A FCS-MPC of an Induction Motor Fed by Indirect

Matrix Converter with Unity Power Factor Control

S. M. Muslem Uddin, Saad
Mekhilef M. Rivera J. Rodriguez
Department of Electrical Engineering Department of Industrial Departamento de Electrónica,
University of Malaya, Kuala Lumpur Technologies
50603, Malaysia Universidad Técnica Federico Santa
Universidad de Talca, Curico, Chile María (USM), Avenida España 1680,
E-mail:
E-mail: marcoesteban@gmail.com Casilla 110-V Valparaíso, Chile
muslem_eee04@siswa.um.edu.my,
saad@um.edu.my E-mail: jrp@usm.cl

Abstract— A torque and flux FCS-MPC (Finite Control Set Model most important improvement of this topology is the simpler and
Predictive Control) of an induction motor with unity power factor less complex in the control compared to DMC and allowing
fed by an indirect matrix converter has been investigated and secure commutation of the system [3] without particular sensing
presented in this paper. The proposed control scheme utilizes the devices as needed for DMC [4].
discrete nature of power converter and predicts the future
Investigation concern to torque control of induction motor
behavior of the system variables. Also the method selects the
switching state that minimizes a cost function to actuate the have been carried out in [5] considering back-to-back and
converter in the next sampling time. The proposed FCS-MPC matrix converters (MCs) with pulse width modulation (PWM)
algorithm is validated through simulations and shows potential schemes accompanied with pure sinusoidal output current and
control and tracking of variables with their respective references. unity power factor. Recently new research works with
multilevel inverter have been introduced in [6-9]. Also,
Keywords—Matrix converters; Predictive control; AC-AC power synchronous PWM for Fly back converter [10] and symmetrical
converters; Support vector machines; Control and modulation sampling digital PWM for power converters have been
strategies investigated in [11]. Furthermore, different applications in
voltage source inverter and MC with predictive control has been
I. INTRODUCTION successfully investigated in recent few years [12-19]. Also
AC-AC power converters are very much important and predictive control technique have been applied in the IMC for
widely used in industries. DC-link storage elements (Capacitors different inquisitions such as; current control [20], current
or Inductors) are advantageous for decoupling of control process control extended by reactive power minimization [21],
but energy storage components are usually large in size and implementation of virtual damping resistance to mitigate the
responsible for reducing the life span of the converter. To resonance effect due to low damped input filter [22], torque-flux
overcome these disadvantages different AC-AC power converter controls with reactive power minimization [23] and imposed
topologies have been investigated in literature and spreads in sinusoidal load and source currents [24] etc. Furthermore a
three main groups, such as: 1) cycloconverter (used in wide comprehensive review on most exoteric control strategies and
range of power); 2) direct matrix converter (DMC) and 3) momentous impact on the resonance of the MC input filter have
indirect matrix converter (IMC) applied in low power range [1]. been discussed in [25]. The successive improvement and rising
Cycloconverter converts AC-AC without any intermediate capabilities of signal processing technologies and semiconductor
energy storage devices hence reduce the sizes and increases life modern microprocessors has been made possible to implement
time but it has a great limitation with output frequency which more complex and sophisticated control algorithm precisely to
consists of significant harmonic contents due to the switching satisfy the high performances in the industries. Predictive
actuations and these contents of harmonics cannot be filtered by control is the method which attained a great contemplation for
load inductance. In the DMC an array of (3×3) bidirectional power converter and drive systems. In this investigation a FCS-
switches consist and directly connected to load from grid supply MPC technique has been proposed to control induction motor
without any intermediate DC-link storage devices but its control flux and torque accurately. A FCS-MPC control is nothing but
strategy is so complex. To overcome all the drawbacks stated take advantages of the system model to make prediction of the
above the indirect matrix converter topology has been future nature of the system variables concern to determine the
investigated which offers similar performance to DMC [2], but cost functions for all possible switching states and selects the

978-1-4673-6322-8/13/$31.00 2013
c IEEE 1769
optimal firing actuation to converter for the next sampling
period based on minimum cost function[26]. This inquisition is
mainly focused on control of induction motor torque, flux and
zero displacement input angle between input voltage and current
in the IMC with FCS-MPC control algorithm.
This paper is organized as follows: Section II is related to
mathematical background of the IMC topology and inductive
load model of the system. Section III presents proposed FCS-
MPC torque and flux control algorithm. Section IV states the
simulation results and analysis of the investigations. Finally a
fruitful conclusion is drawn in section V.

