You are on page 1of 9

Electrical Engineering (2003) 86: 25–33

DOI 10.1007/s00202-003-0179-1

O R I GI N A L P A P E R

H. Altun Æ S. Sünter

Matrix converter induction motor drive: modeling, simulation


and control

Received: 28 April 2003 / Accepted: 17 May 2003 / Published online: 28 June 2003
 Springer-Verlag 2003

Abstract This paper presents a model in Simulink, which connected to any of the three output lines, as shown in
consists of a three-phase matrix converter, an induction Fig. 1, where uppercase and lowercase letters are used to
motor, a field-oriented controller, and a power supply. A denote the input and output, respectively. The switches
simplified Venturini’s modulation algorithm is used in are then controlled in such a way that the average output
the simulation model. This algorithm provides unity voltages are a three-phase set of sinusoids of the re-
fundamental displacement factor at the input regardless quired frequency and magnitude [2].
of the load displacement factor and can be easily The matrix converter can comply with four quadrants
implemented in closed loop operation. Simulation re- of motor operations, while generating no higher har-
sults are presented for both input and output sides of the monics in the three-phase a.c. power supply. Compared
converter. These results demonstrate a high-perfor- with conventional drives, there is potential for reduced
mance matrix converter-fed induction motor drive with cost of manufacture and maintenance, and increased
unity input displacement factor. power/weight and power/volume ratios. The circuit is
inherently capable of bi-directional power flow and also
Keywords Matrix converter Æ Induction motor Æ Field- offers virtually sinusoidal input current, without the
oriented control harmonics usually associated with present commercial
inverters.
The physical realization of the matrix converter is not
straightforward, due to the fact that there are no free-
Introduction wheeling paths. In addition, the number of devices in the
power circuit is high compared with that in the inverter
The matrix converter is the most general converter-type (for instance 18 switches and 18 diodes). Consequently,
in the family of a.c. to a.c. direct converters. On the one the timing of the switch actuation signals is particularly
hand, the matrix converter fulfills the requirements to critical, and protection of the circuit under fault condi-
provide a sinusoidal voltage at the load side and, on the tions requires very careful consideration [3]. No indus-
other hand, it is possible to adjust the unity power factor trial use has been made of this converter type up to now.
on the mains side under certain conditions [1]. Since However, there are a few serious researches on inte-
there is no d.c.-link as in common converters, the matrix grating bi-directional switches into a power module,
converter can be built as a full-silicon structure. How- which will eventually result in industrial use of the ma-
ever, a mains filter is necessary to smooth the pulsed trix converter. In [4], a highly compact converter using
currents on the input side of the matrix converter. Using novel high-power 3-in-1 integrated power modules was
a sufficiently high pulse frequency, the output voltage used. In another paper, the integration of the matrix
and input current both are shaped sinusoidally. The converter and the induction motor in a single unit was
matrix converter is an alternative to an inverter drive discussed [5]. The first matrix converter contained in a
for three-phase frequency control. The converter con- single power module using insulated gate bipolar tran-
sists of nine bi-directional switches, arranged as three sistor technology was presented in [6]. There is now
sets of three so that any of the three input phases can be competition between the matrix converter and the volt-
age-source inverter with a regenerative input rectifier.
Only a few technical papers have dealt with the dy-
H. Altun Æ S. Sünter (&) namic behavior analysis of the field-oriented controlled
Department of Electrical and Electronic Engineering, matrix converter motor drive. In [7, 8], a simulation of
Firat University, 23119 Elazig, Turkey
E-mail: ssunter@firat.edu.tr the matrix converter feeding an induction motor was
26

