IEEE TRANSACTIONS ON INDUSTRIAL ELECTRONICS, VOL. 63, NO.

9, SEPTEMBER 2016 5879

Model Predictive Control Applied to an Improved

Five-Level Bidirectional Converter
Vı́tor Monteiro, Student Member, IEEE, João C. Ferreira, Senior Member, IEEE,
Andrés Augusto Nogueiras Meléndez, Senior Member, IEEE, and João Luiz Afonso, Member, IEEE

Abstract—This paper presents an improved five-level

bidirectional converter (iFBC) controlled by finite control
set model predictive control (FCS-MPC). This control strat-
egy consists in using the discrete time nature of the iFBC to
define its state in each sampling interval. Using FCS-MPC
the switching frequency is not constant; however, it is suit-
able to follow the current reference with low total harmonic
distortion (THD). The iFBC prototype that was specially de-
veloped for obtaining experimental results is described in
detail along the paper, as well as its principle of operation,
power theory, and current control strategy. The iFBC was ex-
perimentally validated connected to the power grid through
a second-order Lf Cf passive filter operating as an active
rectifier and as a grid tie inverter. For both operation modes,
Fig. 1. Topology of the improved five-level bidirectional converter
the experimental results confirm the good performance (in (iFBC).
terms of efficiency, low current THD, and controlled output
voltage) of the iFBC controlled by FCS-MPC.
Index Terms—Bidirectional, five-level converter, model
predictive control (MPC), power quality. lyzed in [7]. A review of PFC bridgeless converters is presented
in [8], where the well-known PFC symmetrical and asymmetri-
I. INTRODUCTION cal bridgeless converters are included [9], [10]. Other relevant
CTIVE rectifiers have some important advantages when groups of PFC converters are the interleaved and the multilevel
A compared with the traditional solutions based on diode
rectifiers and multipulse rectifiers [1]. These advantages are
[11]–[14]. The main advantage of the multilevel converters is
the possibility of reducing the voltage stress in the semiconduc-
the low total harmonic distortion (THD) of the grid current, tors, and the volume and size of the passive filters [15], [16].
the high total power factor, and the controlled output voltage In this context, this paper presents an improved five-level bidi-
(dc-link) [1]–[3]. The power-factor-correction (PFC) converter, rectional converter (iFBC). The circuit topology of the iFBC is
composed of a diode-bridge rectifier followed by a dc–dc boost- presented in Fig. 1.
type converter, is the most employed converter when the control The most famous five-level converter is the cascade
of these three parameters is required. Similar topologies to meet H-Bridge composed of two H-Bridges [17], [18]. Compar-
these requirements can employ a dc–dc three-level converter ing with the iFBC, it can also operate in bidirectional mode,
[4], a dc–dc buck-boost converter [5], and a buck-flyback con- but it has two independent dc-links and requires more hard-
verter [6]. On the other hand, in order to avoid the front-end ware. A new topology of a unidirectional single-phase five-level
diode-bridge rectifier, the PFC bridgeless converters are ana- converter based on the flying-capacitor rectifier is presented
in [19]. However, this new topology can only be connected
to the power grid as an active rectifier and not as a grid-tie
Manuscript received May 31, 2015; revised February 26, 2016 and
December 24, 2015; accepted March 18, 2016. Date of publication April inverter. Similarly, a new topology of a five-level inverter us-
25, 2016; date of current version August 9, 2016. This work was sup- ing a single dc source cascading a flying capacitor and an H-
ported in part by the FCT–Fundação para a Ciência e Tecnologia in the Bridge converter is presented in [20]. However, this is a three-
scope of the project: PEst-UID/CEC/00319/2013. The work of V. Mon-
teiro was supported by the doctoral scholarship SFRH/BD/80155/2011 phase topology that cannot be adopted for single-phase systems.
granted by the FCT agency. Moreover, each phase requires eight insulated-gate bipolar tran-
V. Monteiro and J. L. Afonso are with the Department of Industrial sistors (IGBTs) and two capacitors, besides the dc-link. In this
Electronics, Centro Algoritmi, University of Minho, 4800-058 Guimarães,
Portugal (e-mail: vmonteiro@dei.uminho.pt; jla@dei.uminho.pt). sequence, a novel strategy to control a five-level inverter using
J. C. Ferreira is with the Instituto Universitário de Lisboa (ISCTE-IUL), a flying-capacitor asymmetric H-Bridge converter is proposed
University Institute of Lisbon, ISTAR-IUL, 1649-026 Lisbon, Portugal in [21]. A novel topology of a five-level inverter, which uses
(e-mail: joao.carlos.ferreira@iscte.pt).
A. A. Nogueiras Meléndez is with the Departamento de Tec- two additional IGBTs comparing with the H-Bridge converter
nologı́a Electrónica, University of Vigo, 36310 Vigo, Spain (e-mail: aau- is presented in [22]. Relatively to the iFBC it requires less two
gusto@uvigo.es). diodes, however, depending on the application (connected to
Color versions of one or more of the figures in this paper are available
online at http://ieeexplore.ieee.org. the power grid as an active rectifier or as a grid-tie inverter),
Digital Object Identifier 10.1109/TIE.2016.2558141 the power in the dc-link sources may not be balanced. Besides,
0278-0046 © 2016 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission.
See http://www.ieee.org/publications standards/publications/rights/index.html for more information.
5880 IEEE TRANSACTIONS ON INDUSTRIAL ELECTRONICS, VOL. 63, NO. 9, SEPTEMBER 2016

