Direct Torque Control using Space Vector Modulation and dynamic

performance of the drive, via a Fuzzy Logic controller for speed regulation

Z. Koutsogiannis, G. Adamidis, and A. Fyntanakis

DEMOCRITUS UNIVERSITY OF THRACE

12 Basil. Sofias, GR 67100

Xanthi, Greece
Tel.: +30 / (25410) – 79509.
Fax: +30 / (25410) – 79515.
E-Mail: [email protected], [email protected], [email protected]
URL: http://www.ee.duth.gr

Keywords
«Direct Torque Control», «Space Vector Modulation», «Fuzzy Logic», «Speed Controller».

Abstract
This paper presents a method of controlling the Torque of an asynchronous motor that is fed by a
PWM-Inverter. The control method that is used is a modified Direct Torque Control (DTC) scheme
with constant switching frequency, using Space Vector Modulation (SVM). In this scheme a closed
loop of torque control is applied. Small and rapid changes on motor’s torque are achieved by applying
an appropriate voltage vector, which is responsible for the jerky movement of the stator’s flux vector.
The voltage vector can be of any magnitude and angle and is produced using Space Vector Pulse
Width Modulation. The modified method of DTC using SVM improves the electrical magnitudes of
asynchronous machine, such as reduction of current and flux ripple and also fading out the edges of
torque ripple which are occurred during the sector changes of stator flux vector. Another point of this
paper is the presentation of speed control in dynamic situations. The production of the reference torque
value is achieved by using a fuzzy logic speed controller. The response of the fuzzy PI is compared to
the response of a classic PI speed controller. Results, shows that the fuzzy PI speed controller has a
better response in a wide range area of motor speed. Combining the characteristics of DTC-SVM
scheme and the fuzzy logic speed controller we can have a high performance vector controlled
asynchronous motor drive.

Introduction
During the last decade, a lot of modifications in classic Direct Torque Control scheme [1] have been
made [6], [9], [10], [12], [13], [14]. The objective of these modifications was to improve the start up of
the motor, the operation in overload conditions and low speed region. The modifications also aimed to
reduce the torque and current ripple, the noise level and to avoid the variable switching frequency by
using switching methods with constant switching frequency.

The basic disadvantages of DTC scheme using hysteresis controllers are the variable switching
frequency, the current and torque ripple. The movement of stator flux vector during the changes of
cyclic sectors is responsible for creating notable edge oscillations of electromagnetic torque. Another
great issue is the implementation of hysteresis controllers which requires a high sampling frequency.
When an hysteresis controller is implemented using a digital signal processor (DSP) its operation is
quite different to the analogue one. In the analogue operation the value of the electromagnetic torque
and the magnitude of the stator flux are limited in the exact desirable hysteresis band. That means, the
inverter can change state each time the torque or the flux magnitude are throwing the specified limits.
On the other way, the digital implementation uses specific sample time on which the magnitudes of
torque and flux are checked to be in the desirable limits. That means, very often, torque and flux can
be out of the desirable limits until the next sampling period. For this reason, an undesirable torque and
flux ripple is occurred.

Many researchers are oriented to combine the principles of DTC with a constant switching frequency
method for driving the inverter by using space vector modulation. That means, in such a modified
DTC scheme the main purpose would be to estimate a reference voltage vector, in order to control the
inverter. Before describing the modified DTC scheme it would be wise to take a look at the classic
DTC scheme.

Overview of the classic DTC scheme

The classic DTC scheme is shown in figure 1.

Fig. 1: Classic DTC scheme

DTC based drives require only the knowledge of the stator resistance Rs. Measuring the stator voltage
and current, stator flux vector can be estimated by the following equation:
G G G
ψ s = ∫ (Vs − Rs I s ) dt
(1)

the stator flux magnitude is given by,

G
Ψ s = Ψ 2as + Ψ 2β s
(2)

where the indicators α,β indicates the α-β stationary reference frame. The stator flux angle is given by,

Ψ
θ e = sin −1 G β s
Ψs
(3)

and the electromagnetic torque Te is calculated by,

3P
Te =   ( Ψ α s iβ s − Ψ β s iα s )
2 2  (4)

where P is the number of machine poles.