Fig. 1 Matrix converter


schematic block diagram
representation

performed. and in [9], a control technique for compen- where va, vb, vc are the target phase output voltages.
sating the effects of the input voltage variations on the Alternatively, in a closed loop system (for example a
matrix converter algorithms was described, but closed field-oriented controlled drive), the voltage magnitude
loop operation was not considered. Closed loop opera- and angle may be direct outputs of the control loop.
tion has become an important issue for the viable Then, the voltage ratio is calculated
application of matrix converters [10, 11]. This paper sffiffiffiffiffiffiffiffi
focuses on the dynamic simulation of the field-oriented Vom2

controlled matrix converter drive that has servo per- q¼ 2


ð5Þ
Vim
formance and four-quadrant operation capability.
where q is the desired voltage ratio, and Vim is the peak
input voltage.
Control algorithms Triple harmonic terms are found

Matrix converter algorithm 2 q


K31 ¼ sinðxi tÞ sinð3xi tÞ ð6Þ
9 qm
A simplified version of the Venturini algorithm is used in  
this work [12]. This algorithm is defined in terms of the 2 q 2p
K32 ¼ sin xi t  sinð3xi tÞ ð7Þ
three-phase input and output voltages at each sampling 9 qm 3
instant and is convenient for closed loop operations. For qffiffiffiffiffiffiffiffi 
the real-time implementation of the proposed modula- 2
1 1 1
K33 ¼  Vom cosð3xo tÞ  cosð3xi tÞ ð8Þ
tion algorithm, it is required to measure any two of three 6 4 qm
input line-to-line voltages. Then, Vim and xit are calcu-
lated as where qm is the maximum voltage ratio (0.866).
Then, the three modulation functions for output
2
Vim ¼ 49ðv2AB þ v2BC þ vAB vBC Þ ð1Þ phase a are given as
! 2  
vBC MAa ¼ 13 þ k31 þ 2
ðva þ k33 Þ 23vAB þ 13vBC ð9Þ
xi t ¼ arctan pffiffiffi2  ð2Þ 3Vim
3 3 vAB þ 13 vBC
1 2  
where vAB,vBC are the instantaneous input line voltages. MBa ¼ þ k32 þ 2 ðva þ k33 Þ 13vBC  13vAB ð10Þ
3 3Vim
The target output peak voltage and the output posi-
tion are calculated as MCa ¼ 1  ðMAa þ MBa Þ ð11Þ
2 2 2
Vom ¼ 3ðva þ v2b þ v2c Þ ð3Þ The modulation functions for the other two output
  phases, b and c are obtained by replacing vb and vc
vb  vc with va, respectively in Eqs. (9) and (10). Note that the
xo t ¼ arctan pffiffiffi ð4Þ
3va modulation functions have third harmonic components
27

