Vous êtes sur la page 1sur 6

Aachen, Germany, 2004

2004 35th Annual IEEE Power Electronics Specialists Conference

DTC of Open-End Winding Induction Motor Drive Using Space Vector Modulation
With Reduced Switching Frequency
Arbind Kumar, Shldent Member IEEE

K. Chattejee

B.G. Fernandes

Department of Electrical Engineering,


Indian Institute of Technology Bombay,
Powai, Mumbai- 400 076, India
E-mail: baf@ee.iitb.ac.in

- A new control strategy for reducing the switching


frequency of direct torque controlled (DTC) induction motor
with open-end winding configuration is proposed. Two
independent two-level inverters are feeding both ends of the
three-phase stator winding. In order to achieve the required
voltage vector, one of the inverters is clamped to a particular
voltage vector for a minimum of one sampling interval, while
the other one is controlled using space vector PWM technique.
This arrangement applies a 3-level line voltage waveform to the
machine. Since this configuration is generally used in high
power applications, the effective switching frequency of one of
the inverters is reduced. Also, isolating the input DC source of
the inverters can eliminate the flow of zero sequence current.
Hence a significant reduction in size and improvement in
overall efficiency is achieved. Simulation studies have been
carried out for the proposed scheme and results are compared
with the conventional space vector PWM-DTC by taking the
equivalent 2-level inverter into account.

Abstract

I INTRODUCTION
Induction motor drives are generally used for low and
medium power applications. However, researchers have
moulded this motor even for higher power applications by
making suitable changes in power and control circuit
configurations. One such change is the open-end, threephase stator winding which is fed by two separate inverters.
Direct torque control (DTC) of induction motor was
proposed around two decades ago [I]-[2]. The scheme
presented in [I] was implemented for high power
applications using the induction motor having an open-end
winding configuration [3]. However, the DTC technique
presented in [3] results in:
Torque and speed fluctuations that leads to acoustic
noise and vibrations.
Higher ripple in the stator current that can cause high
power loss and hence heating of the machine.
Use of a three-phase reactor to reduce the zersequence current, thus making the system bulky and
less efficient.
Various control techniques for the inverter-.fed induction
motor drive with open-end windings are also discussed in
[4]-[5]. Direct Self Control technique has been proposed for
this machine with open-end winding configuration [6]-[8].
Both ends of the stator winding are connected to two threelevel inverters. The use of a three-level inverter at both ends
improves the performance of the machine but can lead to an
increase in both cost and complexity. A reactor has been

07803-8399-0/04/$20.00
02004 IEEE.

used on the A C D C side to reduce the flow of zero sequence


currents. In [9]-[ IO], the open-end winding Configuration
has been proposed for high power electric vehiclehybrid
electric vehicle (EVAIEV) propulsion systems. In [ I I],
space vector pulse width modulation technique is used to
control the output voltage of both the inverters connected at
both ends of the motor winding. It should be noted that the
switching frequency capacity of both the inverters is same.
This paper proposes ;I new scheme in which one of the
inverters is switched al. high frequency, and the other at low
switching frequency.
I1 OPEN-END WINDING INDUCTION MOTOR FED BY
2-TWO LEVEL INVERTERS
A schematic of the open-end winding induction
motor drive is shown in Fig.1. A two-level inverter, INVI,
feeds the three ends of the stator winding R Y B, and the
other three ends R Y and B are connected to another two
level inverter called INV2. INVl and INV2 are connected to
separate dc sources (bmeries) of magnitude Vdc/2. This can
be achieved by simply partitioning the set of batteries. This
results in a significant reduction in space, which is a very
important aspect in on-board ship propulsion applications.
Induction motor voltage equations can be derived in
the satne manner as if the machine was connected to a single
two-level inverter as follows [9], [14]:
Phase voltages in tenns of switch position of respective
phases are given by:

where SR,S, & S , = 1 or 0 and it depends on the inverter


leg switching state. Connecting R Yand B together forms
the neutral of the machine, denoted by n, and thus results
in a star connected stator winding. Equation (1) can be
written in terms of the Dole voltaees as:
L

(2)
Voltage equations for open-end winding induction motor
can be written as

1214

2004 35th A n n u l IEEE Power Elecrronics Speciolisrs Conference

RR
YW

+ VRG - VRG

(3)

= VYG + VGG - YG

(4)

= VRC

Stator voltage space vector is given by

V, = vsd + jvsq

(5)
Y B E = VBC + vCc. - V8C
As both inverters are connected to separate dc sources, they
can he assumed to he two independent nodes. Thus the sum
of the three-phase currents can he assumed to he zero. If the
summation of the three-phase currents is zero, summation of
the three-phase voltages can also he assumed zero. Hence
using (3)-(S), and solving for VGGtwe get

vCG

= [(vCc.