In the DTC scheme the electromagnetic torque and stator flux error signals are delivered to two
hysteresis controllers as shown in figure 1. The stator flux controller imposes the time duration of the
active voltage vectors, which move the stator flux along the reference trajectory, and the torque
controller determinates the time duration of the zero voltage vectors, which keep the motor torque in
the defined-by-hysteresis tolerance band. The corresponding output variables HTe, HΨ and the stator
flux position sector θΨs are used to select the appropriate voltage vector from a switching table [1],
which generates pulses to control the power switches in the inverter. At every sampling time the
voltage vector selection block chooses the inverter switching state, which reduces the instantaneous
flux and torque errors.

DTC-SVM with closed-loop torque control

In this section, the DTC-SVM scheme will be presented which uses a closed loop torque control. The
block diagram of this scheme is shown in figure 2.

Fig. 2: DTC-SVM with closed-loop torque control

The objective of the DTC-SVM scheme, and the main difference between the classic DTC, is to
estimate a reference stator voltage vector V*S in order to drive the power gates of the inerter with a
constant switching frequency. Although, the basic principle of the DTC is that the electromagnetic
torque of the motor can be adjusted by controlling the angle δΨ between the stator and rotor magnetic
flux vectors. Generally, the torque of an asynchronous motor can be calculated by the following
equation.

3 P  L
Te =   m ' Ψ r Ψ s sin δψ (5)
2  2  Lr Ls

Where L's = Ls Lr − L2m . The change in torque can be given by the following formula,

3 P L G G
∆Te =   m ' Ψ r Ψ s + ∆Ψ s sin ∆δψ (6)
2  2  Lr Ls

where the change in the stator flux vector, if we neglect the voltage drop in the stator resistance, can
be given by the following equation,
G G
∆Ψ s = Vs ∆t
(7)

where ∆t=Ts, is the sampling period.

 (a) (b)

Fig. 3: Principle of Space Vector Modulation (SVPWM)

(a) reference stator vector
(b) modulation of space vector during one switching period which is equal to sampling time of the
DTC-SVM method.

Generally, the classic DTC employs a specific switching pattern by using a standard switching table.
That means the changes in the stator flux vector and consequently the changes in torque would be
quite standard because of the discrete states of the inverter. That is happens because the inverter
produces standard voltage vectors.

The objective of the DTC-SVM scheme, and the main difference between the classic DTC, is to
estimate a reference stator voltage vector V*S and modulate it by SVM technique, in order to drive the
power gates of the inerter with a constant switching frequency. Now, in every sampling time, inverter
can produce a voltage vector of any direction and magnitude. That means the changes in stator flux
would be of any direction and magnitude and consequently the changes in torque would be smoother.

According to above observations, and bearing in mind figure 2, we can see that torque controller
produces a desirable change in angle ∆δΨ between stator and rotor flux vectors. The change in angle
∆δΨ is added in the actual angle of stator flux vector, so we can estimate the reference stator flux
vector by using the following formula, in stationary reference frame.

G G j ω t +∆δψ )
ψ s* = ψ s* e (e

(8)

Applying a phasor abstraction between the reference and the actual stator flux vector we can estimate
the desirable change in stator flux ∆ΨS. Having the desirable change in stator flux, it is easy to estimate
the reference stator voltage vector:
G
G * ∆Ψ s G
Vs = + Rs I s
TS (9)

If the reference stator voltage vector is available, it is easy to drive the inverter by using the SV-PWM
technique. So, it is possible to produce any stator voltage space vector (figure 3). As it mentioned
G
before, in the classic DTC scheme, the direction of stator flux vector changes ∆ψ S are discrete and
are almost in the same direction with the discrete state vectors of the inverter. Consequently, in DTC-
G G
SVM, stator flux vector changes ∆ψ S can be of any direction, which means the oscillations of ψ S
would be more smoother.
Speed regulation using a fuzzy logic controller
So far, two methods were described for controlling the electromagnetic torque of an asynchronous
motor drive. When we need to regulate the speed of such a drive a speed controller is needed. The
speed controller takes the error signal between the reference and the actual speed and produces the
appropriate reference torque value. That means, the drive changes mode from torque control to speed
control. So, now the mechanical load on motor shaft defines the electromagnetic torque of the motor.
In torque control mode the mechanical load on motor shaft defines the rotor speed.

