Paper 21

You might also like

Download as pdf or txt
Download as pdf or txt
You are on page 1of 7

2019 10th International Power Electronics, Drive Systems and Technologies Conference (PEDSTC)

12-14 February, Shiraz University, Iran

A Novel Space Vector Modulation Strategy for Direct


Torque Control of Induction Motors
Mehrdad Mirzaei Ali Dastfan
Faculty of Electrical And Robotic Engineering Faculty of Electrical And Robotic Engineering
Shahrood University of Technology Shahrood University of Technology
Shahrood, Semnan Province, IRAN Shahrood, Semnan Province, IRAN
mirzaei.m@shahroodut.ac.ir dastfan@shahroodut.ac.ir

Abstract - A novel Direct Torque Control (DTC) scheme for efforts to reduce the flux and electromagnetic torque
an induction motor drive system is proposed in this paper. This fluctuation by applying fixed switching frequency methods.
method is based on direct load angle control to determine needed
stator voltage vector for compensation of electromagnetic torque In [5], the DTC scheme proposes a decoupling of the
and stator flux error by means of new modified Space Vector
Modulation (SVM) inverter controller. Objectives of motor drive
stator flux and the torque. The main issue here is to utilize
control, fixed switching frequency, low torque, and stator flux equation of torque as a function of the angular frequency of
ripple, consequently low current distortion and simplicity in slip and rotor in order to achieve the control of flux angle in
implementation are achieved using new modified SVM strategy. parallel with maintaining the magnitude of the reference of
Matlab / Simulink simulation verifies the capacity of the proposed stator flux constant. Using this structure, quick dynamic torque
motor drive to evaluate potentials such as improvement in torque response and low torque and flux ripple are achieved in the
and stator flux ripple, low stator current harmonic contents expense of losing the independence of the control scheme to
during full load and one step speed reference.
the motor parameter and high computational requirements. In
Index Terms - torque control, load angle control, modified [8], two PI controllers are used respectively to control torque
space vector modulation, low ripple. and flux simultaneously. As a result, produced reference
voltage vector in the rotary frame makes the scheme similar to
I. INTRODUCTION the field oriented control. An improved DDC (Discrete Duty
In the mid-80s, new strategies for controlling the torque of Cycle Control) method is used in [9], in addition, an effective
the induction motor are presented by Takahashi and Noguchi voltage vector selection is proposed and by inserting zero
under the title of Direct Torque Control (DTC) [1] and voltage vector along with active voltage vector, a finite set of
Depenbrock as Direct Self Control (DSC) [2]. The founders of new candidate voltage vectors with different duty cycle is
this new strategy believed that using hysteresis controller synthesized. Also, a finite-state model predictive torque
could avoid the existing complexity of the Field Oriented control is employed to select the optimal duty cycle in order to
Control (FOC) scheme, and achieved a less detailed and low reduce torque and flux ripple. In [10], the 5-phase induction
computational requirement structure, along with simplicity, motor is fed by a seven-level torque comparator and a three-
new techniques are able to improve the dynamic response of level5-phase inverter, and virtual voltage vectors are also
the induction motor. These two referred methods above, are formed to select the actual voltage vectors depending on the
the basic direct torque control approaches and have been location of the x-y stator flux, which in turn eliminates the x-y
developed and improved since 1985 by many researchers stator flux.
without interruption. The contribution in this paper is to utilize extremely basic
DTC and FOC methods are very popular as two high- one-step direct load angle control technique for induction
performance vector control scheme. The DTC has surpassed motor drive with DTC and a novel modified SVM got from the
its popularity due to its intrinsic advantages, simplicity, blend of space vector modulation and hysteresis control.
robustness, and high dynamic behavior in competing against This paper is arranged as follow, section II portrays three-
its long-standing rival, the FOC technique. phase induction motor dynamic equations in the two-phase
But unlike the special benefits of this approach, requires a stator reference frame and the principle of fundamental DTC
large sampling time and variable switching frequency and control scheme. Section III is dedicated to clarifying the
torque fluctuation, which results in production of harmonics in principle of basic DTC-SVM. The principle of one step load
the current, and as consequence, the temperature of the drive angle control of an induction motor and the proposed new
system and motor depreciation is boosted, and acoustic noise, modified SVM are presented in section IV. Comparative
especially at low speed operating conditions, will have a performance results using Matlab/Simulink simulation
significant amount. The variable switching frequency also between basic DTC, DTC-SVM and the proposed DTC-
raises the losses and heat in the drive. Therefore, in recent modified SVM are given in section V. In conclusion, the
years, numerous studies have been done to eliminate the merits obtained from simulation results are presented.
disadvantages of this control plan, which has focused on