this topology uses an IGBT between the two sources of the dc- TABLE I
POSSIBLE STATES OF THE IFBC
link, which can be a great disadvantage for some applications,
e.g., interface from renewables, as photovoltaics. Moreover, this
g1 g2 g3 g4 g5 g6 vc v
topology, at most, can be expanded only to a seven-level topol-
ogy. A new buck-boost five-level converter is presented in [23], Active rectifier vg > 0 0 0 0 0 0 0 +V d c
[24]. However, it also has the dc-link constituted by two inde- 0 0 0 0 1 0 V d c /2
0 0 1 0 0 0 0
pendent sources, such as the cascade H-Bridge converter, limit- vg < 0 0 0 0 1 0 0 0
ing its range of applications. A bidirectional five-level converter 0 0 0 0 0 1 −V d c /2
specially dedicated for photovoltaic applications is proposed in 0 0 0 0 0 0 −V d c
[25], however, it requires two more IGBTs and two diodes com- Grid-tie inverter vg > 0 1 0 0 1 0 0 +V d c
1 0 0 0 0 1 V d c /2
paring with the iFBC. A new single-phase five-level inverter 1 0 0 0 0 0 0
with a dead-beat controller is presented in [26]. This converter vg < 0 0 1 0 0 0 0 0
uses a split dc-link and is composed of an H-Bridge converter 0 1 0 0 1 0 −V d c /2
0 1 1 0 0 0 −V d c
and a bidirectional cell connected between the middle point of
the dc-link and a leg of the H-Bridge converter. This new inverter
was experimentally validated only operating as an inverter [26].
In counterpart, the iFBC uses an enhanced bidirectional cell, be selected, e.g., trajectory based or MPC with finite control set.
i.e., instead of an IGBT with a diode-bridge rectifier, there are Comparing both current control strategies, in the scope of this
two legs, one formed by IGBTs and another by diodes. In the paper, the MPC with finite control set (FCS-MPC) is used to
proposed improved bidirectional cell, only one IGBT is used at define the state of the iFBC in each sampling period, due to its
each time, i.e., it is possible to split the losses among the two simplicity, flexibility, and possibility to include nonlinearities
IGBTs. Moreover, with the improved bidirectional cell, during (e.g., the input Lf passive filter) [30], [31]. Taking into account
each sampling period, instead of an IGBT and two diodes, a that the grid current is the variable that should be controlled
branch with an IGBT and a diode is used. This implies that, us- during both the operation modes (active rectifier and grid-tie
ing the improved bidirectional cell, the losses in the diodes are inverter), a cost function to minimize the error between the
reduced to the half. Another advantage is the possibility of op- measured grid current and its reference is used [32]. According
erating as an active rectifier or as a grid-tie inverter. During the to the converter and the final application, it is important to note
operation as an active rectifier, the iFBC works with a sinusoidal that the cost function can also be used to minimize the switch-
grid current, in phase with the power grid voltage, and with a ing frequency, output voltage, torque, flux, and active or reactive
controlled output voltage. On the other hand, when operating power [33]. In the FCS-MPC a longer prediction horizon (more
as a grid-tie inverter the iFBC also works with a sinusoidal grid than one) and a delay compensation can also be considered
current, but in phase opposition with the power grid voltage. In [34]. As example, a new delay time compensation method for
both operation modes, the voltage produced by the iFBC (i.e., the MPC, applied to a voltage source inverter in order to control
the voltage between the Lf Cf passive filter and the IGBTs leg) an induction motor, is proposed in [35]. A review about the use
can assume five different levels (+Vdc , +Vdc /2, 0, −Vdc /2, and of FCS-MPC for power electronics applications is presented in
−Vdc ). Moreover, adding additional IGBTs and diodes to the [33], where it is shown that it can be successfully applied to ac–
iFBC the number of levels can be increased indefinitely. For a dc, ac–ac, dc–dc, and dc–ac converters. Taking into account its
topology with n distinct voltage levels, the number of IGBTs versatility, the MPC has been employed in a wide range of appli-
is n–1, the number of diodes is n–3, and the number of dc-link cations [36]–[38]. It can be applied to control the grid current in
capacitors is ((n + 1)/2)−1. voltage-source and current-source converters [39], [40], and for
The classical current control strategies for active rectifiers different applications, e.g., uninterruptible power supplies and
or grid-tie inverters are mainly based on hysteresis-band con- shunt active power filters [41], [42]. In the scope of this paper,
trollers and linear-control with pulse width modulation [27], the FCS-MPC was validated with a proposed improved bidirec-
[28]. However, nowadays, due to the advances in microcon- tional multilevel converter (iFBC) during the operation as an
trollers, increasingly, predictive schemes are proposed to control active rectifier and as a grid-tie inverter. The main contributions
power electronics converters. When compared with the classical of this paper are: An improved bidirectional cell associated with
current control strategies, predictive schemes are more intuitive a H-Bridge converter to perform a five-level converter (iFBC);
and allow to include the nonlinearities of the controlled system; the experimental validation of the iFBC, i.e., a converter capable
however, the digital implementation is more complex [29]. In of producing five distinct voltages (+Vdc , +Vdc /2, 0, −Vdc /2,
this context, several predictive control strategies can be imple- and −Vdc ); and the experimental validation of the FCS-MPC
mented, including dead-beat control, hysteresis based, trajectory applied to the iFBC during the operation as an active rectifier
based, and model predictive control (MPC), where the continu- (i.e., controlling the grid current and the dc-link voltage) and as
ous control set and the finite control set is included [29]. In order a grid-tie inverter (i.e., controlling the grid current).
to control the grid current of the iFBC, any of the aforementioned The rest of this paper is organized as follows. Section II
current control strategies can be used. However, using a mod- describes the principle of operation of the iFBC during the
ulator to obtain the five voltage levels is complex. Therefore, a operation as an active rectifier and as a grid-tie inverter. In
current control strategy that does not require a modulator should Section III the FCS-MPC is described in detail, i.e., the power
MONTEIRO et al.: MODEL PREDICTIVE CONTROL APPLIED TO AN IMPROVED FIVE-LEVEL BIDIRECTIONAL CONVERTER 5881

Fig. 2. Operation stages of the iFBC working as an active rectifier: (a)–(c) when v g > 0; (d)–(f) when v g < 0.