Fig. 4: Speed regulation using a speed controller

In figure 4 we can see the block diagram of the proposed drive, in speed control mode. A reference
speed signal ω*r is given or in other words the speed command. The actual speed ωr is estimated or is
measured with a speed encoder. This is depended on precision requirement of each application. The
speed is estimated directly from state equations. The dynamic α-β frame state equations of a machine
can be manipulated to compute speed signal directly [2], [4]. Consequently, the speed computation is
given by,

1  d d  L 
ωr =   Ψ α r Ψ β r − Ψ β r Ψα r  − m ( Ψα r iβ s − Ψ β r ias )
ˆ 
Ψ 2
dt dt  Tr 
r (8)

where Tr = Lr/Rr

This method of speed computation requires the knowledge of the machine parameters Lr, Lm and Rr
which are the rotor inductance, the magnetizing inductance and rotor resistance respectively.

The speed controller can be a classic PI controller or a fuzzy PI controller. In [8], a detailed
presentation and comparison of the two controllers is presented and operates with a classic DTC drive.
In this paper the fuzzy PI controller is also used for the comparison between the classic DTC and
DTC-SVM.
Simulation results
The DTC schemes, that are presented so far, are designed and simulated using Matlab/Simulink
(figure 5). The proposed scheme is simulated and compared to the classic one. The dynamic and also
the steady state behavior is examined in a wide range of motor speed and operating points.

(a)

(b)

Fig. 5: Simulink models of (a) classic DTC and (b) DTC-SVM

In figure 6 an analytical comparison of the dynamic performance of the control system is presented. It
can be seen the system behavior for increasing the motor speed while the load torque in motor’s shaft
remains constant at 50% of the nominal load. In more detail, it is presented the dynamic performance
of the two speed controllers, classic PI and fuzzy PI, for increasing the motor speed of 30%, 20% and
10% step commands of the nominal speed. In this figure, it can also be seen the improvement in
motor’s acceleration and the change in motor’s torque using the fuzzy PI controller. Classic PI
controller shows an undesirable overshoot of the actual speed. On the other hand fuzzy PI controller
has a smoother response. The output of each controller is the value of the reference electromagnetic
torque Te*. The change in motor’s speed is the result of applying the produced reference torque to the
DTC scheme.
The motor operation in steady state, it can be seen in figures 7 and 8. In figure 7 the motor is running
in low speed region and the load torque is also in low level. DTC and DTC-SVM have almost the
same switching frequency at this operating point. The ripple in stator current is very low in DTC-SVM
scheme compared to the classic DTC where the current ripple is on high level. Figure 8 shows the
motor operation in normal mode. The switching frequency is also at the same value in order to
 Classic PI Fuzzy PI

2 2

1.5 1.5 Te

Torque (pu)

Torque (pu)
Te*
1 1 TL
0.5 0.5

0 0
0.3 0.35 0.4 0.45 0.5 0.55 0.6 0.3 0.35 0.4 0.45 0.5 0.55 0.6

1.1 1.1
Rotor Speed (pu)

Rotor Speed (pu)

1 1
ωr*
0.9 0.9 ωr
0.8 0.8

0.7 0.7
0.3 0.35 0.4 0.45 0.5 0.55 0.6 0.3 0.35 0.4 0.45 0.5 0.55 0.6
Time (sec) Time (sec)

(a1) (a2)
2 2

1.5 1.5
Torque (pu)

Torque (pu)
Te
1 1 Te*
TL
0.5 0.5

0 0
0.3 0.35 0.4 0.45 0.5 0.55 0.6 0.3 0.35 0.4 0.45 0.5 0.55 0.6

1.1 1.1
Rotor Speed (pu)

Rotor Speed (pu)

1 1

0.9 0.9 ωr*

ωr
0.8 0.8

0.7 0.7
0.3 0.35 0.4 0.45 0.5 0.55 0.6 0.3 0.35 0.4 0.45 0.5 0.55 0.6
Time (sec) Time (sec)

(b1) (b2)
2 2

1.5 1.5
Torque (pu)

Torque (pu)

Te
1 1 Te*
TL
0.5 0.5

0 0
0.3 0.35 0.4 0.45 0.5 0.55 0.6 0.3 0.35 0.4 0.45 0.5 0.55 0.6

1.1 1.1
Rotor Speed (pu)

Rotor Speed (pu)