at the input and output frequencies added to them to Field-oriented controller


produce output voltage, vo. This is a requirement to
get the maximum possible voltage ratio [2]. It should The field-oriented control strategy was implemented in
be noted that in Eq. (3) there is no requirement for the matrix converter induction motor drive. The induc-
the target outputs to be sinusoidal. In general, three- tion machine is controlled in synchronously rotating d–q
phase output voltages and input currents can be de- axis frame with the d axis oriented along the stator flux
fined in terms of the modulation functions in matrix vector position. In this way, a decoupled control be-
form as tween the electrical torque and the rotor excitation
current is obtained. The indirect field-oriented control
voph ¼ Mviph technique using impressed voltages and control of field
2 3 2 32 3
va MAa MBa MCa vA and torque current components was implemented in the
6 7 6 76 7 ð12Þ drive system. The control requires the measurements of
4 vb 5 ¼ 4 MAb MBb MCb 54 vB 5
the stator currents and the rotor position. Equa-
vc MAc MBc MCc vC tions (17), (18), and (19) are the fundamental equations
for field-oriented control [13], and allow the induction
motor to act like a separately excited d.c. machine with
iiph ¼ M T ioph decoupled control of torque and flux, making it possible
2 3 2 32 3 to operate the induction motor as a high-performance
iA MAa MAb MAc ia four-quadrant servo drive.
6 7 6 76 7 ð13Þ
4 iB 5 ¼ 4 MBa MBb MBc 54 ib 5 dimrd
iC MCa MCb MCc ic isd ¼ sr þ imrd ð17Þ
dt
where the superscript T denotes a transpose, and M is  
1
the instantaneous input-phase to output-phase transfer xsl ¼ isq ð18Þ
sr imrd
matrix of the three-phase matrix converter. viph and
voph are the input and output phase voltage vectors,   2
P L0
and iiph and ioph represent the input and output phase Te ¼ 3 isd isq ð19Þ
current vectors. Alternatively, from Eqs. (12) and (13), 2 Lr
the output-line voltages and input-line currents can be Figure 2 shows a schematic block diagram for the
expressed as field-oriented controlled matrix converter induction
voLine ¼ mviLine motor drive where three output currents and rotor po-
2 3 2 32 3 sition are required to be measured. The field-oriented
vab mAb mBb mCb vAB control method shown in Fig. 2 imposes a rotor flux
6 7 6 76 7 ð14Þ
4 vbc 5 ¼ 4 mAc mBc mCc 54 vBC 5 vector angle Qe which is aligned to the d axis. The motor
vca mAa mBa mCa vCA speed, xr is measured and compared with the demanded
speed xr*. The resulting speed error is then processed by
and a proportional-integral (PI) controller to produce an isq*
demand, which in the constant torque region is pro-
iiLine ¼ mT ioLine portional to the torque demand, providing that the
2 3 2 32 3 system is field-oriented. The flux current demand is
iAB mAb mAc mAa iab
6 7 6 76 7 ð15Þ maintained constant at just under saturation level when
4 iBC 5 ¼ 4 mBb mBc mBa 54 ibc 5 the machine runs below synchronous speed. However,
iCA mCb mCc mCa ica field weakening must be introduced above synchronous
speed at base frequency operating conditions so that the
where flux current reference is reduced as the speed is increased
above its synchronous base.
mAb ¼ 13ðMAa  MAb Þ  13ðMBa  MBb Þ The transformation of the instantaneous stator cur-
mBb ¼ 13ðMBa  MBb Þ  13ðMCa  MCb Þ rents into field-oriented d and q axis components is carried
out in two stages. First, the three instantaneous currents
mCb ¼ 13ðMCa  MCb Þ  13ðMAa  MAb Þ isa(t), isb(t), and isc(t) are transformed to the stationary two
mAc ¼ 13ðMAb  MAc Þ  13ðMBb  MBc Þ axis currents, isa(t) and isb(t). These are then transformed
mBc ¼ 13ðMBb  MBc Þ  13ðMCb  MCc Þ ð16Þ into the rotating d–q axis currents, isd and isq. The equiv-
1 1
alent complex operator e)jQe is used in this transforma-
mCc ¼ 3ðMCb  MCc Þ  3ðMAb  MAc Þ tion. Qe denotes the instantaneous flux vector angle,
mAa ¼ 1 1
3ðMAc  MAa Þ  3ðMBc  MBa Þ
which is determined by summing the rotor position signal
1 1 and the commanded slip position obtained by integrating
mBa ¼ 3ðMBc  MBa Þ  3ðMCc  MCa Þ Eq. (18). The inverse transformation of d and q axis values
1 1
mCa ¼ 3ðMCc  MCa Þ  3ðMAc  MAa Þ to the instantaneous stator reference frame is represented
28