II. THE INDIRECT MATRIX CONVERTER TOPOLOGY Fig. 1 Indirect matrix converter topology
A. Mathematical Background of the Topology
The indirect matrix converter topology is shown in Fig. 1
valid states but also inverter is associated with eight valid states.
which consists of a rectifier and an inverter. Inverter part of the
There are some operating restrictions in the indirect matrix
IMC is connected to the rectifier and power supply through an
converter such as: no short circuit in the input, no open circuit at
intermediate virtual DC-link. This is one of the most important
the output and positive DC-link ( Vdc > 0 ) between the rectifier
advantages of the indirect matrix converter. Positive DC-link
voltage is synthesized by the converter selecting the rectifier and inverter parts. To satisfy the first two constraints stated
switching states which connect one phase to positive (+) and above, 72 possible switching states are applicable but if the
other phase to negative (-). DC-link voltage is obtained as a constraint regarding positive DC-link is included then possible
function of rectifier switches ( S r ) and input voltages ( Vi ) of switching states are reduced to only 24, because rectifier valid
switching states have reduced to three from nine possible states
the converter, to maintain positive DC-link between rectifier and inverter [20].
Vdc = [S1r − S4r S3r − S6r S5r − S2r ]Vi (1) It is notable that IMC topology provides an extra freedom for
reducing the commutation sequence complexity, the zero DC-
where S 1 r to S 6 r are the rectifier switching states and input link current Commutation [3]. In addition, rectifier should
include an input LC filter to avoid the over-voltages [23] and
voltage Vi = [Vi a Vi b Vi c ]T . The rectifier input current I ri is filtering the effect of high order harmonics of input currents
defined as a function of the rectifier switches and DC-link generated by the commutations and inductive features of the
current I dc as, load.
B. Inductive Load Model
ª S 1r − S 4 r º
(2) A two dimensional space vector (SV) can be represented
I ri = «« S 3 r − S 6 r »» I dc from three phase quantities of the converter to obtain the system
¬« S 5 r − S 2 r »¼ of the model. For example complex space vector described from
the three phase quantities y A , y B and y C are given as,
In equation (2) the input currents are given as,
y = yα + jy β (5)
I ri = I [ i
a
I b
i
c T
]
I . On the other hand, inverter part is modeled
i
as, where, yα = 1 (2yA − yB − yC ), y = 1 ( y − y ) (6)
β B C
3 3
ª S 1i − S 4 i º In equation (6), stationary reference frame ( αβ -reference frame)
(3)
V o = «« S 3i − S 6 i »»V dc is considered for expected SV. The model of the induction
«¬ S 5 i − S 2 i »¼ motor referred to stator is obtained as described in [13, 27].
Fixed coordinate rotor and stator voltage equations (induction
motor model in SV representation) are presented as,
I dc = [S1i S3i S5i ]I o (4) .
Vo = Rs I o + Ls ψ s (7)
Here [
Vo = VoA VoB VoC ]T
is the output voltage and .
Vr = Rr I r + Lr ψ r − jpωψ s (8)
[ T
]
Io = I oA I oB I oC corresponds to the output current.
where R s , R r , p and ω are the stator resistance, rotor resistance,
In this topology not only rectifier model corresponds to nine number of pole pairs and rotor angular frequency, respectively.