1 1

0.9 0.9 ωr*

ωr
0.8 0.8

0.7 0.7
0.3 0.35 0.4 0.45 0.5 0.55 0.6 0.3 0.35 0.4 0.45 0.5 0.55 0.6
Time (sec) Time (sec)

(c1) (c2)

Fig. 6: Simulation results of the speed controller response in various speed step commands.(1) Classic
PI controller, (2) Fuzzy PI controller.
(a) 30% , (b) 20% and (c) 10% of the nominal speed step command

have a right comparison. Current ripple has also a notable reduction in DTC-SVM compared to classic
DTC. Also, at this operating point it can be seen that in classic DTC the torque ripple of the
electromagnetic torque which is resulted by the cyclic sector changes of stator flux vector and
produces sharp edges, is now eliminated by using DTC-SVM.
 Classic DTC DTC-SVM

(a) (b)

Fig. 7: Steady state of the motor in an operation point where the motor has the 10% of the nominal
speed and 10% of nominal load.
G
(a) Classic DTC with HBψ = ± 0.015( 2% Ψ Sn ) , HBTe = ±0.65(  3%Ten ) for the hysteresis
band controllers and TS _ DTC = 12 µ sec the sampling time for discrete implementation.
Inverter produces 16780 states/sec.
(b) DTC with space vector modulation. Switching frequency is equal to 2.5kH and inverter
produces 15000 states/sec.
 Classic DTC DTC-SVM

Ψs vector α - β axis (pu) Vs vector α - β axis (pu) Ψs vector α - β axis (pu) V*ref vector α - β axis (pu)

1 1 1 1

0.5 0.5 0.5 0.5

0 0 0 0
-0.5 -0.5 -0.5 -0.5
-1 -1 -1 -1
-1 0 1 -1 0 1 -1 0 1 -1 0 1
Ψr vector α - β axis (pu) Is vector α - β axis (pu) Ψr vector α - β axis (pu) Is vector α - β axis (pu)

1 1 1 1

0.5 0.5 0.5 0.5

0 0 0 0

-0.5 -0.5 -0.5 -0.5

-1 -1 -1 -1
-1 0 1 -1 0 1 -1 0 1 -1 0 1

1.1 1.1
Torque (pu)

Torque (pu)

1 1

0.9 0.9

0.8 0.8
0 5 10 15 20 0 5 10 15 20
Time (sec) -3 Time (sec) -3
x 10 x 10
1 1
Current ia (pu)

Current ia (pu)

0.5 0.5

0 0

-0.5 -0.5
-1 -1

0 5 10 15 20 0 5 10 15 20
Time (sec) -3 Time (sec) -3
x 10 x 10
1 1

0.5 0.5
Vab (pu)

Vab (pu)

0 0

-0.5 -0.5

-1 -1
0 5 10 15 20 0 5 10 15 20
Time (sec) -3 Time (sec) -3
x 10 x 10

(a) (b)

Fig. 8: Steady state of the motor in normal operation point where the motor has the 100% of the
nominal speed and 100% of nominal load.
G
(a) Classic DTC with HBψ = ± 0.015( 2% Ψ Sn ) , HBTe = ±0.65(  3%Ten ) for the hysteresis
band controllers and TS _ DTC = 12 µ sec the sampling time for discrete implementation.
Inverter produces 14950 states/sec.
(b) DTC with space vector modulation. Switching frequency is equal to 2.5kH and inverter
produces 15000 states/sec.
Conclusion
This paper has presented a modified Direct Torque Control method for PWM-Inverter fed
asynchronous motor drive using constant switching frequency. Constant-switching-frequency is
achieved by using space vector modulation and finally an SVM based DTC system is compared to the
classic DTC scheme for torque control. DTC-SVM schemes improves considerably the drive
performance in terms of reducing torque and flux pulsations, reliable startup and low-speed operation,
well-defined harmonic spectrum, and radiated noise. Therefore, DTC-SVM is an excellent solution for
general-purpose asynchronous motor drives. Instead, the short sampling time required by the classic
DTC schemes makes them suited to very fast torque- and flux-controlled drives because of the
simplicity of the control algorithm. When a speed control mode is needed instead of torque control, a
speed controller is necessary for producing the reference electromagnetic torque value. For this
purpose a fuzzy logic based speed controller is hired. Fuzzy PI speed controller has a more robust
response, compared to the classic PI controller, in a wide range area of motor speed.