Fig. 3. Operation stages of the iFBC working as a grid-tie inverter: (a)–(c) when v g > 0; (d)–(f) when v g < 0.

theory, the predictive model, the cost function and the gate pulse voltage vcv can also assume three distinct values: 0, −Vdc /2
patterns. Section IV presents the developed iFBC prototype, and −Vdc . Analyzing all these cases, it is possible to identify
while in Section V the experimental validation of the FCS- five different voltage levels and two forms for obtaining the level
MPC applied to the iFBC in both operation modes is presented. 0 (which is useful for dividing the losses between the IGBTs g1 ,
Finally, in Section VI the main conclusions are presented. g2 , g3, and g4 ). Analyzing Table I, it is possible to observe that,
using the iFBC as an active rectifier it is necessary to use four
IGBTs (g3 and g5 when vg > 0, and g4 and g6 when vg < 0),
II. PRINCIPLE OF OPERATION OF THE IFBC and using the iFBC as a grid-tie inverter it is necessary to use six
This section presents the principle of operation of the iFBC. IGBTs (g1 , g3 , g4 , and g6 when vg > 0 and g2 , g3 , g4 , and g5
It is described in detail when the IGBTs are on (1) or off (0) when vg < 0). In both operation modes, the maximum voltage
during a specific sampling period, in accordance with the FCS- in the IGBTs g1 and g2 is Vdc , and in the IGBTs g3 , g4 , g5 , and
MPC. The different states of the iFBC operating as an active g6 is Vdc /2. The maximum voltage in the diodes d1 and d2 is
rectifier or as a grid-tie inverter are summarized in Table I. also Vdc /2.
Fig. 2 shows the equivalent circuits and the current paths when Fig. 4 shows some simulation results, obtained with PSIM
the iFBC is operating as an active rectifier. More specifically, v9.0, of the iFBC during the operation as an active rectifier.
Fig. 2(a)–(c) shows the current path when the power grid voltage Fig. 4(a) shows the power grid voltage vg and the grid current
is positive vg > 0, and Fig. 2(d)–(f) when is negative vg < 0. ig . In this figure, it is possible to observe that the grid current is
On the other hand, Fig. 3 shows the equivalent circuits and the increasing without sudden variations. In this case, the rms value
current paths when the iFBC is operating as a grid-tie inverter. of the grid current increases from 9 to 17 A. Fig. 4(b) shows
More specifically, Fig. 3(a)–(c) shows the current path when the grid current ig and its reference i∗g in detail. This simulation
vg > 0, and Fig. 3(d)–(f) when vg < 0. result shows that the grid current ig tracks its reference i∗g very
When the power grid voltage is positive vg > 0, the voltage well even with the variable switching frequency.
produced by the iFBC vcv can assume three distinct values: Fig. 5 shows some simulation results of the iFBC during the
0, +Vdc /2 and +Vdc . On the other hand, when vg < 0, the operation as a grid-tie inverter. Fig. 5(a) shows the power grid
5882 IEEE TRANSACTIONS ON INDUSTRIAL ELECTRONICS, VOL. 63, NO. 9, SEPTEMBER 2016

Fig. 4. Simulation results during the iFBC operation as an active recti-

fier: (a) power grid voltage v g ; grid current ig ; (b) detail of the grid current Fig. 6. Simulation results during the operation as an active rectifier:
ig and its reference i∗g . (a) grid current ig with step changes in the load; (b) detail of the current
reference i∗g and the grid current ig using the iFBC and the H-Bridge
converter.

switching frequency. In the first sudden variation the measured

time delay was 0.21 ms and in the second 0.13 ms.
Fig. 6 shows some simulation results during the operation as
an active rectifier. Fig. 6(a) shows the grid current ig with step
changes in the load. In the first period a load of 1000 W was
used, in the second period a load of 400 W, in the third period
a load of 800 W, in the fourth period a load of 200 W, and in
the last period a load of 600 W. In order to show the advantages
of the iFBC, it was also simulated the H-Bridge converter for
the same conditions. Fig. 6(b) shows in detail the grid current ig
using the iFBC and the H-Bridge converter in comparison with
the reference i∗g . As it can be seen, the grid current produced
by the iFBC is more accurate than the current produced by the
H-Bridge converter. For the operation power values of 200, 400,
600, 800, and 1000 W the rms error of the grid current using the
iFBC was, respectively, 0.6%, 0.3%, 0.1%, 0.09%, and 0.01%.
Using the H-Bridge it was, respectively, 1.66%, 0.68%, 0.68%,
Fig. 5. Simulation results during the iFBC operation as a grid-tie in- 0.2%, and 0.2%.
verter: (a) power grid voltage v g ; grid current ig ; (b) detail of the grid
current ig and its reference i∗g .
The measured THD of the grid current using the iFBC and the
H-Bridge converter during the operation as an active rectifier and
as grid-tie inverter for the aforementioned five periods is shown
in Table II. The measured THD is lower using the iFBC for all
voltage vg and the grid current ig . In this simulation result a
the operation power values, and for both operation modes, as an
dc–dc boost-type converter was used to interface a solar photo-
active rectifier or as a grid-tie inverter.
voltaic system, which is responsible for controlling the dc-link
voltage and for defining the operation power value of the iFBC.
Due to the power variation, it is possible to observe that the III. MODEL PREDICTIVE CONTROL
injected grid current has sudden variations. In the first variation In this section, the FCS–MPC applied to the iFBC is de-
the instantaneous value of the reference current i∗g changes from scribed in detail. It is important to refer that this analysis is
1.4 to 7.2 A, and in the second it changes from 7.2 to 4.3 A. valid for the iFBC in both operation modes, i.e., operating as
Fig. 5(b) shows in detail the grid current ig and its reference i∗g . an active rectifier or as a grid-tie inverter. During the operation
This simulation result shows that the grid current ig tracks its as an active rectifier, the iFBC is used to control the grid cur-
reference i∗g very well with a low time delay even with variable rent and the dc-link voltage, i.e., a power theory is necessary in
MONTEIRO et al.: MODEL PREDICTIVE CONTROL APPLIED TO AN IMPROVED FIVE-LEVEL BIDIRECTIONAL CONVERTER 5883