Fig. 2 Block diagram for vector controlled matrix converter induction motor drive
2 3 2 32 3
by the complex operator ejQe. The two current controllers wsq Ls 0 Lo 0 Isq
which employ PI control process the isd and isq errors to 6 wsd 7 6 0 Ls 0 Lo 7 6 Isd 7
6 7 6 76 7
give V¢sd and V¢sq. Voltage compensation terms are added 4 wrq 5 ¼ 4 Lo 0 Lr 0 54 Irq 5
ð21Þ
to the output of each current controller to get the resulting
wrd 0 Lo 0 Lr Ird
voltage reference signals Vsd* and Vsq*. These voltages are
then converted to the three-phase voltages using the Ls ¼ Lls þ Lo
complex operator ejQe. The three-phase voltages Va*, Vb*, ð22Þ
Vc*, and instantaneous flux vector angle Qe are used as the Lr ¼ Llr þ Lo
input signals for the Venturini algorithm (Eqs. 1, 2, 3, 4, 5,
6, 7, 8, 9, 10, 11) to generate the duty cycles for each switch P Lo
Te ¼ 3 ðIsq wrd  Isd wrq Þ ð23Þ
in the matrix converter. 2 Lr
dxmech
J ¼ Te  TL  fv xmech ð24Þ
dt
Induction motor d–q model
where quantities with subscript q or d denote q axis or d
The simulation equations for an induction motor in the axis quantities and quantities with subscript s or r de-
d–q synchronously rotating reference frame are given as note stator or rotor quantities. w denotes flux linkage, R
[14] is resistance, Lls and Llr are the stator and rotor leakage
2 3 2 32 3 inductances, respectively, and Lo is the magnetizing
Vsq Rs 0 0 0 Isq inductance. Ls and Lr denote the self-inductances of the
6 Vsd 7 6 0 Rs 0 0 76 Isd 7 stator and rotor, respectively. Te and TL are the motor
6 7 6 76 7
6 7¼6 76 7 torque and load torque, respectively. P is the number of
4 0 5 4 0 0 Rr 0 54 Irq 5
poles, and fv is the friction constant coefficient. J is
0 0 0 0 Rr Ird inertia and xmech is the mechanical speed of the motor.
2 32 3 All rotor quantities are referred to the stator.
p xs 0 0 wsq
6 xs p 0 0 76 w 7
6 76 sd 7
þ6 76 7
4 0 0 p ðxs  xr Þ 54 wrq 5 Simulink model
0 0  ðx s  x r Þ p wrd
The simulation program constructed in Matlab/Simu-
ð20Þ link software package [15] comprises three main parts,
29

which are the matrix converter, induction motor, and have logic levels. Logic gates are used at the output to
field-oriented controller. Figure 3 shows the Simulink get three gate signals proportional to the duty cycle of
blocks of the whole drive system, where the d–q model the power switches for one output phase. Three
of the induction motor is placed on top of the figure. ‘‘switches’’ blocks are used to form one output phase of
The simulation model of the matrix converter can be the converter as shown in Fig. 4. The ‘‘switch’’ option is
seen among the mid blocks. Remaining parts of Fig. 3 a Simulink building block. It has three inputs [in (1),
belong to the field-oriented controller and measurement in(2) and in(3)] and one output. It operates in accor-
processes. dance with the following logic
– if in(2)>0 then output=in(1)
– else output=in(3)
Matrix converter model
The output of the switches connected together forms
Figure 4 shows the Simulink model for one output phase one output phase of the matrix converter and feeds the
of the matrix converter in detail. The other two phases induction motor.
are not shown for clarity, since they are same except for
a phase shift of 120 and 240. Ideal switches have been
assumed in the simulation of the matrix converter power Induction motor model
circuit. In Fig. 4, the input variables of the matrix con-
verter are the clock and target output voltages obtained The Simulink model of a squirrel cage induction motor
from the vector controller. The Simulink blocks, ‘‘MAa’’, in the d–q synchronously rotating reference frame with
‘‘MBa’’, and ‘‘MCa’’, represent Eqs. (9), (10), and (11), Vsq, Vsd, and xs as input variables is illustrated in Fig. 6.
respectively. Similarly, the blocks ‘‘k31–k32’’ and ‘‘k33’’ Here, Eqs. (20), (21), (22), (23), and (24) are imple-
represent Eqs. (6), (7), and (8), respectively. The block mented using Simulink blocks. The output variables of
for ‘‘q’’ calculates the required voltage ratio using the motor model are the stator currents and rotor speed.
Eq. (5). The switching frequency of the matrix converter In addition, the rotor currents, motor flux, and elec-
is determined by the ‘‘Sawtooth’’ block (in this case tromagnetic torque are output for monitoring purposes.
2 kHz). Equations (1), (2), (3), and (4) are represented The specifications of the motor are given in the
by blocks labelled ‘‘Vim^2-wit’’ and ‘‘Vom^2-wot’’. The Appendix.
block, ‘‘M functions’’ consists of the modulation func-
tions and is taken out for calculating the input currents
using Eq. (15). Simulation results
The block ‘‘Duty cycle generator’’ is shown in Fig. 5
in detail. The modulation functions are compared with Simulation of the matrix converter fed induction motor
the sawtooth waveform at the input and arranged to with field-oriented controller was performed for various