1770 2013 IEEE 8th Conference on Industrial Electronics and Applications (ICIEA)
The stator flux and rotor flux are related with the following
equations,
ψ s = Ls I o + Lm I r , ψ r = Lr I r + Lm Io (9)
where L s , L r are self-inductances and L m is the mutual
inductance of the induction motor. Finally, electrical torque is
developed in the induction motor can be represented by current
and flux terms,
3 (10)
T = p (ψ × I )
e s o
2

III. PROPOSED FCS-MPC TORQUE AND FLUX CONTROL

ALGORITHM
FCS-MPC algorithm uses the finite number of valid actuation
states of the power converter. The proposed scheme maintains
the predictive values closed to their respective references at the Fig. 2 Proposed control scheme
end of the sampling instant and maintain positive DC-link
voltage between the rectifier and inverter stages. The proposed where T s is the sampling period; so, rotor and stator flux can be
FCS-MPC control scheme and algorithm are presented in Fig. 2 estimated from (7)-(8) resulting in,
and Fig. 3, respectively. Predictive controller satisfies all the
aforementioned constraints by using the following five steps ψ sk = ψ sk −1 + Vok Ts − Rs I ok Ts (12)
[27]:
• For the k th sampling instant, input current I rik , supply L s Lr L (13)
ψ rk = I ok ( L m − ) + r ψ sk −1
k k k Lm Lm
voltage V , input voltage V i , stator currents I and
s o

speed ω k of the induction motor are measured. Therefore, in accordance with [17] predictive stator flux is as
follows,
• Stator reference flux ψ ref , reference speed ω ref , and
ψ sk +1 = ψ sk + V ok +1T s − R s I ok +1T s (14)
reference reactive power qref are known values. PI
controller is used to set nominal torque T nom from the where Vok +1 is derived from (3) and predictive stator current [17]
error signal between the measured and reference speeds is,
of the induction motor.
• Ts Tr (15)
A flux estimator has been used to estimate the stator I ok +1 = ψ rk (Vok +1 + (τ r k r − jk r ω )) + I ok (1 − s σ )Ts
and rotor flux. σL s σLs
• For each valid switching states of indirect matrix
converter, values of torque T e k + 1 and stator flux where rσ = Rs + Rr k r2 , τ = Lr , k = L m , k = L m and
r s r
Rr Ls Lr
ψ s k +1 are predicted in the next sampling period (k+1).
σ = 1 − kskr .
• All the predictive values are compared with their
respective references and determined the cost functions Therefore predictive torque of induction motor for the next
of all possible switching states. The switching state sampling period is as follows,
corresponds to the minimum cost function is selected in
the next sampling time actuation. 3
Tek +1 = p (ψ sk +1 × I ok +1 ) (16)
2
A. Predictive Torque and Flux Calculation
A first order differential equation is used for prediction of the
The induction motor model has explained in equations (7)- input voltages and input currents [23] have shown in (17) and
(10) are used with first order approximation for derivatives (18),
along with the first order nature of state equations,
.
y k +1 − y k (11) Vi k +1 = φ11Vi k + φ12 I sk + γ 11V sk + γ 12 I rik (17)
y= T
s
I rik +1 = φ 21V i k + φ 22 I sk + γ 21V sk + γ 22 I rik (18)

2013 IEEE 8th Conference on Industrial Electronics and Applications (ICIEA) 1771
 reactive power compensation is analyzed. While in the second
case reactive power compensation has taken into consideration.
In both the cases, the induction motor starts at 0.01s without any
load torque, varying the reference speed from 0 to 149.75 rad/s
and the torque limited to 51 N-m. A load torque of 40 N-m is
applied at time 0.25s and a reverse torque at 0.35s is applied to
change the speed in the reverse direction from 149.75 rad/s to -
149.75 rad/s. In this paper, stator reference flux has been
assumed as 1.1 Wb. The speed controller generates torque
references at transients which is different from zero and be
appreciated as a well tracker of torque given in Fig. 4(a) and
Fig. 5(a), of speed indicated in Fig. 4(b) and Fig. 5(b) and of
flux depicted in Fig. 4(c), Fig. 5(c).

Fig. 3 FCS-MPC torque and flux control flow diagram.

The real coefficients φ ij and γ ij are defined to obtain predicted

values of currents correspond to those of the continuous time (a)
system after one period of Ts , as depicted in [20].