TABLE II
THD COMPARISON BETWEEN THE IFBC AND THE H-BRIDGE CONVERTER

Active rectifier Grid-tie inverter

Power iFBC H-Bridge iFBC H-Bridge

200 W 5.71% 14.6% 6.41% 12.4% Fig. 7. Representation of the voltages and currents between the power
400 W 2.87% 7.04% 3.62% 6.39% grid and the iFBC.
600 W 1.95% 4.57% 2.29% 4.36%
800 W 1.49% 3.33% 1.86% 3.46%
1000 W 1.48% 2.72% 1.48% 2.64%
grid current reference i∗g is directly influenced by the waveform
of the power grid voltage vg . Taking into account that the power
order to obtain the instantaneous value of the grid current ref- grid voltage can have harmonic content, a phase-locked loop
erence. On the other hand, when the iFBC is used as a grid-tie algorithm is used in order to obtain a grid current reference pro-
inverter, the value of the grid current reference is provided to the portional only to the fundamental component of the power grid
digital control, e.g., from a dc–dc back-end converter used to voltage [46]. Therefore, the grid current reference is sinusoidal
interface renewable energy sources. In this operation mode the even when a distorted power grid voltage is present. The dc-link
dc-link voltage is controlled by the dc–dc back-end converter voltages (Vdc1 and Vdc2 ) are controlled through a PI controller.
and, therefore, it is not necessary to use the power theory for These voltages are controlled to the same voltage (85 V in each
obtaining the grid current reference. In terms of the FCS-MPC, capacitor) independently of the iFBC operation power value.
the main difference between both the operation modes is the
selection of the gate pulse patterns. B. Predictive Model
The predictive model is based on the circuit equations and is
A. Power Theory aimed to predict the grid current as a function of its actual and
The power theory is used for obtaining the grid current previous values. Analyzing the voltages and currents identified
reference i∗g , i.e., a signal that is directly proportional to the in Fig. 7, it can be established that
power grid voltage vg to maintain a unitary power factor [43], vg = vL f 1 + vL f 2 + vcv (5)
[44]. The Fryze–Buchholz–Depenbrock method, proposed in
[45], consists in considering the power converter as a conduc- ig = iC f + iL f 1 . (6)
tance from the power grid point of view. Therefore, it can be From (6), substituting the current in the Cf filter, (5) can be
established: rewritten as
i∗g = Gvg (1) dig d 2 vg
vg = Lf − Lf Cf 2 + vcv (7)
dt dt
where G denotes a conductance that represents the iFBC. Using
the rms values of the power grid voltage VG and the grid current where Lf = Lf 1 + Lf 2 . Taking into account that a digital con-
IG , the conductance G can be defined by troller is used, the discrete implementation of (7) using the
forward Euler method is
IG
G= . (2) ig [k + 1] − ig [k]
VG vg [k] = Lf + vcv [k]
T
Taking into account that the grid current ig is in phase (oper-
vg [k + 1] − 2vg [k] + vg [k − 1]
ation as an active rectifier), or in phase opposition (operation as − Lf Cf . (8)
a grid-tie inverter), with the power grid voltage vg , (2) can be T2
rewritten in terms of active power PG This method can be applicable to discretize (7) [29]. However,
the backward Euler method can also be applied for the same
PG
G= . (3) purpose [30]. Rewriting (8) in terms of the predict current, i.e.,
VG2 the current in the instant [k + 1] is obtained
Neglecting the losses, this active power corresponds to the T
power in the dc side Pdc and the power PC to maintain the dc- ig [k + 1] = ig [k] + (vg [k] − vcv [k])
Lf
link voltage controlled. Substituting (3) in (1), the grid current
reference i∗g is defined by +
Cf
(vg [k + 1] − 2vg [k] + vg [k − 1]) (9)
T
PG
i∗g = vg . (4) where the predicted power grid voltage in the [k + 1] can be
VG2 obtained from [39]
The reference current determined by (4) changes according
vg [k + 1] = 3vg [k] − 3vg [k − 1] + vg [k − 2] . (10)
to the operation power value PG . Therefore, the reference cur-
rent is dynamically adjusted according to the operation mode. Substituting (10) in (9) the equation that allows us to predict
Analyzing (4), it is possible to infer that the waveform of the the grid current in the instant [k + 1] is obtained.
5884 IEEE TRANSACTIONS ON INDUSTRIAL ELECTRONICS, VOL. 63, NO. 9, SEPTEMBER 2016

Fig. 8. Example of the state selection for minimizing the grid current
error.

C. Cost Function
After obtaining the grid current reference and the predicted
grid current, the final stage consists in using a cost function for
minimizing the grid current error, i.e., the difference between Fig. 9. Bode plot of the second-order low-pass L f C f passive filer
characteristic with passive damping.
both. The grid current reference in the instant [k + 1] can be
extrapolated from the previous samples with the same reasoning
of (10). The cost function used in this paper is defined by
 
g [k + 1] = i∗g [k + 1] − ig [k + 1] . (11) gate driver through a command board. In this board a protection
circuit is also included, and it consists in comparing the mea-
Taking into account that only one of the variables is con- sured voltages and currents with the predefined references. The
trolled, no weighting factors are necessary for this cost func- measured voltages and currents are acquired with a sampling
tion, as well as no tuning processes. This type of cost function frequency of 40 kHz. For such purpose, a timer programmed to
is very common when is required to control the grid current of obtain interruptions at 25 µs is used. Therefore, considering that
voltage-source converters [47], [39]. During each sampling pe- in the maximum the states of the IGBTs will change from one
riod (25 µs), the gate pulse patterns for the IGBTs are defined. interruption to another, then the maximum switching frequency
Therefore, the state of the iFBC during such sampling period is is limited to 20 kHz. In the digital controller, a digital-to-analog
defined in order to minimize the grid current error. The error is converter (DAC TLV5610 from Texas Instruments) is also used
zero when the cost function is zero. in order to visualize the digital values in an oscilloscope.