References
[1] I. Takahashi and T. Noguchi, “A new quick-response and high efficiency control strategy of an induction
machine,” IEEE Trans. Ind. Applicat., vol. 22, pp. 820–827, Sep./Oct. 1986.
[2] Bimal K.Bose, Modern Power Electronics And AC Drives. Prentice Hall 2002.
[3] Andrzej M. Trzynadlowsky, Control of Induction Motors. Academic Press 2001, pp. 137-157.
[4] I.Boldea, S.A.Nasar. “Electric Drives”, CRC Press, 1998.
[5] Domenico Casadei, Giovanni Serra and Angelo Tani, “FOC and DTC: Two viable schemes for induction
motors torque control”, IEEE Trans. Power Electron., vol. 17, pp. 779–787, Sept. 2002.
[6] Domenico Casadei, Giovanni Serra and Angelo Tani, “Implementation of a Direct Torque Control Algorithm
for Induction Motors Based on Discrete Space Vector Modulation”, IEEE Trans. Ind. Applicat., vol. 15, No.4 pp.
769–777, July 2000.
[7] Giuseppe S. Buja, Marian P. Kazmierkowski, “Diret Torque Control of PWM Inverter-Fed AC Motors – A
survey,” IEEE Trans. Ind. Applicat., vol. 51, pp. 744–757, August 2004.
[8] Z. Koutsogiannis, G. Adamidis, and A. Fyntanakis, “Computer Analysis of a Direct Torque Control
Induction Motor Drive Using a Fuzzy Logic Speed Controller” XVII International Conference on Electrical
Machines, September 2006.
[9] T. Brahmananda Reddy, B. Kalyan Reddy, J. Amarnath, D. Subba Rayudu, and Md. Haseeb Khan,
“Sensorless Direct Torque Control of Induction Motor based on Hybrid Space Vector Pulsewidth Modulation to
Reduce Ripples and Switching Losses – A Variable Structure Controller Approach”, IEEE Power India
Conference, 10-12 April 2006.
[10] Lin Chen, Kang-Ling Fang, Zi-Fan Hu “A scheme of fuzzy direct torque control for induction machine”,
IEEE Proceedings of the Fourth International Conference on Machine Learning and Cybernetics, Guangzhou,
18-21 August 2005.
[11] E.D. Mitronikas, A.N. Safacas, “A Hybrid Sensorless Stator-Flux Oriented Control Method for Induction
Motor Drives”, in 35th Annual IEEE Power Electronics Specialists Conference (PESC’04), June 20-25, 2004,
Aachen, Germany, pp. 3481-3485.
[12] P. Z. Grabowski, B. K. Bose and F. Blaabjerg, “A Simple Direct Torque Neuro Fuzzy Control of PWM
Inverter Fed Induction Motor Drive” IEEE Trans. Ind. Electron., Vol. 47, No. 4, pp. 863-870, August 2000.
[13] L. Romeral, A. Arias, E. Aldabas, M. G. Jayne, Tani, “Novel Direct Torque Control (DTC) Scheme With
Fuzzy Adaptive Torque-Ripple Reduction,” IEEE Trans. Ind. Electron., vol.50, pp.487–492,Jun. 2003.
[14] Marcel Ortega, Jos´e Restrepo, Julio Viola, Mar´ıa I. Gim´enez, Victor Guzm´an, “Direct Torque Control of
Induction Motors using Fuzzy Logic with current limitation”, IEEE Industrial Electronics Society, IECON 2005.
32nd Annual Conference.
[15] Epaminondas D. Mitronikas, Athanasios Safacas, Member, IEEE, and Emmanuel Tatakis, “A New Stator
Resistance Tuning Method for Stator-Flux-Oriented Vector-Controlled Induction Motor Drive”, IEEE
Transaction on Industrial Electronics, Vol. 48, No. 6, December 2001, pp. 1148 – 1157.
[16] Epaminondas D. Mitronikas, Athanasios Safacas, Member, IEEE, “An improved Sensorless Vector Control
Method for an Induction Motor Drive”, IEEE Transaction on Industrial Electronics, Vol. 52, Νο. 6, December
2005, pp. 1660-1668.