this drive have
proved beyond doubt the usefulness and simplicity of the
modulation scheme and it is almost a standard feature in
many ac drives products.
But all of them have precision low speed operational
problems due to the errors induced by varying stator
resistance in the flux and its angle calculator that is central to
the control scheme. The stator resistance change has a wide
variation of 0.75 to 1.7 times its nominal value due to
temperature variation to a large extent and stator frequency
variation to a smaller degree. It deteriorates the drive
performance by introducing errors in the estimated flux
linkages magnitude and position and hence in the
electromagnetic torque estimation, particularly at low speeds.
Note that at low speeds, the stator resistance voltage drops
constitute significant proportion of the applied voltages.
A few control schemes have been proposed to overcome
this parameter sensitivity, which restricts the speed control
range of the drives. Partial but operating frequency dependent
hybrid flux estimator has been proposed for stator resistance
tuning [9], which has the problem of convergence and
slowness of response. Adjustment of the stator resistance
based on the difference between the flux current and its
command [101 has the problem in identifjmg the actual flux
current. Finding stator resistance based on the steady state
voltage equation [111 has the shortcoming of using direct axis
flux linkages which itself has been affected by stator
resistance variations. A recent study [ 121 proposes the use of
stator current phasor error with PI and fuzzy estimators
presenting a good performance for tuning the stator
resistance. The fuzzy estimator seems to outperform the PI
based estimator thereby necessitating the use of intelligent

control techniques to synthesize the adaptation mechanism.

There has been no mention of how the stator current phasor

423

reference is obtained to realize this scheme. No mention of

stability is made in all the proposed schemes either implying
that global stability is guaranteed or no instability is
encountered in realizations.
Due to parameter sensitivity, there is a possible instability
problem arising in this scheme. It is shown by simulation in
this paper that the drive system becomes unstable if the stator
resistance value used in the controller is higher than that of
the machine resistance. Such cases may arise in practice. Two
possible scenarios are: (i) The stator resistance may be lower
than its controller set point nominal value in externally
housed drive system in colder climates as the operating
temperature may be different from the motor temperature at
starting. The controller resistance may correspond to the
operating temperature as that is justifiable for operation in the
parameter uncompensated system. (ii) If the drive system has
parameter adaptation and if its performance is poor, then the
estimated resistance may be higher than the actual motor
resistance at times, leading to instability.
This paper is concerned with a solution to track the stator
resistance so that the performance degradation and a possible
instability problem can be avoided. Proportional-Integral (PI)
adaptive compensator using only the measured stator currents
is applied [12]. An analytic expression to evaluate the stator
current command from the torque and stator flux linkage
commands is derived and presented. This scheme requires
two filters and one PI controller apart from the existing
electromagnetic torque and stator flux linkages calculator in
the drive controller. A signal proportional to stator resistance
change is developed using the error between the reference
and actual stator current phasor. This error is processed

through a PI controller for application in the controller. Stator

resistance parameter adaptation results in restoring the
precise and accurate estimation of stator flux linkage
magnitude and its position in the controller. Note that this is
the key to successful operation at low speeds. A set of high
bandwidth current and voltage sensors or estimators is
required for the implementation of the solution. The
performance of the controller is validated by extensive
dynamic simulation studies for a wide variety of operating
conditions including flux weakening.
VECTOR
CONTROL
SCHEME
11. DIRECT

The principle of operation of the direct vector controlled

induction motor drive is briefly discussed in this section. A
block diagram schematic of the drive scheme is shown in Fig.
1. The block enclosed in dotted lines is excluded in the basic
scheme. The scheme uses the feedback control of torque and
stator flux linkages, which are computed from the measured
motor voltages and currents. The method uses stator
reference frame model of the induction motor and the same
reference frame is used in the implementation thereby
avoiding the trigonometric operations encountered in the
coordinate transformations of other reference frames. This is
one of the advantages of the control scheme. The stator q and
d axis flux linkages h,,, hds can be obtained through the
integration of the difference between the phase voltage and
the voltage drop in the stator resistance as,

424

7
1
:1

and flux linkage phasor is,

R ,,
R

(3)

0.1
0

T k ,s

and electromagnetic torque is given by,

Te = q2 i2q S A d s-idsAqs)