978-1-5386-9254-7/19/$31.00 ©2019 IEEE

114
II. THE PRINCIPLE OF BASIC DTC The following relations are showing the flux hysteresis
controller that has got two dimensions of digitalized yield as:
A. Dynamic Model of Induction Motor
Hψ S = 1 for Eψ S > + HBψ S (6)
The dynamic equations of an induction motor can be
Hψ S = −1 for Eψ S < − HBψ S (7)
written in two-phase frame variables within the stationary
reference frame. The motor's stator voltage vector VS can be Likewise, a three-dimension hysteresis controller which has
expressed as follows: got three digital output is devoted to torque controller, as in
d (8)-(10).
VS = RS I S + ψS (1) HTe = 1 for ETe > + HBTe (8)
dt
d HTe = −1 for ETe < − HBTe (9)
0 = Rr I r + ψ r − jωmψ r (2)
dt HTe = 0 for − HBTe < ETe < + HBTe (10)
ψ S = LS I S + Lm I r , LS = Lls + Lm (3) The figuring of assessed estimation of stator flux and
ψ r = Lr I r + Lm I S , Lr = Llr + Lm (4) electromagnetic torque is performed by some manipulating of
Where Rs and Rr are the stator and rotor resistance, Lls, Llr, (1) and (2). Torque and flux estimator block also compute the
and Lm are the stator leakage, rotor leakage, and mutual sector number N in which flux vector lies. There are six π/3
inductances respectively and ωm is the rotor speed. radian sectors as indicated in Fig. 2. Suitable voltage vector
The electromagnetic torque equation of an induction (switching states) for the inverter is produced by the voltage
motor in terms of stator flux and rotor flux vector is written as vector table block in Fig. 1 utilizing a lookup table, which is
follow: appeared in table I.
3 L 3 L By refusing to consider stator resistance we can write:
Te = P m ψ S ×ψ r = P m ψ S ψ r sin(δ ) d
2 σ LS Lr 2 σ LS Lr VS ≈ ψS (11)
dt
L2
(5) Or
σ = 1− m
LS Lr Δψ S = VS * Δt (12)
Where P is the number of pole sets, δ is load angle and σ Plainly from (12), it is comprehended that stator flux can
is known as the leakage factor. be varied gradually by applying stator voltage vector for time
B. The principle of Basic DTC increase Δt.
High bandwidth torque control can likewise be gotten by The conduction time of each of the voltage vectors applied to
utilizing the immediate torque control (basic DTC) techniques the inverter is calculated by predetermined hysteresis bands.
as an option in contrast to the vector control plot. Fig. 1 shows So the smaller hysteresis caused to the higher switching
a basic DTC induction motor drive schematic, which uses two frequency.
hysteresis controller. In Fig. 1, induction motor provided V3 ( 010 ) V2 (110 )
through a VSI inverter and stator linkage flux and
electromagnetic torque are controlled straightforwardly and
autonomously by choice of inverter switching modes. The N3 N2
choice is made to confine the flux linkage and torque error
inside their hysteresis bands to get a quick torque reaction. The V7 (111)
V4 ( 011) N4 N1 V1 (100 )
output commands of the flux and torque hysteresis V0 ( 000 )

comparators (Hψs, HTe) are used to pick up inverter switching


N5 N6
states from the lookup table, which also uses information about
the sector N where sector stator flux space vector is located in.
Alluding to Fig. 1 estimated value of stator flux and torque
V5 ( 001) V6 (101)
amplitudes are subtracted from their reference comments, and
error is applied to hysteresis controllers to get the switching Fig. 2 Inverter voltage vectors
states of inverter prepared. TABLE I
SWITCHING TABLE OF INVERTER VOLTAGE VECTORS
Te ETe
ω r* Hψs HTe N1 N2 N3 N4 N5 N6