B. Cost Function Calculation

Absolute error is considered for computational simplicity in
cost function. The predicted torque with its nominal value and
predicted flux correspond to reference can be combined in a
single term to express the cost function.
h k +1 = X (Tek +1 − Tnom ) + Y (ψ sk +1 −ψ ref ) (19)
h k +1 = XΔTek +1 + YΔψ sk +1 (b)

In equation (19), X and Y are the priority control factors and

(19) can be used for predictive control of the torque and flux. To
maintain unity power factor, reactive power term can be
included in the cost function which is expressed in (20),
Δ q sk + 1 = ( I skβ+ 1V skα+ 1 − I skα+ 1V skβ + 1 ) − 0 (20)
Thus, the cost function implies:
h k +1 = X Δ Tek +1 + YΔ ψ sk +1 + Z Δ q sk +1 (21)
(c)
Z is the new weighting factor with reactive power term.
Different values of cost functions have been calculated by (19)
and (21) correspond to all possible switching states. The
switching state which produces minimum cost function is
selected for firing actuation in next T s [12-14].

IV. SIMULATION RESULTS AND ANALYSIS

The FCS-MPC technique is simulated in MATLAB Simulink
to justify the performance of the proposed control scheme. The
parameters used in simulation are given in Table I and the (d)
simulation has been carried out with Ts =20 μS. Two cases are Fig. 4 Simulation studies without reactive power compensation: (a) Nominal
torque, ( Tnom ) [Nm] and induction motor torque, ( Te ) [Nm] (b) Reference speed,
analysed. First, the validation of the control scheme without

1772 2013 IEEE 8th Conference on Industrial Electronics and Applications (ICIEA)
( ω ref ) [rad/s] and motor speed, ( ω ) [rad/s]; (c) Stator reference flux, (ψ ref ) A. FCS-MPC Control Without Reactive Power Compensation
Term in Quality Function
[Wb] and stator flux, (ψ s ) [Wb]; (d) Supply voltage, ( V s ) [V] and supply
current ( I s ) [A].
In Fig. 4(d), a chaoitic behavior is observed in supply
current of indirect matrix converter because of Z=0, i.e;
regardless of reactive power compensation in the cost function.
The angle between the input voltage and input current is not
equal to zero and a notable magnitude of angle existed between
the two variables. Consequently, a significant amount of
reactive power creates in the input side of the converter and
responsible for such a chaotic characteristics of the input current
. This draw back can be improved in section B.
B. FCS-MPC Control With Reactive Power Compensation
Term in Quality Function
(a)
The second analysis with reactive power minimization,
improve the chaotic characteristics of the input current of IMC.
Weighting factor (Z = 1) and reactive power reference deem
negligible ( qref = 0). Reactive power generated at the input side
are compared with its reference and tracks to reference value
which effects to minimize the cost function. Therefore,
minimum reactive power will be generated for the next
operating period. Fig. 5(d) shows the compensate behavior of
the input current of the converter. Another important feature is
(b)
substantial in Fig. 5(d) that unity power factor is maintainted
strictly in the input of the power converter because angle
between input voltage and input current is zero without the
reverse condition of the load torque which is expected.

V. CONCLUSIONS
A robust and simple FCS-MPC algorithm has been presented
in this paper in order to control torque and flux of induction
motor fed by an indirect matrix converter with input reactive
power minimization. The control scheme has utilized the
(c) discrete features of input filter, power converter and inductive
load to predict the future behavior of the flux, torque and input
current of the system. It has also employed to obtain the cost
function for all 24 possible switching states from where the
switching state is selected corresponding to minimum cost
function. However, the system behaviour is highly changeable
with the values of weighting factor and till to date no analytical
method is existed for designing this issue. A future study will be
done by calculating the optimum weighting factor to control
sensitibity of the system variables precisely.
(d)

Fig. 5 Simulation studies with reactive power compensation: (a) Nominal ACKNOWLEDGEMENT
torque, ( Tnom ) [Nm] and induction motor torque, ( Te ) [Nm]; (b) Reference
The authors wish to thank the financial support from the
speed, ( ω ref ) [rad/s] and motor speed, ( ω ) [rad/s]; (c) Stator reference flux, University of Malaya through HIR-MOHE project
UM.C/HIR/MOHE/ENG/17.
(ψ ref ) [Wb] and stator flux, (ψ s ) [Wb]; (d) Supply voltage, ( V s ) [V] and
supply current ( I s ) [A].