(4)

where the direct and quadrature axis components are

obtained from the abc variables using the following
transformation,

lqs = I,,

Ids

= -(its - ib,)

loo

-50

J5

I
2

Thne,s'

4 ,

where Rs is estimated stator resistance, i,, ib, and i,, are

stator phase currents, P is the number of poles. Similar
transformation is applicable to voltages and flux linkages as
well. It is assumed that the voltages and currents are balanced
and hence zero sequence voltages and currents are
nonexistent. To obtain uniformly rotating stator flux, note
that the motor voltages have to be rotated uniformly too. This
imposes a requirement of continuously variable stator
voltages with infinite steps, which is not usually met by the
inverter due to the fact that it has only eight switching states
and its switching frequency is limited by thermal factors. The
limited states of the inverter create distinct discrete
movement of the stator voltage phasor, vs , consisting of the
resultant of vqs and Vds . In spite of this limitation, almost
continuous and uniform flux linkage phasor is feasible with
these discrete voltage states due to their integration over time
as seen from equations (1) and (2). Stator resistance is the
only motor parameter that affects all the key variables for
feedback in the direct vector controller is seen from equations
(1) to (4).
A mismatch between the controller set stator resistance and
its actual value in the machine can create instability as shown
in Fig. 2. This figure shows the simulations for a step stator
resistance change from 100% to 80% of its nominal value at
0.5 s and rated torque, command is applied at 0.1 s. The drive
system becomes unstable if controller instrumented stator
resistance is higher than its actual value in the motor. This
may be reasoned as follows. As the motor resistance
decreases in the machine, then its current increases for the
same applied voltages, which increases the flux and
electromagnetic torque. The controller has the opposite effect
in that the increased currents, which are inputs to the system,
cause increased stator resistance voltage drops in the

.s

0I

-""Av*,

Tk,s'

Fig. 2. Instability due to parameter mismatch

for a step stator resistance change.
calculator resulting in lower flux linkages and
electromagnetic torque estimation. They are compared with
their command values giving larger torque and flux linkages
errors resulting in commanding larger voltages and hence in
larger currents leading to a run off condition.
The instability result under a step change is questionable as
the stator resistance does not change in a step manner in
practice. A linearly decreasing stator resistance is simulated
and the performance is shown in Fig. 3. Even for such a
gradual change of stator resistance, note that the system
becomes unstable. The controller calculated electromagnetic
torque and stator flux linkages are equal to set values and
contrary to the real situation in the machine. Therefore, any
scheme using these values for parameter compensation would
not be effective. For example, the airgap power feedback
control for parameter compensation in indirect vector
controlled drive system is successful but it will not work in
the DTC drive system.
The parameter mismatch between the controller and
machine also results in a nonlinear characteristic between
torque and its reference making it a nonideal torque
amplifier. This will have undesirable consequences in a
torque drive and to a smaller extent in the speed controlled
drive systems. The motor resistance adaptation is essential to
overcome instability and to guarantee a linear torque

425

.______

0.3

R , o.2
OJ
0

is

Tk,s

-50

L....

""

lis,

2--

v.s

1.-

T&,s

Tk,s

/--/-A
1

T&,s

-20
- .

15

,j

-~
--

N .m

0.4

Th,r

1.

, 0.3

Low Pass

Limiter

Filter

50 --

f,,

PI Controller
I
and Limiter
R,,

Fig. 4. Block diagram schematic of the adaptive

stator resistance compensator.

is,

ow pass
Filter

0.2
0.1
04

I
I

motor temperature and to a smaller extent by the varying

stator frequency. The incremental value of stator resistance
for correction is obtained through a PI controller and limiter.
The current error goes through a low pass filter, which has
very low cutoff frequency in order to remove high frequency
components contained in the stator feedback current. This
low pass filter does not generate any adverse effect on the
stator resistance adaptation if the filter time constant is
chosen to be smaller than that of the adaptation time constant.
This incremental stator resistance, AR,, is continuously
added to the previously estimated stator resistance, Rso . The
final estimated value, kS,is obtained as the output of
another low pass filter and limiter. This low pass filter is
necessary for a smooth variation of the estimated resistance
value. This final signal is the updated stator resistance and
can be used directly in the controller. The above algorithm
requires the stator current phasor command, which is a
function of the commanded torque and commanded stator
flux linkages. An analytic procedure to evaluate the stator
current command from the torque and stator flux linkages
commands is presented in the following.
B. Stator Current Phasor Command