+ vYrG + v

~ )G
- CVRG + VyG + vBG )I / 3

v,,

Substituting this value of

(6)

in equation (3), (4) and (5),

-y

the expression for the phase voltages are given hy

[=I]:![

-1

-I

2
-1

-1

-RG]

VYG

-1

(7)

-vrG

VBG

BE

Aochen, Gennony, 2004

and the corresponding d-q components are

(8)

(9)
There are two independent two-level inverters. The total
possible enerated voltage vectors given hy equation (9)
will he 2 X2 ( 4 4 ) . However, in actual practice there is
only one zero voltage vector, six small voltage vectors (S),
six medium voltage vectors (M) and six large voltage
vectors (L). The remaining vectors overlaps with these
voltage vectors. These vectors are shown in Fig.2. The six
small voltage vectors form the inner hexagon of Fig.3,
whose centre is 0 and they each lie at the vertices P, Q, R, S,
T and U. These small voltage vectors can he generated by
clamping INV2 to the zero voltage vector (VO or V7) and
switching INV1. The medium and large voltage vectors are
formed when both the inverters are in switching mode. For
example, for achieving voltage vector V14, INVl is
switched to position V1 and INV2 is switched to position
V4. The six large voltage vectors form the outer large
hexagon of Fig.3. The entire region of Fig.3 can he
subdivided into 24 equilateral triangles. They are known as
sub-sectors. These sub sectors form one inner hexagon
and six outer hexagons whose centres are 0, P, Q, R, S, T
and U. The position of the voltage space vector can he
determined by the method proposed in [ll].

INV.

(High Switching Frequency)

INDUCTION MOTOR

(Low Swilching Frequency)

Fig. 1 Schematic ofopen-end winding induction motor drive fed by 2 -WO level inverters
Y-Phase Axis
V36

V26V35

1i I

V25

Axis

VS2

V62VS3

t -

V63

B-Phase Axis
Fig.3 Partition of hexagon in to Sub-sectors

Fig.2 Generated Voltage-Vectors

121s

Anchen, Germany, 2004

2004 35rh Annual IEEE Power Elecrronics Specialisrs Conference

I11

PROPOSED SVM-DTC

Though the reference voltage vector is generated using the


method proposed in [IS], for the sake of convenience of the
reader, it is briefly described here. It is has been derived in
[I] that the rate of change of torque is proportional to the
slip speed. Using the same concept, the reference
synchronous speed for the reference stator flux is derived.
The actual flux vector is derived from the motor model. The
difference of these two-stator flux vectors divided by the
sampling time generates the reference voltage vector. Thus
in each sampling interval, the generated voltage vector will
compensate this error.
The reference voltage vector can be constructed by
the application of the two nearest voltage vectors and zero

Fig.4 Constnrction of reference voltage vector


voltage vectors for certain duration and in some particular
sequence. The duration of these voltage vectors depends on
the magnitude and the position of the reference voltage
vector. In order to reduce the overall switching frequency
without compromising on the current or torque ripple, the
following technique is used. Let us assume that the
reference voltage vector VOMis lying in sub-sector 24 as
shown in Fig.4. The sub-sector 24 is lying in the hexagon
whose center is P. Therefore this voltage vector can be
realized by applying vectors VOL.VOAand Vop i.e.V64, VI4
and VI (or V04) in some sequence for a pre-calculated time
known as
dwelling time. This dwelling time can be
determined by using the method suggested in [I21 for three
level inverter. It can be noted that if V64, VI4 and V04 are
chosen to realize the voltage vector VoM, INV2 can be
clamped at 4 and INVI can be switched according to the
conventional SVM. For calculating the dwelling time of
INVI, Vop is subtracted from the reference voltage vector
VOM.Thus the remaining voltage vector is either VpMor
VOW. Voltage vector VOM. lies in the inner hexagon.
Therefore it can be synthesized in the normal way of space
vector PWM technique. Hence if the voltage vector lies in
between -30 and +30, INV2 is clamped to V4 position.
This results in a significant reduction in the switching
frequency of INV2. INV2 may be continued to V4 for the
angle-30 to +30 if the reference voltage vector lies inside
the outer hexagon. The voltage vectors to which INV2 is to
be clamped, and the switching sequences of INVI for
various sub-sectors are given in TABLE-I. If the reference
voltage vector is lying in the inner hexagon, INV2 is either
clamped to position VO or V7. When the reference voltage