Fig. 3 Simulink model of the


drive system
30

Fig. 4 Detail of the Simulink


block diagram of the matrix
converter for one output phase

Fig. 5 Detail block diagram for


the duty cycle generator

operating conditions. Figure 7 shows the motor speed As seen from Fig. 8 before the speed change command
and torque variations for various speed reference con- is received the input current and input voltage are in
ditions on no load. In this case, the motor is accelerated phase resulting in unity input displacement factor.
from standstill to 1,000 rpm and stays in steady state for However, the input current suddenly changes its phase
a while (first region). Then, the reference is changed to to 180 with respect to the input voltage as the
)1,000 rpm and the motor speed decelerates following reference speed change command is received, and
this reference signal. Therefore, transition from the first regenerative energy of the motor is transferred to the
to the second region occurs, and the machine operates as input supply. Similar results for the transition from
a generator. The motor torque remains in saturation the third to the fourth region were obtained as shown
until the speed reaches its final value. After the motor in Fig. 9.
speed reaches )1,000 rpm and remains at that speed in Figure 10 demonstrates the simulation waveforms for
for a certain time (third region), the reference command the induction motor in steady-state operation showing
signal is again changed to 1,000 rpm resulting in a asynchronous operation of the matrix converter with
transition from the third to the fourth region. Eventu- unity input displacement factor regardless of the load
ally, the motor reaches 1,000 rpm and steady-state power factor. This is the main advantage of the matrix
operation is provided in the first region. As can be seen converter compared with voltage-source inverters with
from this figure, the controller provides four-region an uncontrolled input rectifier. The torque–speed curve
operation for the drive system. of the matrix converter-fed induction motor for all four-
Figure 8 illustrates the input phase voltage, input quadrant regions is shown in Fig. 11. The input power
line current, motor speed, and motor current for the variation of the matrix converter in constant torque
transition from the first region to the second in detail. region for four-quadrant operation is given in Fig. 12.
31

Fig. 6 Detail Simulink block diagram of the induction motor

Fig. 7 Motor torque and speed in four regions Fig. 9 Waveforms for the transition from the third to the fourth
region

Fig. 8 Waveforms for the transition from the first to the second Fig. 10 Input voltage, input current, and motor current waveforms
region for 1,000 rpm in steady-state operation

In this simulation result, it can be seen that variation of The simulation waveforms of the input current, out-
the input power is linear with the speed demand. Note put current, and voltage of the vector-controlled matrix
that, while the negative input power corresponds to ei- converter induction motor drive are illustrated in
ther the second or fourth region operations, positive Fig. 13, where the motor operates with a constant
power corresponds to the first or third region. speed of 1,500 rpm. The voltage and current spectra
32