D. Gate Pulses Patterns

B. Power Converter
The gate pulses pattern is selected according to the iFBC
operation mode (active rectifier or grid-tie inverter) and the The iFBC uses the IGBTs IXXR110N65B4H1 from IXYS
result of the cost function. Fig. 8 shows a simulation result of and the diodes IR HFA15PB60 from International Rectifier.
an example of the state selection. In this situation, the smallest The dc-link is composed of ten capacitors, each one with a
error Δig 2 is obtained when the state of the iFBC defined by nominal value of 560 µF/450 V. The gate drivers are composed
{0, 0, 0, 0, 1, 0} is selected. Therefore, during the time interval of optocouplers HCPL3120 from Avago Technologies and iso-
[k, k + 1] only the IGBT g5 is ON. lated dc–dc sources NMV1515SC from Murata. The iFBC is
connected to the power grid through a second-order low-pass
IV. DEVELOPED PROTOTYPE Lf Cf passive filter, which is composed of an inductor of 3 mH
and a capacitor of 3 µF.
In this section, the developed prototype of the iFBC is de-
The capacitor is split in two (2 and 1 µF) and it is used a
scribed in detail, i.e., the digital controller and the power
damping resistor of 120 Ω in series with the capacitor of 2 µF
converter.
in order to smooth the gain response of the passive filter at the
cutoff frequency. The bode plot of this passive filter is shown in
A. Digital Controller Fig. 9. The cutoff frequency of this filter is around 1.6 kHz.
The digital controller is mainly based on the DSP In the inductor design, two cores T300-60D from MI-
TMS320F28335 from Texas Instruments, where the aforemen- CROMETALS were used. Analyzing (9), it is possible to ob-
tioned FCS-MPC is implemented. The digital controller is also serve that the predicted current is influenced by the inductor
composed of a signal conditioning board to convert the analog value. Using a fixed value for the inductor, the control system
voltages and currents for the digital values used in the DSP. can select a state for the iFBC that effectively does not cor-
In this board, the ADC MAX1320 from MAXIM is used. The respond to the state that minimizes the error between the grid
voltages are measured with sensors LV-25 P and the currents current and its reference. In order to contribute to the stability
with sensors LA-55 P, both from LEM. The output signals from of the digital control system, instead of using a fixed value for
the DSP (gate pulse patterns) are adjusted to input signals of the the inductor, it is used the real value of the inductor in function
MONTEIRO et al.: MODEL PREDICTIVE CONTROL APPLIED TO AN IMPROVED FIVE-LEVEL BIDIRECTIONAL CONVERTER 5885

TABLE III
SPECIFICATIONS OF THE EXPERIMENTAL SETUP

Parameters Value Unit

Power grid voltage 115 V

Grid frequency 50 Hz
Maximum output power 1000 W
Dc-link voltage 170 V
Total power factor at full Load 0.99 –
THDi% at Full load 1.4 –
Sampling frequency 40 kHz
Maximum switching frequency 20 kHz

TABLE IV Fig. 10. Experimental setup used during the experimental validation of
TIME REQUIRED BY THE MAIN TASKS OF THE CONTROL ALGORITHM the iFBC controlled with FCS-MPC.

Tasks Value Unit

ADCs reading 690 ns

PLL synchronization 820 ns
Dc-link voltages digital filtering 1800 ns
Power theory 450 ns
Current reference prediction 820 ns
Cost function 500 ns
Gate pulses 800 ns

of the grid current, described by

 
L = 0.001 0.2i3g − 7.5i2g − 96.2ig + 7.59. (12)
Therefore, the equation (12) is used in (9) in order to intro-
duce the nonlinearity of the inductor in the FCS-MPC. Besides
the effect of the inductor value variation, other criteria can be
considered for the stability analysis. As example, the analysis
of the stability and performance of the MPC applied to an active
rectifier is presented in detail in [48].

V. EXPERIMENTAL VALIDATION
Fig. 11. Experimental results of the power grid voltage (v g : 1 V/div),
In this section, the main experimental results obtained with grid current (ig : 500 mA/div) and voltage produced by the iFBC (v c v :
the iFBC are presented. These results were acquired with a 50 V/div) operating as an active rectifier.
digital Yokogawa DL708E digital oscilloscope, with a Fluke
435 Power Quality Analyzer, and with a current probe Tek-
tronix AM5031. The specifications of the experimental setup
voltage vg , therefore the iFBC operates with a unitary power
are shown in Table III. Table IV shows the time required by the
factor. It is important to note that these results are not exactly
main tasks of the digital control algorithm. These times were
as the results shown in Fig. 8, because the observations were in
measured using one of the features available in the Code Com-
different instants and in different contexts. Moreover, as it can
poser Studio from Texas Instruments. The experimental setup
be seen, the grid current does not follow exactly a simple ramp
used during the experimental validation of the iFBC as an active
form due to the nonlinearity of the second-order low-pass Lf Cf
rectifier and as a grid-tie inverter is shown in Fig. 10.
passive filter. In this specific case, i.e., during the transition from
the positive to the negative semicycle, the minimum ripple in the
A. iFBC Operating as an Active Rectifier grid current is 688 mA and the switching frequency is 6.944 kHz
In this item, the experimental results of the iFBC operating (about 35% of the maximum switching frequency). In Fig. 11 it
as an active rectifier are presented. For such purpose the iFBC is possible to observe that when vg > 0, during the state {0, 0,
was connected to the power grid (115 V) and the dc-link voltage 0, 0, 1, 0} the voltage vcv is +Vdc /2, and during the state {0, 0,
was controlled to a voltage of 170 V, i.e., 85 V in each capacitor. 1, 0, 0, 0}, the voltage vcv is 0. On the other hand, when vg < 0,
Fig. 11 shows the grid current ig , the power grid voltage vg , and during the state {0, 0, 0, 0, 0, 1} the voltage vcv is −Vdc /2, and,
the voltage produced by the iFBC vcv (i.e., the voltage between during the state {0, 0, 0, 1, 0, 0}, the voltage vcv is 0.
the Lf Cf passive filter and the IGBTs leg) in detail. As it can Fig. 12 shows in a detail of 50 ms, the power grid voltage
be observed the grid current ig is in phase with the power grid vg , the grid current ig , the voltage of the iFBC vcv , and both
5886 IEEE TRANSACTIONS ON INDUSTRIAL ELECTRONICS, VOL. 63, NO. 9, SEPTEMBER 2016