vector moves from the outer hexagon to the inner hexagon,


V0 or V7 is selected lor the least change in the switching
state. The block diagram of the proposed method is shown
in FigS.
In order to suppress the zero sequence currents,
three methods have bem reported in the literature:
( I ) Use of inter-phase reactors, either on the AC or DC
side of PWM inverter [I], 161-181.
(2) Rectifying the AC supply through transformers
thus providing isolation between the input DC
voltages to the inverters [ 5 ] .
(3) Use of auxiliary switches in order to provide
imaginary neutral [13].
The inter-phase reactox has some disadvantages like acoustic
noise and losses, resui.ts in a bulky system and it requires
more space. Though the isolating transformer seems to be a
better alternative, it also makes the system bulky and is
costly, it creates more acoustic noise and requires more
space. The method reported in 1131 uses auxiliary switches
to block the zero sequence currents. This method restricts
some of the switching sequences that cause triplen
harmonics. Due to thi!; reason, the dc bus voltage is under
utilized and hence the output of the multilevel inverter
decreases. In order to fully utilize the PWM inverter, the
voltage of the dc bus n(:eds to be increased.
In some applications like EV/HEV where motors
are driven by electric batteries, it may be convenient to
partition the batteries in order to isolate the dc sources and
mitigate the problem of zero sequence currents. But
partitioning of batterieii is not a wise decision [IO], as it will
require additional relays, fuses etc. This may not be true in
general, specifically for high power applications such as
submarines. Relays, fuses and other safety issues are much
cheaper and occupy less space as compared to other
methods of isolation. The soft-switched, high switching
frequency isolating transformer can also be used for
isolating dc sources that will eliminate many disadvantages
of the conventional methods.
IV

RESULTS

Simulation studies have been carried out using SIMULINK


software in MATLAB environment. Figs. 6(a) & (b) are the
results of the proposed method at a load torque of 6 N-m
applied at 0.5 sec and having a reference speed command of
600 rpm. Figures 7 (a) & (b) shows the results for a twolevel inverter under identical condition. In figures 8 and 9, a
change in reference fliur from 0.7 to 0.4Wb is applied at 0.5
sec. It is done at light load condition in order to improve the
efficiency and the drive now starts to operate in two-level
mode. In figures IO and I I , a change in reference speed
from 600rpm to -500 rpm is shown. Though the results of
the open-end winding configuration are superior to the
conventional SVPWM. method, the main advantage is the
reduction in switching frequency and hence the device
rating is reduced to half that makes it less expensive and
easily available.

1216

2004 35th Annual IEEE Power Electronics Specialists Conference

Aarhpn. Germany, 2004

Ill

Fig.5 Block Diagram of the proposed methad


decreases and one of the inverters goes into the off state
automatically, thereby improving its efficiency. Results also
show that the torque control using open-end winding
configuration generates three level line voltages that can
improve the performance of the drive. The proposed strategy
mentioned earlier may be suitable for battery operated
propulsion systems such as submarines. And by simply
partitioning the batteries, the problem of zero sequence
currents can be eliminated.

TABLE-I

REFERENCES
I Takahashi and T Noguchi. A New Quick-Response and HighEfficiency Conml of an Induction Motor, LEEE Trans. Industry
Applications, Vol. IA-22, No.5, , pp 820-827, 1986.
M Depenbroek, Direct Self Control (DSC) of Invener-Fed
Induction Machines, IEEE Trans. Power Electronics, Vol. PE -3,
N0.4, pp 420429, 1988.
I Takahashi and Youchi Ohmari, High -Performance Direct
Torque Control of an Induction Motor,. IEEE Tram. Industry
Applications, Vol. IA-25, No.2, pp 257-264, 1989.
H Stemmler and P. Guggenbach, %onfigurations of High -power
Voltage Source Inverter Drives, Proc. EPE93, Vol., pp 7-14,
1993.
T kawabata, New Open-Winding Configurations For High-Power
Inverters, Proc. ISIE97. Va1.2, pp 457 462, 1997.
Y Kawabata, M Nasu, T Nomoto, E C Ejiogu and T Kawabata,
High-Emciency and IOW Acoustic Noise Drive System Using
Oven-Windine AC Motor and Two Soace-Vector-Modulated
Invencrs, IEEE Trans. On Industrial Electronics, Vo1.49,Na.4, pp
783-789, Aug.2002.
M Tanpen and Andreas Steimel. Direct Self Control With
Minimum Toraue Rioole and Hieh h a m i c s for Double three level GTO I n v k e r Drive, E E E Yrans.. on Indushial Electronics,
Vo1.49, Na.5,pp 1065- 1071,Oct. 2002.
Xiao Q. Wu and Andreas Steimel, Direct Self Control of
Induction Machines Fed by a Double Three-Level Inverters,
IEEE Trans. cm Industrial Electronics. Vol. 44,Na.4, pp 519 -521,
Aug. 1997.
KA Corzine, SD Sudhoff, CAWhitcamb, Performance
Characteristics of a Cascaded Two-Level Converter, E E E Trans.
on Energy Conversion, Vo1.14, N0.3, pp 433 -439, Sept. 1999.
Brain A Welchka and James M Nagashima, A comparative
Evaluation of Motor Drive Topologies for Low-Voltage, HighPower EVIHEV Propulsion Systems, IEEE International
I