ωr Eψ S 1 1 V2 V3 V4 V5 V6 V1
*
ψS
Vdc 1 0 V0 V7 V0 V7 V0 V7
ψS θψ States
S
ψ S =  (VS − RS I S ) dt Ia 1 -1 V6 V1 V2 V3 V4 V5
3
Te = P (ψ S × I S )
2 -1 1 V3 V4 V5 V6 V1 V2
ψ 
θψ S = arctan  β S 
Ib
ψαS  -1 0 V7 V0 V7 V0 V7 V0
Fig. 1 Basic DTC control block diagram

115
-1 -1 V5 V6 V1 V2 V3 V4
β axis
Therefore, flux pulsations consequently torque pulsations are V2 (110 )
directly controlled and core losses and also noise may be
controlled through the stator flux level selection.
III. PRINCIPLE OF DTC-SVM
N1
Due to the fact that the three-phase inverter has six power T2
V2 VS*
switches, so can consider eight different switching states for TSW
three phase. Now, by drawing the voltage vectors α V1 (100 )

corresponding to each combination of switching states, the T1 α axis


V1
vectors plane is shaped like a hexagon as given in Fig. 3. And, TSW
also the on and off modes of lower power switches are exactly Fig. 4 Calculation of duration of applied voltage vector V1 and V2
opposite of the switching modes of upper ones.
Space vector modulation refers to a special switching TSW T1 T1 +T2 TSW

sequence of the upper three power switches of a three-phase  V *S dt =  V1dt +  V2 dt + 


T1 +T2
V0 dt (15)
power inverter. In the SVM algorithm, the goal is to determine 0 0 0

the duration of the conduction of each of the adjacent voltage Where


vector in every sector in order to generate a voltage at the Tsw = T1 + T2 + T0 (16)
inverter output, which has an average value equal to the mean T1 and T2 are used to determine the driving duration of each
value of the reference voltage vector at the input. The space two adjacent vectors in every sector. The reason for using zero
vector plane is separated into six areas by voltage vectors voltage vectors in switching pattern is only to maintain the
created utilizing a three-phase PWM inverter as appeared in balance in the on and off cycle of power switches and is also a
Fig. 3 [7]. useful factor in reducing switching power losses.
π π In sector N1, using (15) and based on Fig. 4, driving times
(i − 1) < Ni < i , i = 1,...,6 (13)
3 3 are calculated as follows:
By applying (14), the position of the reference voltage TSW V *S = (T1.V1 + T2 .V2 ) (17)
vector where it lies can be determined.  π 
cos( ) 
θV
V*
= arctan( S*α )  cos(α )  2 1  2  3 π
TSW *
V .  = T1 3 Vdc  0  + T1 3 Vdc   (0 ≤ α ≤ ) (18)
*
S
VS β (14) S
 sin(α )     sin( π )  3
The SVM technique determines the duration of  3 
conduction of each voltage vector applied to the inverter The expression of T1 and T2 are as follows:
through the projection of the V̅S* on two contiguous vectors in  π
 sin( − α )
each sector. The values of this projection determine the 3
T1 = TSW .a. π
desired switching times Tn and Tn+1 and correspond to two  sin( )
 3 (19)
none zero inverter switching states. To maintain a constant
 sin(α )
switching frequency, in the case of Tn + Tn + 1 < TSW, zero  T2 = TSW .a.
 π
inverter status is applied during a complementary duration of sin( )
 3
the TSW period, that is T0 = TSW – (Tn + Tn+1).
Where
Assuming that the reference voltage vector is located in
sector N1, the vector V̅S* can be calculated according to Fig. 4 V *S (20)
T0 = TSW − (T1 + T2 ), a=
by inserting V̅1 and V̅2 in equation (15). 2
Vdc
β axis 3
V3 ( 010 ) V2 (110 ) Consequently, the switching time for all sectors is given
as:
N2
3TSW V *S   π n − 1 
N3 VS*β * N1 T1 =  sin  − α + π 
V Vdc  3 3 
S
(21)
V4 ( 011) θV V1 (100 )
V7 (111) *
*
V0 ( 000 )
S
3TSW V  n n 
α axis S
sin( 3 π ) cos(α ) − cos( 3 π )sin(α ) 
VS*α =
N4 N6 Vdc
N5 3TSW V *S   n − 1 
T2 =  sin  α − π  (22)
V5 ( 001) V6 (101) Vdc   3 
3TSW V *S  n −1 n −1 
Fig. 3 Switching vectors and sectors =  sin(α ) cos( π ) − cos(α )sin( π )
Vdc  3 3 