Fig. 11 Torque-speed characteristic in four-region operation of the


drive system

Fig. 14a, b Harmonic spectrum corresponding to Fig. 13: a output


line voltage; b input current

Fig. 12 Input power variation of the matrix converter for four-


quadrant operation

Fig. 15 Three-phase steady-state motor currents for 500 rpm

out. A high-performance vector-controlled drive


employing the matrix converter was presented. Instead
of the inverter with a d.c. link in the vector-controlled
drive system, the use of the matrix converter made the
drive system capable of operating in all four-quadrant
Fig. 13 Output line voltage, motor current, and input current of regions. It was demonstrated that the matrix converter is
the matrix converter in steady state for 1,500 rpm
capable of operating with unity input displacement
factor regardless of the load power factor at the output.
corresponding to those results are demonstrated in The simulation results agree with the nature of high-
Fig. 14. Here, dominant harmonics are around the performance motor drive system.
switching frequency of 2 kHz as expected. Figure 15
shows three-phase motor currents in steady state for
500 rpm operation. Appendix

The ratings of the three phase 50 Hz, 415 V, 4 kW, delta


Conclusions connected, 1,420 rpm squirrel cage induction motor are:
Rs=5.32 W, Rr¢=4.14 W, T=27 Nm, Ls=0.6 H, Lr=
In this paper, a simulation study of the field-oriented 0.59 H, J=0.4 kg m2, Lo=0.565 H, B=0.707 Nm s/rad,
matrix converter-fed induction motor drive was carried cos /=0.83, Is=8.1 A, P=4 poles.
33

8. Zuckherberger A, Weinstock D, Alexandrovitz A (1997)


References Single-phase matrix converter. IEE Proc Trans Electr Power
Applic 144:235–240
1. Venturini M (1980) A new sine wave in sine wave out conver- 9. Sünter S, Altun H, Clare JC (2002) A control technique for
sion technique which eliminates reactive elements. In: Pro- compensating the effects of input voltage variations on matrix
ceedings of Powercon 7, San Diego, Calif., pp E3-1, E3-15 converter modulation algorithms. Electric Power Components
2. Alesina A, Venturini M (1981) Solid-state power conversion: a Syst 30:807–822
Fourier analysis approach to generalized transformer synthesis. 10. Matsuo T, Bernet S, Colby RS, Lipo TA (1998) Modeling and
Proc IEEE Trans Circuit Syst 28:319–330 simulation of matrix converter/induction motor drive. Math
3. Empringham L, Wheeler PW, Clare JC (1998) Intelligent Comput Simul 46:175–195
commutation of matrix converter bi-directional switch cells 11. Bouchiker S, Capolino GA, Poloujadoff M (1998) Vector
using novel gate drive techniques. In: IEEE PESC’98, control of a permanent-magnet synchronous motor using
Fukuoka, Japan, 17–22 May. IEEE, Piscataway, N.J., pp 707– AC–AC matrix converter. Proc IEEE Trans Power Electron
713 13:1089–1099
4. Chang J, Sun T, Wang A (2002) Highly compact AC–AC 12. Sünter S (1995) A vector controlled matrix converter induction
converter achieving a high voltage transfer ratio. Proc IEEE motor drive. PhD Thesis, Department of Electrical and Elec-
Trans Ind Electron 49:345–352 tronic Engineering, University of Nottingham
5. Klumpner C, Nielson P, Boldea I, Blaabjerg F (2002) A new 13. Leonhard W (1985) Control of electrical drives. Springer,
matrix converter motor (MCM) for industry applications. Proc Berlin Heidelberg New York
IEEE Trans Ind Electron 49:325–335 14. Krause PC, Thomas CH (1965) Simulation of symmetrical
6. Simon O, Mahlein J, Muenzer MN, Bruckmann M (2002) induction machinery. Proc IEEE Trans Power Apparatus Syst
Modern solutions for industrial matrix-converter applications. 84:1038–1053
Proc IEEE Trans Ind Electron 49:401–406 15. Math Works (1999) MATLAB for Microsoft Windows. Math
7. Zuckherberger A, Weinstock D, Alexandrovitz A (1996) Sim- Works, Mass.
ulation of three-phase loaded matrix converter. IEE Proc Trans
Electr Power Applic 143:294–300

You might also like