Fig. 12. Experimental results of the power grid voltage (v g : 50 V/div), Fig. 13. Experimental results of the power grid voltage (v g : 50 V/div)
grid current (ig : 5 A/div), voltage produced by the iFBC (v c v : 50 V/div), and peak-to-peak voltages in the dc-link (V d c : 20 V/div) for the iFBC
and dc-link voltages (V d c : 20 V/div) for operation as an active rectifier. operating as an active rectifier.

dc-link voltages (Vdc1 and Vdc2 ). In this experimental result, it

is possible to confirm that the grid current is sinusoidal even
when the power grid voltage has harmonic content. It is also
possible to confirm the five distinct voltages of the iFBC, i.e.,
+Vdc , +Vdc /2, 0, −Vdc /2, and −Vdc . In this experimental re-
sult, it is possible to observe that the dc-link voltages (Vdc1 and
Vdc2 ) are balanced with a maximum peak-to-peak voltage of
4.8 V, i.e., about 5% of the dc-link voltage in each capacitor.
This relation can be improved increasing the capacitance of the
dc-link capacitors. If the dc-link voltages (Vdc1 and Vdc2 ) are
imbalanced, then the voltage produced by the converter vcv will
be directly influenced. It should be mentioned that the dc-link
Fig. 14. Spectral analysis and THD% of the grid current for a power
voltages are not used directly in the PI controller to obtain the of 980 W (grid current of 8.5 A) with the iFBC operating as an active
grid current references, but its mean value. The dc-link volt- rectifier.
ages (Vdc1 and Vdc2 ) are acquired with a sampling frequency of
40 kHz and then a digital filter is used to obtain the mean value
of these voltages. Therefore, the influence of the peak-to-peak ig , i.e., a variation from 0 to 10 A during 0.14 s (a first stage
voltage is avoided to obtain the grid current reference. (1) that corresponds to seven cycles of the power grid voltage).
Fig. 13 shows the power grid voltage and the peak-to-peak In this situation, the iFBC increases the grid current ig without
voltage in each dc-link capacitor. The voltage in each capacitor sudden variations in the current, which is favorable to the power
is controlled independently according to the power grid voltage quality maintenance. As it can be seen, the grid current ig is
vg . If the power grid voltage is positive, then the voltage Vdc1 is in phase with the power grid voltage vg during all the transient
adjusted. If the power grid voltage is negative, then the voltage response.
Vdc2 is adjusted. Fig. 14 shows the spectral analysis and the During the operation as an active rectifier, the measured effi-
THD% (1.4%) of the grid current for a power of 980 W, i.e., for ciency and the grid current THD% for a range of operation from
a grid current of 8.5 A. Fig. 15 shows the variation of the grid 200 to 1000 W are presented in Fig. 17(a) and (b), respectively.
current ig as a function of the power grid current voltage vg . As The maximum measured efficiency was 96.2% at 395 W and
it can be seen, the grid current varies almost linearly with the the minimum efficiency was 92.7% at 205 W. The efficiency
power grid voltage. Due to the harmonic content of the power of the iFBC as an active rectifier was measured using the afore-
grid voltage (THD = 2.4%), the variation is more critical during mentioned power quality analyzer and the current probe. For
the maximum amplitude of the power grid voltage and the grid such purpose, the iFBC was connected to the power grid to
current. Fig. 16 shows the transient response of the grid current a voltage of 115 V, the dc-link voltage was controlled to a
MONTEIRO et al.: MODEL PREDICTIVE CONTROL APPLIED TO AN IMPROVED FIVE-LEVEL BIDIRECTIONAL CONVERTER 5887

Fig. 15. Experimental result of the grid current (ig : 5 A/div) and power
grid voltage (v g : 50 V/div) in X–Y mode for the iFBC as an active rectifier.
Fig. 17. Experimental results of iFBC for operation as an active recti-
fier: (a) measured efficiency; (b) measured grid current THD%.