Th,s

Fig. 3. Instability due to parameter mismatch for a linearly

decreasing stator resistance change.
amplifier in the direct torque controlled drive. A stator
resistance parameter adaptation scheme is presented in the
next section to achieve these objectives.
111. ADAPTIVE
STATOR
RESISTANCE
COMPENSATION

A. Scheme

A block diagram schematic of the applied stator resistance

compensation scheme is shown in Fig. 4 and its incorporation
in the drive schematic is given in Fig. 1 and shown in dotted
lines. This technique is based on the principle that the error
between the measured stator feedback current phasor

The stator feedback current phasor magnitude is is

obtained from the q and d axis measured currents as,
is = Ji;,

+i&)

The stator command current phasor magnitude i,* is

derived from the dynamic equations of the induction motor in
the synchronously rotating reference frame using the torque
command T: and stator flux linkage command 1;. The flux
linkages, rotor equations, and torque are given by,

magnitude is and its command i: [121 is proportional to the

stator resistance variation which is mainly caused by the

426

flux linkages, ,.,A:

i:,

are q-d axis rotor flux linkages,

i& are q-d axis stator currents, i:,. , iz,. are q-d axis rotor

currents, osl is the slip speed given by (a, -a,), and P is

the number of poles. The resultant stator flux linkage h, is
assumed to be on the direct axis. This step is to reduce the
number of variables in the equations by one. Moreover, it
corresponds with the reality that the stator flux linkages is a
single resultant phasor. Hence aligning the &axis with stator
flux linkage phasor,

A,

= 0, pheg, = 0,

= h,

Similarly rotor d axis equation can be derived in the steady

state using (13), (14), and (16) as,

R J , .e+
*
-__
Lm Ids -aslLmi(

XZ,

~ecause

and :i

are lm values,

(14)

( 6,

229,s

are

(23)

C-bh

(13)

(15)

led:

(12)

Substituting (12) in (7) to ( l l ) , the resulting dynamic

equations are,

3P
T =--ie

4
1 and

found by solving (21) and (22) simultaneously, and are given

by,

o = L,i& + L,i$, h, = LSi&+ Lmi&

he,, = L,i;, + LmiES,hd,= L,izr + Lmi&

(22)

l -LsLr
z ) +R
Lm
,h:=O

Equation (23) gives two solutions for izi and the appropriate
solution is the one that gives a smaller value. Finally, the
stator current command magnitude i: is calculated using the
equations (18) and (23) as,

(17)

ii

=,/.d;:p)

Then the q-axis current command is directly obtained from

(17) Using the torque COI11Illilflc1 T,* and stator flux linkage

command A*, as,

T,*
2_
2_
ie* = _
qs
3P

(18)

Note that dq axis stator current commands are independent of

stator and rotor resistances and so is the stator current phasor
command. In the implementation, these values are calculated
and stored for various operating conditions including fluxweakening region.

Iv.RESULTS
Because h, is a constant, the following relations are derived
from (13) as,
= -Lmpi&

(19)

L,pi& = -LmpiZr

(20)

L,pi:,

In steady state, pi& = pi:, = 0 . Therefore, all rotor variables

in (15) can be removed using (13), (14), (19) and (20). The
resultant rotor q axis equation that is a function of stator
variables only is given by,

Dynamic simulations are performed to validate the

performance of the torque controlled drive system with the
compensation technique. The induction motor and drive
system details used in the simulation are given in the
appendix. In the implementation of the drive algorithm, only
six nonzero voltage vectors are used and two zero voltage
vectors are excluded. It is noted that this has no impact on the
basic Performance of the system. Because the stator
resistance voltage drop is a considerable portion of the
applied voltage at low speed, the performance of the system
at low speed due to variation in the stator resistance
deteriorates much more compared with that in the high speed

427

0.6

0.4

-.