116
 n = 1,...,6 (sec tor 1 − 6)  3 Lm
K= P ψS ψr
T0 = TSW − (T1 + T2 ) ,  π  (23) 2 σ LS Lr (25)
 0≤α ≤ 
 3  Therefore using (24) and output of torque comparator ΔTe , Δδ
In order to keep the balance in switching pattern, driving can be obtained as the following expression.
duration of each zero voltage vectors (V̅0, V̅7) is considered ΔTe ≈ K * Δδ (26)
equal, and for none zero vectors (V̅2, V̅1), each driving time Δδ ≈ ΔTe / K (27)
corresponds to the inverter voltage vectors is divided into two, From fig. 6, Δδ is the amount that load angle must be
because these vectors are applied twice to the inverter as it is increased to compensate the torque error. For calculating Δδ
demonstrated in Fig. 5. Referring to (21)-(23), conduction time by means of electromagnetic torque error, a PI controller has
of each adjacent vectors in every sectors are given in Table II. been designed and its optimal parameters are achieved using
According to Table II, conduction time of each upper try and error.
switches of the inverter are given and conduction time of lower Fig. 8 delineates the block diagram of the direct load
switches are toggle of upper ones. angle control of an induction motor utilizing fundamental
IV. DIRECT LOAD ANGLE CONTROL AND PROPOSED SVM space vector modulation. To meet torque reference tracking, a
PI torque controller incites over load angle δ. The block that
A. The principle of Load Angle Control computes the reference flux vector is given by,
Based on (5), clearly, it can be seen that the flux vector of ψ S * = ψ *S * cos(Δδ + θ S ) + j ψ *S * sin(Δδ + θ S ) (28)
the stator and the rotor are the determinant parameters in
calculating electromagnetic torque. Therefore, simply As shown in Fig. 8, the reference flux signal produced in
assuming that the amplitude of stator and rotor flux vectors are accordance with equation (28) is compared with its
constant, it is highly possible to achieve to the control of the corresponding estimated value, and the result ΔψS is divided
torque and speed of an induction motor by adjusting the load into Δt. By adding a stator resistance drop, the vector of the
angle δ. reference voltage (as in (29)) needed to track the reference
In this method, the objective is looking for a voltage torque can be taken. The SVM block generates appropriate
vector that acts on motor directly to affect the stator flux. This switching states for an inverter with the supply of V̅S*.
allows adjusting the load angle to provide the required torque. ψ * −ψ S
VS = S + RS I S
Meanwhile, SVM technique has been used to apply the voltage Δt (29)
vector set to the inverter. B. Proposed Modified SVM
From (5), δ, load angle is the angle among stator and rotor Generally, the basis of the proposed method is as similar
flux vectors and (5) can be approximated by (24) as follows: as basic SVM. The only difference between the proposed and
Te ≈ K * δ (24) basic SVM is in how to involve with the determination of
conduction time of power switches. In the proposed algorithm,
Where using (29) the stator reference voltage vector that needs to
V0 ( 000 ) V1 (100 ) V2 (110 ) V7 (111) V2 (110 ) V1 (100 ) V0 ( 000 )
compensate the torque and flux error, is calculated and applied
to the proposed modified SVM as an input. In the proposed
Sa algorithm the calculation of commutation times of switches T1
and T2 is not required. The modified SVM block uses
Sb calculated V̅S* by (29) and a hysteresis band to determine the
switching state of ithe nverter. As appeared in Fig. 7, space
Sc vector plane is divided into six equal π/3 radian sectors which
sector N1 is located in the range of -π/6 to π/6 radian. Thus
voltage vectors compatible with each switching states are
T0
located in the sector with its corresponding number.
T1 T2 T0 T2 T1 T0
Internal circle with radius Vh, shown in Fig. 7 is known as
4 2 2 2 2 2 4
hysteresis band. This hysteresis band is determined in terms of
Fig. 5 Sequence of the switching states in sector N1
β axis ψ *S
TABLE II Δψ S ∝ VS
CONDUCTION TIME OF UPPER SWITCHES OF INVERTER
) S
ψ S sin ( Δδ + θψ