Fig. 16. Experimental results during transient of the power grid voltage
(v g : 50 V/div) and grid current (ig : 5 A/div) for operation as an active
rectifier.
Fig. 18. Experimental results of the power grid voltage (v g : 50 V/div),
grid current (ig : 5 A/div), voltage produced by the iFBC (v c v : 50 V/div),
and dc-link voltages (V d c : 20 V/div) for operation as a grid-tie inverter.

voltage of 170 V, and a variable resistive load was used to

change the iFBC operation power value. In terms of the grid
current THD%, the maximum value registered was 4.2% (at Fig. 18 shows the grid current ig , the power grid voltage
205 W) and the minimum was 1.4% (at 980 W). vg , and the voltage produced by the iFBC vcv in detail. In
this operation mode, the grid current ig is in phase opposition
with the power grid voltage vg . Also in this operation mode,
B. iFBC Operating as a Grid-Tie Inverter it is possible to confirm that the grid current is sinusoidal even
In this item, the experimental results of the iFBC operating when the power grid voltage has harmonic content. It is also
as a grid-tie inverter are presented. For such purpose, the ac side possible to confirm the five distinct voltages of the iFBC, i.e.,
of the iFBC was connected to the power grid (115 V) and the +Vdc , +Vdc /2, 0, −Vdc /2, and −Vdc . Fig. 19 shows the tran-
dc-link was connected to a dc voltage source of 170 V (85 V in sient response of the grid current ig , i.e., from the first to the
each capacitor). second stage it corresponds to a variation from 3 to 10 A and
5888 IEEE TRANSACTIONS ON INDUSTRIAL ELECTRONICS, VOL. 63, NO. 9, SEPTEMBER 2016

was measured using the same aforementioned description. The

iFBC was connected to the power grid to a voltage of 115 V,
the dc-link voltage was controlled to a voltage of 170 V, and a
variable power reference was used to change the iFBC operation
power value. In terms of the grid current THD%, the maximum
value registered was 4.6% (at 210 W) and the minimum was
1.8% (at 1010 W).

VI. CONCLUSION
This paper has presented an improved five-level bidirectional
converter (iFBC) controlled by finite control set model predic-
tive control (FCS-MPC). The paper presented in detail the iFBC
hardware, the principle of operation, the power theory used for
obtaining the grid current reference, the FCS-MPC, and the
cost function for minimizing the error between the grid current
and its reference. The experimental results, in the steady state
and during transient operation, were obtained with the iFBC
connected to the power grid and operating both as an active
rectifier and as a grid-tie inverter. For both operation modes,
Fig. 19. Experimental results during transient of the power grid voltage the efficiency and the THD% were verified for different oper-
(v g : 50 V/div) and grid current (ig : 5 A/div) for operation as a grid-tie
inverter. ation power values. The experimental results confirm that the
FCS-MPC can be applied with success (in terms of efficiency,
low grid current THD, and controlled output voltage) to the
iFBC.

REFERENCES
[1] B. Singh, B. N. Singh, A. Chandra, K. Al-Haddad, A. Pandey, and
D. P. Kothari, “A review of single-phase improved power quality AC-
DC converters,” IEEE Trans. Ind. Electron., vol. 50, no. 5, pp. 962–981,
Oct. 2003.
[2] O. Garcı́a, J. A. Cobos, R. Prieto, P. Alou, and J. Uceda, “Single phase
power factor correction: A survey,” IEEE Trans. Power Electron., vol. 18,
no. 3, pp. 749–755, May 2003.
[3] F. Beltrame, L. Roggia, L. Schuch, J. R. Pinheiro, “A Comparison of high
power single-phase power factor correction pre-regulators,” in Proc. IEEE
Int. Conf. Ind. Technol., Mar. 2010, pp. 625–630.
[4] M. Narimani and G. Moschopoulos, “A new single-phase single-stage
three-level power-factor-correction — AC-DC converter with phase-shift
modulation,” IEEE Trans. Ind. Electron., vol. 60, no. 9, pp. 3731–3735,
Sep. 2013.
[5] X. Liu, J. Xu, Z. Chen, and N. Wang, “Single-inductor dual-output buck–
boost power factor correction converter,” IEEE Trans. Ind. Electron.,
vol. 62, no. 2, pp. 943–952, Feb. 2015.
[6] X. Xie, C. Zhao, Q. Lu, and S. Liu, “A novel integrated buck–flyback
nonisolated PFC converter with high power factor,” IEEE Trans. Ind.
Electron., vol. 60, no. 12, pp. 5603–5612, Dec. 2013.
[7] A. De Bastiani Lange, T. B. Soeiro, M. S. Ortmann, and M. L. Heldwein,
Fig. 20. Experimental results of iFBC during the operation as a grid-tie “Three-level single-phase bridgeless PFC rectifiers,” IEEE Trans. Power
inverter: (a) measured efficiency; (b) measured grid current THD%. Electron., vol. 30, no. 6, pp. 2935–2949, Jun. 2015.
[8] L. Huber, Y. Jang, and M. Jovanovic, “Performance evaluation of bridge-
less PFC boost rectifier,” IEEE Trans. Power Electron., vol. 23, no. 3,
pp. 1381–1390, May 2008.
from the second to the third stage to a variation from 10 to 7 [9] R. Martinez and P. N. Enjeti, “A high-performance single-phase rectifier
A. In this situation, the grid current ig changes instantaneously with input power factor correction,” IEEE Trans. Power Electron., vol. 11,
no. 2, pp. 311–317, Mar. 1996.
between the stages. As it can be seen, the grid current ig is in [10] J.-W. Lim and B.-H. Kwon, “A power-factor controller for single-phase
phase opposition with the power grid voltage vg in both the PWM rectifiers,” IEEE Trans. Ind. Electron., vol. 46, no. 5, pp. 1035–1037,
stages. During the operation as a grid-tie inverter, the measured Oct. 1999.
[11] S. Dusmez, S. Choudhury, M. Bhardwaj, and B. Akin, “A modified dual-
efficiency and the grid current THD% for a range of operation output interleaved PFC converter using single negative rail current sense
from about 200 to 1000 W are presented in Fig. 20(a) and (b), for server power systems,” IEEE Trans. Power Electron., vol. 29, no. 10,
respectively. In this operation mode, the maximum measured pp. 5116–5123, Oct. 2014.
[12] M. Narimani and G. Moschopoulos, “A novel single-stage multilevel type
efficiency was 95.2% at 420 W and the minimum efficiency was full-bridge converter,” IEEE Trans. Ind. Electron., vol. 60, no. 1, pp. 31–
92.9% at 210 W. The efficiency of the iFBC as a grid-tie inverter 42, Jan. 2013.
MONTEIRO et al.: MODEL PREDICTIVE CONTROL APPLIED TO AN IMPROVED FIVE-LEVEL BIDIRECTIONAL CONVERTER 5889