02 ..
0 ,
0.0

0.5

1.o

04
0

T k s

lo

Thre.s'

T&,s'

-I

N . m -10o~
0.0

0.5

1.o

Tk,s

0.6

A , , 0.4
. s 0.2

a s , 0.4

v .s

0
0.0

0.5

I .o

v+q!nd--

--

1
I

--

0.2

Th,s

Fig. 5a. Step response for a parameter uncompensated

system
range. Therefore, the rotor speed is limited to 0.1 p.u. in the
simulations to study the most affected region of performance
in this paper.
Figs. 5a and 5b show the simulations for a step change in
stator resistance in a parameter uncompensated torque drive
system, and a compensated system, respectively. The system
controller has the nominal value of stator resistance and after
0.5 s, stator resistance is changed to twice its nominal value
and the corresponding effects are studied. Rated torque
command is applied at 0.1 s after the stator flux linkage has
reached the steady state. Parameter adaptation is initiated
after 0.2 s. In the uncompensated system, right after the
change of resistance, the generated electromagnetic torque
decreases nearly by 25% in steady state with much higher
drop during the transient. The stator flux linkages and stator
current show similar worsening. In the compensated system,
it is noticed that stator resistance estimate suffers in the initial
transient state and converges gradually to its final actual
value in steady state. All the other variables also have initial
transient state, but reach their final values in steady state. A
step variation in the stator resistance is rather an extreme test
and not a significant case encountered in practice.
In actual operating conditions, the rate of change of
temperature is very slow and so is the stator resistance. Fig. 6
covers this situation. Stator resistance is increased linearly

20
Tc,

N.m

-.

10 --

0 .

15

20

Time,s'

1
~

II

-10

0.6

a s, 0.4

10

--

v . s 0.2 --

,
I

from nominal value to twice its nominal value in 7 s, and

decreased linearly from twice to 0.8 times its nominal value,
The stator resistance estimate, stator current, torque, and

428

0.6

O J
0

20 --

is,
A

1
10 --

s - ~

U),

.-

m.
10 0 .i

-10

v .s
A s 3

:/:____I

12

T k s

0.6

0.2

04
0

I2

Fig. 7. Ramp response with the step change of torque

command.

l2

-.-A

Tk,s

Fig. 8. Ramp response in field weakening mode.

It is demonstrated, for the first time, that instability
occurs in the direct self-control drive if controller set
stator resistance is greater than its actual value in the
machine regardless whether the stator resistance varies
linearly or in step fashion.
An adaptive stator resistance compensation scheme is
applied to eliminate the stator resistance parameter
sensitivity using only the existing stator current
feedback. It is simple to implement and its realization is
indirectly dependent on stator inductances. Since flux is
controlled in the machine, the inductances used in the
computation of stator phasor current command will be
constants.
A procedure for finding the stator current phasor
command from the torque and stator flux linkage
commands is derived to realize the adaptation scheme.
Stator current phasor command derived from this
procedure is independent of stator and rotor resistances
of induction motor.
The scheme is verified with dynamic simulation for
various operating conditions including flux-weakening
mode. This scheme is successful even in the face of
rapid changes in the stator resistance such as step
changes. The scheme adapts very well without transients
even for simultaneous variations of torque and flux
linkages command while the stator resistance is varying.
The PI controller in the stator resistance adaptation loop
gives a performance equal to that of the intelligent

stator flux linkages are tracking the references very closely

and experience no oscillation and transient during tracking.
Note that the instability problem encountered with no
parameter adaptation is solved with this adaptive
compensation scheme.
Step changes in torque cctmmand from 1 to 2 p.u. and then
2 p.u. to -1 p.u. while the stator resistance is ramped up have
no adverse effects on the drive system as shown in Fig. 7.
The flux linkages hardly vary during these changes and
torque linearity is perfectly maintained between the torque
and its command in this adaptation scheme.
Finally the dynamic performance in the flux weakening
range is shown in Fig. 8. The stator resistance is ramped fiom
nominal value to twice its nominal value. Stator flux linkages
command and torque command are proportionally decreased
from and increased linearly again to its original reference
values. The tracking of motor variables and stator resistance
is achieved with hardly any transients thus proving the
effectiveness of the adaptive controller in the fluxweakening region also.

V.CONCLUSION
The following are considered to be the original
contributions of this study:

429

R. Krishnan, Lecture Notes on Analysis & Design of

controller such as a fuzzy controller. Since the design

and implementation of a PI controller is easier compared
to the intelligent controller, this study demonstrates the
sufficiency of the PI controller for parameter
compensation. It is also possible that the PI controller
performance is equally good because the derived stator
current phasor reference is independent of stator and
rotor resistances which may not have been the case with
the intelligent controller implementation.