V CONCLUSION
A new control strategy is proposed for reducing the
switching frequency of one of the inverters. It clamps one of
them for a minimum period of one switching cycle. As
shown in the results, it reduces the ripples in torque and flux
Harmonic spectrum of the phase current has improved
significantly. Under light load condition, the reference flux

1217

Aachen, Germany, 2004

2004 3Srh Annunl IEEE Power Eiecrronics Specialisrs Conference

'
7

81

'

4,

er

a1

0 2

I 2

s1

a'

I'

,
L
-

<

DI

116

114

o,

I.."

/I

01

or

01

01

I-

Fig. 6 (a) Stator nux space Vector: X-Axis (D-Component),


Y-Axis (Q-Component) (Proposed)

Fig. 7 (a) Stator Fiux space Vector. X-AXIS(D-Component),


Y-Axis (Q-Comporient)(Two-Level SVM)

i
0.75 . . . . . . . . . . . . . . . . . . . . . . . . .
.........................

0.25 . . . . . . . . . . . . . . . . . . . . . . . . .

@6"

: :............

.........

.]

-1.

.........
. . f. . . .
.....

. . . . . . . . . . . . . . . .

.......................

C.15

ai5

a3

0.45

. . . . .

0.6

0.45

nme (sec)

0.6

me (sec)

Fig. 6 (b) Change in Load torque fmm 0 to 6 N-m at 0.5


at wnsmt stator flux 010.7 wb (Proposed)

0.3

Fig. 7 (b) Change in Load torque from 0 to 6 N - m at 0.5 Sec


Sec

at constant stator flux of0.7 Wb

(Two-Level SVM)

""11

.
:
I
"

aj

,
.I

,
a.

as

0,

02

"

"

OI

01

Or

'

111

.M

I**

Fig. 9 (a) Chsnge in stator flux vector from 0.7wb t o 0.4 at


0.5 Sec (Two-Level SVM)

Fig. 8 (a) Change in stator flux vector fmm 0.7W to 0.4 at


0.5 Sec (Pmpased)

1218

Aachen, Germany, 2004

2004 351h A n n u l IEEE Power Elecrronics Specialists Conference

"=.eI,

Tlme (sec)

Fig. 8 (b) Change in stator

flux vector from 0.7%

to 0.4 Wb

Fig 9 (b) Change in stator flux vector from 0.7 to 0.4 Wb at 0.5Sec

0.5 Sec (Proposed)

at

0.4

0.6
nrne (Sec)

Fig.10 Change in Speed from 600 Ipm to

1.2

Tlms IS64

Fig. I I Change in Speed from 600 'pm to -600 rpm

4 0 0 Ipm at no-load

condition (Proposed)

[Ill

[I21

[I31

[I41

[IS]

Symposium on Industrial Electronics, ISIE'03, Brazil, pp 1-6,


June9-12,2003.
EG Shivakumar. K Gopakumar, SK Sinha, VT Rangmathan.
"Space Vector PWM Control of Dual' Inverter Fed Open -End
Winding Induction Motor Drive," IEEE-APEC, V d l , pp 399-405,
2001.
JH Seo. CH Chai, DS Hyun. "A New Simplitled Space-Vector
PWM Method for Three-Level Inverters", IEEE-APEC, Vol.l, pp
515-520, 1999.
VT Somasekhar. K Gopakumar. A Pillet and VT Rangmathan,
'' PWM inverter switching strategy for a dual Two-level inverter
fed open-end winding induction motor drive with a switched
neutral", IEE hoe-Electr. Power Appl. Va1.149, No.2, pp 152 -160.
March 2002.
P Vas. Sensorless Vector and Direct Torque Control", Oxford
University Press, Inc., New York, 1998.
Arhind Kurnar, BG Fernandes, K Chatterjee. "Simplified Hybrid
SVM Based Direct Torque Control of Thee Phase Induction
Motor," National conference on CClS 2 W . Goa (India) Vol.l, pp
137-142.2004.
I'

1219

Vous aimerez peut-être aussi