2013 IEEE 8th Conference on Industrial Electronics and Applications (ICIEA) 1773
 REFERENCES [19] J. Rodriguez, J. Pontt, R. Vargas, P. Lezana, U. Ammann, P. Wheeler,
and F. Garcia, "Predictive direct torque control of an induction motor
fed by a matrix converter," in European Conference on Power
[1] P. W. Wheeler, J. Rodriguez, J. C. Clare, L. Empringham, and A. Electronics and Applications, 2007, pp. 1-10.
Weinstein, "Matrix converters: a technology review,"IEEE [20] P. Correa, J. Rodríguez, M. Rivera, J. R. Espinoza, and J. W. Kolar,
Transactions on Industrial Electronics, vol. 49, pp. 276-288, 2002. "Predictive control of an indirect matrix converter," IEEE
[2] T. Friedli, M. Heldwein, F. Giezendanner, and J. Kolar, "A high Transactions on Industrial Electronics, vol. 56, pp. 1847-1853, 2009.
efficiency indirect matrix converter utilizing RB-IGBTs," in 37th [21] J. Rodriguez, J. Kolar, J. Espinoza, M. Rivera, and C. Rojas,
IEEE Power Electronics Specialists Conference, PESC'06. 2006, pp. "Predictive current control with reactive power minimization in an
1-7. indirect matrix converter," in IEEE International Conference on
[3] J. W. Kolar, F. Schafmeister, S. D. Round, and H. Ertl, "Novel three- Industrial Technology (ICIT), 2010, pp. 1839-1844.
phase AC–AC sparse matrix converters," IEEE Transactions on [22] M. Rivera, J. Rodriguez, B. Wu, J. R. Espinoza, and C. A. Rojas,
Power Electronics, vol. 22, pp. 1649-1661, 2007. "Current control for an indirect matrix converter with filter resonance
[4] P. W. Wheeler, J. C. Clare, L. Empringharn, M. Bland, and M. Apap, mitigation," IEEE Transactions on Industrial Electronics, vol. 59, pp.
"Gate drive level intelligence and current sensing for matrix 71-79, 2012.
converter current commutation," IEEE Transactions on Industrial [23] J. Rodriguez, J. Kolar, J. Espinoza, M. Rivera, and C. Rojas,
Electronics, vol. 49, pp. 382-389, 2002. "Predictive torque and flux control of an induction machine fed by an
[5] K.-B. Lee and F. Blaabjerg, "Sensorless DTC-SVM for induction indirect matrix converter with reactive power minimization," in IEEE
motor driven by a matrix converter using a parameter estimation International Symposium on Industrial Electronics (ISIE), 2010, pp.
strategy," IEEE Transactions on Industrial Electronics, vol. 55, pp. 3177-3183.
512-521, 2008. [24] M. Rivera, J. Rodriguez, J. R. Espinoza, T. Friedli, J. W. Kolar, A.
[6] S. Mekhilef and M. A. Kadir, "Novel vector control method for Wilson, and C. A. Rojas, "Imposed Sinusoidal Source and Load
three-stage hybrid cascaded multilevel inverter," IEEE Transactions Currents for an Indirect Matrix Converter," IEEE Transactions on
on Industrial Electronics, vol. 58, pp. 1339-1349, 2011. Industrial Electronics, vol. 59, pp. 3427-3435, 2012.
[7] S. Mekhilef and A. Kadir, "Voltage control of three-stage hybrid [25] J. Rodriguez, M. Rivera, J. W. Kolar, and P. W. Wheeler, "A review
multilevel inverter using vector transformation,", IEEE Transactions of control and modulation methods for matrix converters," IEEE
on Power Electronics, vol. 25, pp. 2599-2606, 2010. Transactions on Industrial Electronics, vol. 59, pp. 58-70, 2012.
[8] M. Menshawi, M. Menshawi, and S. Mekhilef, "Voltage Vector [26] A. Linder and R. Kennel, "Model predictive control for electrical