Sectors S1 S2 S3 ψS
N1 T1+T2+T0/2 T2+T0/2 T0/2
Δδ
*

N2 T2+T0/2 T1+T2+T0/2 T0/2


δ
N3 T0/2 T1+T2+T0/2 T2+T0/2 θψ S ψr
N4 T0/2 T2+T0/2 T1+T2+T0/2
N5 T2+T0/2 T0/2 T1+T2+T0/2 ψ S cos ( Δδ + θψ ) α axis
*
S

N6 T1+T2+T0/2 T0/2 T2+T0/2 Fig. 6 Flux vectors diagram

117
the magnitude of DC bus voltage of the inverter and used only V. SIMULATION RESULTS
to investigate the amplitude condition of applied voltage
In order to test torque, dynamic response of the proposed
vector and also does not affect the switching frequency. In this
method a closed loop performance analysis is performed and
algorithm, toward the start of the sampling time frame
contrasted with basic DTC and DTC-SVM scheme from point
sufficiency state of V̅S* is performed. If applied V̅S* is lower
of view of speed reference tracking, line current FFT,
than Vh, zero voltage vector is applied to the inverter,
electromagnetic torque and stator flux fluctuation.
otherwise, depending on the sector number where the V̅S* is
For this simulation, an induction motor whose
located, the voltage vector with the same number is selected.
specifications and parameters are described in detail in Table3
Selected voltage vector in either of the two mentioned modes
is used. For this test, the assumptions like sampling frequency
are valid for the whole current sampling period and until the
of 100 KHz, dc link voltage of inverter equal to 600 v and also
beginning of the next sampling period, switching states will
a step input for speed reference, which at time of 200 ms is
not change. Despite using hysteresis band in the proposed
ωm=150 rad/s, are considered identical for the three models.
method, the switching frequency remains constant as in basic
In Fig. 9, three of main and measurable results of an
SVM and it is apart from motor operating condition. The
induction motor (speed, torque response, and flux amplitude)
algorithm of selecting appropriate voltage vector is given as
from each of the three models are shown. The upper part of
the following expression:
each figure is dedicated to the speed tracking reference, in the
if VS* > Vh then state = VN
 middle torque response is given and in the end section, the flux
VS* in sec tor ( N ):  (30)
*
 if VS < Vh then state = V0 domain is shown.
In view of Fig. 9, it is comprehended that basic DTC and
Therefore, using the proposed SVM technique, the torque
the proposed strategy have comparative powerful reaction,
fluctuation is anticipated to be almost diminished, while using
great speed reference following while at the same time keeping
a fixed switching frequency technique the error is nearly zero.
steady stator flux sufficiency. Flux and torque ripple of three
analyzed technique as shown in the third graph of Fig. 9 and
V3 ( 010 ) V2 (110 ) Fig. 10 separately, from high to low is altogether as
fundamental DTC, DTC-SVM and the proposed DTC-
modified SVM.
N3 N2 TABLE III
INDUCTION MOTOR PARAMETERS
Vh
V7 (111) 3-phase 4KW Llr = 0.005839 H
V4 ( 011) N4 N1 V1 (100 )
V0 ( 000 ) 50Hz 4-poles RS = 1.405 Ω
400V Lm= 0.1722 H Rr = 1.395 Ω
N5 N6
ωr = 150 rad/s Lls = 0.005839 H J = 0.0131 Kg.m2
In order to compare the speed of the dynamic response
V5 ( 001) V6 (101)
among these three techniques, once the speed of the rotor
reaches the reference value, one step load torque with the
Fig. 7 Proposed voltage vectors plane and a hysteresis band property of 12 Nm at time of 600 ms is applied to the motor.
The results are given in Fig. 10.