Motion Control System, VPI&SU, Blacksburg, 1997.

T. G. Habetler, F. Profumo, M. Pastorelli, and L. M.
Tolbert, Direct torque control of induction machines
using space vector modulation, IEEE Trans. on
Industry Applications, Vo1.28, No.5, pp. 1045-1053,
Sept./Oct. 1992.
M. P. Kazmierkowski and A.B. Kasprowicz, Improved
direct torque and flux vector control of PWM inverterfed induction motor drives, IEEE Trans. on Industrial
Electronics, Vo1.42, No.4, pp.344-350, Aug. 1995.
T. G. Habetler, F. Profumo, G. Griva, M. Pastorelli, and
A. Bettini, Stator resistance tuning in a stator flux field
oriented drive using an instantaneous hybrid flux
estimator, Conf. Record, EPE Conf., Brighton, UK,
V01.4, pp.292-299, 1993.
T. Okuyama, N. Hujimoto, and H. Hujii, A simplified
vector control system without speed and voltage sensors
- Effect of saturating errors of control parameters and
their compensation, Electrical Engineering in Japan,
V01.110, No.4, pp.129-139, 1990.
R. J. Kerkman, B. J. Seibel, T. M. Rowan, and D.
Schlegel, A new flux and stator resistance identifier
for AC drive systems, Conf. Record, IEEE-IAS,
Orlando, Florida, pp.310-318, Oct. 1995.
S. Mir, M. E. Elbuluk, and D. S. Zinger, PI and fuzzy
estimators for tuning the stator resistance in direct
torque control of induction machines, IEEE Trans. on
Power Electronics, Vol. 13, No.2, pp.279-287, March
1998.

APPENDIX
TABLE I
INDUCTION MOTOR
AND DRIVESYSTEM
PARAMETERS

I 5 HP. 200 V, 60 Hz. 3 Phase. Y Connection, 4 Poles, 1766.9 RFM

Stator resistance
Rotor resistance
Stator self inductance
Rotor self inductance
Mutual inductance
Rotor inertia

0.277 n
0.183 n
0.0553 H
0.0560 H
0.0538 H
0.01667 ~g . m 2

Viscous friction coefficient

Rated torque
Rated stator flux linkage
Ratedstator current
Adaptation controller proportional gain
Adaptation controller integral gain
Stator resistance sampling time
Cutoff frequency for current low-pass filter
Cutoff frequency for resistance low-pass filter

1 0 N.m.s/rad
20.178 N . m

I
I 0.433 v .s
I 18.48A
I 0.01

0.0005
5ms
2Hz
0.24 Hz

NOMENCLATURE

REFERENCES

Subscripts indicating stator d, q quantities.

Subscripts indicating rotor d, q quantities.
Subscripts indicating stator phase quantities.
Electromagnetictorque and command, N-m.

M. Depenbrock, Direct self-control (DSC) of inverterfed induction machine, IEEE Trans. on Power
Electronics, Vo1.3, No.4, pp.420-429, Oct. 1988.
I. Takahashi and T. Noguchi, A new quick response
and high efficiency control strategy of an induction
motor, IEEE Trans. on Industry Applications, Vol. IA22, NOS,pp.820-827, STt./Oct. 1986.
1. Takahashi and Y. Ohmori, High performance direct
torque control of an induction motor, IEEE Trans. on
Industry Applications, Vol. IA-25, No.2, pp.257-264,
Mar/Apr 1989.
R. Joetten and G. Maeder, Control methods for good
dynamic performance induction motor drives based on
current and voltage as measured quantities, IEEE
Trans. on Industry Applications, Vol. IA-19, No.3,
pp.356-363, MayIJune 1983.
R Krishnan and P. Pillay, Parameter sensitivity in
vector controlled AC motor drives, Proc. IEEE
Industrial Electronics Conference, Vol. 1, pp.2 12-218,
Nov. 1987.

Estimated electromagnetic torque, N-m.

Stator flux linkage and command, V-s.
Estimated stator flux linkage, V-s.
Stator feedback and command current, A.
Stator and rotor resistance, SZ .
Estimated stator resistance, 0.
Stator and rotor self-inductancesper phase, H.
Mutual inductance per phase, H.
Stator and rotor electrical angular speeds, rads.
Time constant for current low-pass filter, s.
Time constant for resistance low-pass filter, s.

430