Approximation Control of Multistage-Multilevel Inverter Using drives," in IEEE 36th Power Electronics Specialists Conference,
Simplified Logic Implementation." PESC'05. 2005, pp. 1793-1799.
[9] S. Mekhilef, M. AbdulKadir, and Z. Salam, "Digital Control of Three [27] J. Rodriguez, J. Kolar, J. Espinoza, M. Rivera, and C. Rojas,
Phase Three-Stage Hybrid Multilevel Inverter," 2013. "Predictive torque and flux control of an induction machine fed by an
[10] A. M. Omar, N. A. Rahim, and S. Mekhilef, "Three-phase indirect matrix converter," in IEEE International Conference on
synchronous PWM for flyback converter with power-factor Industrial Technology (ICIT), 2010, pp. 1857-1863.
correction using FPGA ASIC design," IEEE Transactions on
Industrial Electronics, vol. 51, pp. 96-106, 2004.
[11] S. Mekhilef, N. Rahim, and A. Omar, "Modelling of three-phase APPENDIX
uniform symmetrical sampling digital PWM for power converter," in The parameters used in simulation are given in Table I.
IEEE 35th Annual Power Electronics Specialists Conference, PESC
04. 2004, pp. 3499-3503.
[12] J. Rodriguez, J. Pontt, C. A. Silva, P. Correa, P. Lezana, P. Cortés, TABLE I
and U. Ammann, "Predictive current control of a voltage source SIMULATION PARAMETERS
inverter," IEEE Transactions on Industrial Electronics, vol. 54, pp. Description Variables Values
495-503, 2007. Ts
Sampling time 20 μs
[13] P. Correa, M. Pacas, and J. Rodriguez, "Predictive torque control for
inverter-fed induction machines," IEEE Transactions on Industrial Supply phase voltage (RMS) Vs 500 V
Electronics, vol. 54, pp. 1073-1079, 2007. Supply frequency fs 50 Hz
[14] P. Cortes, J. Rodriguez, D. E. Quevedo, and C. Silva, "Predictive Rf
Input filter resistance 0.5 Ÿ
current control strategy with imposed load current spectrum," IEEE
Transactions on Power Electronics, vol. 23, pp. 612-618, 2008. Input filter inductance Lf 400 μH
[15] R. Vargas, M. Rivera, J. Rodriguez, J. Espinoza, and P. Wheeler, Input filter capacitance Cf 21 μF
"Predictive torque control with input PF correction applied to an
induction machine fed by a matrix converter," in IEEE Power Reference speed ω ref 149.75 rad/s
Electronics Specialists Conference, PESC 2008. 2008, pp. 9-14. Nominal torque Tnom 51 Nm
[16] M. Rivera, R. Vargas, J. Espinoza, J. Rodriguez, P. Wheeler, and C.
Nominal current I nom 16 A
Silva, "Current control in matrix converters connected to polluted AC
voltage supplies," in IEEE Power Electronics Specialists Conference, Nominal power Pnom 4000 Watt
PESC 2008. 2008, pp. 412-417. Stator inductance Ls 0.2861
[17] M. E. Rivera, R. E. Vargas, J. R. Espinoza, and J. R. Rodriguez,
"Behavior of the predictive DTC based matrix converter under Stator resistance Rs 1.35 Ÿ
unbalanced AC supply," 42nd IAS Annual Meeting in Industry Number of poles p 2
Applications Conference, 2007. Conference Record of the 2007 IEEE, Priority control factor X 63
2007, pp. 202-207. Priority control factor Y 13500
[18] M. Rivera, J. Rodriguez, P. W. Wheeler, C. A. Rojas, A. Wilson, and
Priority control factor Z 1
J. R. Espinoza, "Control of a matrix converter with imposed
sinusoidal source currents," IEEE Transactions on Industrial
Electronics, vol. 59, pp. 1939-1949, 2012.

1774 2013 IEEE 8th Conference on Industrial Electronics and Applications (ICIEA)