ω r* ψ *S 1 V *S
ωr Te Δt

ψS
*
ψS RS I S
Vdc
θψ S
States

ψ S =  (VS − RS I S ) dt
3
Te = P (ψ S × I S ) Ia
2
ψ  Ib
θψ = arctan  β S 
ψα S 
S

Fig. 8 Control diagram of DTC-SVM motor drive

118
( a)

( a)

( b)

( b) ( c)
Fig. 10 Torque response comparison. (a) Basic DTC, (b) DTC-SVM, (c)
DTC-modified SVM
A noteworthy capability of the proposed DTC-modified
SVM over other two strategies is that the lower computational
power it needs, the higher switching frequency can be utilized.
Accordingly, this legitimacy prompts low torque and stator
flux ripple. So the lower ripple in torque and stator flux of
induction motor, the lower THD is caused in line current as
shown in Fig. 11. Low scope of harmonics delivered by DTC-
modified SVM as for two other analyzed scheme can produce
fewer unpleasant acoustic noise.
VI. CONCLUSION
Although the proposed method does not provide
significant improvement over the SVM method in terms of
torque and flux ripple reduction, application of the proposed
method has advantages such as minimal dependence on motor
( c)
Fig. 9 Comparison of dynamin response: (a) Basic DTC, (b) DTC-SVM, (c) parameters, very simple structure, ease of implementation, and
DTC-modified SVM also the most important advantage is the low computational
Fig. 10 shows that by employing the proposed SVM in the requirement which allows maximizing the switching
DTC drive, the rate of the torque tracking response is well frequency and can be easily changed to sensorless drive
suited to the DTC-SVM method and has improved system. In addition, this method can solve known defects,
significantly compared to the basic DTC. high torque ripple, and variable switching frequency.

119
( a) ( a)

( b) ( b)
Flux Space Vector Trajectory

0.5

-0.5

stator flux rotor flux


-1
-1 -0.5 0 0.5 1
Flux [Wb]
( c) ( d)
Fig. 11 Line current FFT comparison. (a) Basic DTC, (b) DTC-SVM, (c) Fig. 12 Stator and rotor Flux vector trajectory. (a) Basic DTC, (b) DTC-SVM,
DTC-modified SVM (c) DTC-modified SVM
REFERENCES [6] Jun Zhang and M. Rahman, "Direct flux vector control scheme for
[1] Takahashi and T. Noguchi, "A New Quick-Response and High- induction machine drives with space vector modulation", Fourtieth IAS
Efficiency Control Strategy of an Induction Motor", IEEE Transactions Annual Meeting. Conference Record of the 2005 Industry Applications
on Industry Applications, vol. -22, no. 5, pp. 820-827, 1986. Conference, 2005.
[2] M. Depenbrock, "Direct self-control (DSC) of inverter-fed induction [7] A. Ouarda and F. Ben Salem, "Induction machine DTC-SVM: A
machine", IEEE Transactions on Power Electronics, vol. 3, no. 4, pp. comparison between two approaches", 10th International Multi-
420-429, 1988. Conferences on Systems, Signals & Devices 2013 (SSD13), 2013.
[3] F. Ben Salem, A. Yangui and A. Masmoudi, "On the reduction of the [8] Yen-Shin Lai and Jian-Ho Chen, "A new approach to direct torque
commutation frequency in DTC: a comparative study", European control of induction motor drives for constant inverter switching
Transactions on Electrical Power, vol. 15, no. 6, pp. 571-584, 2005. frequency and torque ripple reduction", IEEE Transactions on Energy
[4] J. Rodríguez, J. Pontt, C. Silva, R. Huerta and H. Miranda, "Simple Conversion, vol. 16, no. 3, pp. 220-227, 2001.
direct torque control of induction machine using space vector [9] M. Nikzad, B. Asaei and S. Ahmadi, "Discrete Duty-Cycle-Control
modulation", Electronics Letters, vol. 40, no. 7, p. 412, 2004. Method for Direct Torque Control of Induction Motor Drives With
[5] Y. Kumsuwan, W. Srirattanawichaikul, and S. Premrudeepreechacharn, Model Predictive Solution", IEEE Transactions on Power Electronics,
"Reduction of Torque Ripple in Direct TorqueControl for Induction vol. 33, no. 3, pp. 2317-2329, 2018.
Motor Drives Using Decoupled Amplitude and Angle of stator flux [10] Y. Tatte and M. Aware, "Torque Ripple and Harmonic Current
Control", ECTI TRANSACTIONS ON ELECTRICAL ENG., Reduction in a Three-Level Inverter-Fed Direct-Torque-Controlled Five-
ELECTRONICS, AND COMMUNICATIONS, vol. 8, no. 2, pp. 187-196, Phase Induction Motor", IEEE Transactions on Industrial Electronics,
2010. vol. 64, no. 7, pp. 5265-5275, 2017.

120

You might also like