[13] J. Rodrı́guez, J.-S. Lai, and F. Z. Peng, “Multilevel inverters: A survey [35] M. Uddin, S. Mekhilef, M. Nakaoka, and M. Rivera, “Model predictive
of topologies, controls, and applications,” IEEE Trans. Ind. Electron., control of induction motor with delay time compensation: An experimental
vol. 49, no. 4, pp. 724–738, Aug. 2002. assessment,” in Proc. IEEE Appl. Power Electron. Conf. Expo., Mar. 2015,
[14] J.-S. Lai and F. Z. Peng, “Multilevel converters—A new breed of power pp. 543–548.
converters,” IEEE Trans. Ind. Appl., vol. 32, no. 3, pp. 509–517, May/Jun. [36] J. Scoltock, T. Geyer, and U. K. Madawala, “Model predictive direct power
1996. control for grid-connected NPC converters,” IEEE Trans. Ind. Electron.,
[15] G. H. P. Ooi, A. I. Maswood, and Z. Lim, “Five-level multiple-pole PWM vol. 62, no. 9, pp. 5319–5328, Sep. 2015.
AC–AC converters with reduced components count,” IEEE Trans. Ind. [37] M. D. P. Akter, S. Mekhilef, N. M. L. Tan, and H. Akagi, “Modified model
Electron., vol. 62, no. 8, pp. 4739–4748, Aug. 2015. predictive control of a bidirectional AC-DC converter based on Lyapunov
[16] R. S. Alishah, D. Nazarpour, S. H. Hosseini, and M. Sabahi, “Novel topolo- function for energy storage systems,” IEEE Trans. Ind. Electron., vol. 63,
gies for symmetric, asymmetric, and cascade switched-diode multilevel no. 2, pp. 704–715, Feb. 2016.
converter with minimum number of power electronic components,” IEEE [38] D.-K. Choi and K.-B. Lee, “Dynamic performance improvement of
Trans. Ind. Electron., vol. 61, no. 10, pp. 5300–5310, Oct. 2014. AC/DC converter using model predictive direct power controlwith finite
[17] C. Buccella, C. Cecati, M. G. Cimoroni, and K. Razi, “Analytical method control set,” IEEE Trans. Ind. Electron., vol. 62, no. 1, pp. 757–767,
for pattern generation in five-level cascaded h-bridge inverter using selec- Feb. 2015.
tive harmonic elimination,” IEEE Trans. Ind. Electron., vol. 61, no. 11, [39] J. Rodrı́guez, et al., “Predictive current control of a voltage source in-
pp. 5811–5819, Nov. 2014. verter,” IEEE Trans. Ind. Electron., vol. 54, no. 1, pp. 495–503, Feb. 2007.
[18] M. Calais, L. J. Borle, and V. G. Agelidis, “Analysis of multicarrier PWM [40] P. Zavala, et al., “Predictive control of a current source rectifier with
methods for a single-phase five-level inverter,” in Proc. 32nd Annu. IEEE imposed sinusoidal input currents,” in Proc. 39th Annu. Conf. IEEE Ind.
Power Electron. Spec. Conf., Jun. 2001, vol. 3, pp. 1351–1356. Electron. Soc., Vienna Austria, Nov. 2013, pp. 5842–5847.
[19] C. A. Teixeira, D. G. Holmes, and B. P. McGrath, “Single-phase [41] P. Cortés, G. Ortiz, J. I. Yuz, J. Rodrı́guez, S. Vazquez, and L. G. Franquelo,
semi-bridge five-level flying-capacitor rectifier,” IEEE Trans. Ind. Appl., “Model predictive control of an inverter with output LC filter for UPS
vol. 49, no. 5, pp. 2158–2166, Sep./Oct. 2013. applications,” IEEE Trans. Ind. Electron., vol. 56, no. 6, pp. 1875–1883,
[20] P. Roshankumar, P. P. Rajeevan, K. Mathew, K. Gopakumar, Jose I. Leon, Jun. 2009.
and L. G. Franquelo, “A five-level inverter topology with single-DC supply [42] P. Zanchetta, P. Cortes, M. Perez, J. Rodriguez, and C. Silva, “Finite states
by cascading a flying capacitor inverter and an h-bridge,” IEEE Trans. model predictive control for shunt active filters,” in Proc. IEEE IECON
Power Electron., vol. 27, no. 8, pp. 3505–3512, Aug. 2012. Ind. Electron. Conf., pp. 581–586, Nov. 2011.
[21] Y. Zhang and L. Sun, “An efficient control strategy for a five-level inverter [43] F. Blaabjerg, R. Teodorescu, M. Liserre, and A. V. Timbus, “Overview
comprising flying-capacitor asymmetric h-bridge,” IEEE Trans. Ind. Elec- of control and grid synchronization for distributed power generation
tron., vol. 58, no. 9, pp. 4000–4009, Sep. 2011. systems,” IEEE Trans. Ind. Electron., vol. 53, no. 5, pp. 1398–1409,
[22] S. Camur, B. Arifoglu, E. Kandemir Beser, and E. Beser, “A Novel topol- Oct. 2006.
ogy for single-phase five-level inverter compared with h-bridge inverter,” [44] A. A. Valdez-Fernández, P. R. Martı́nez-Rodrı́guez, G. Escobar,
in Proc. Int. Symp. Power Electron., Elect. Drives, Autom. Motion, May C. A. Limones-Pozos, and J. M. Sosa, “A model-based controller for
2006, pp. 556–560. the cascade h-bridge multilevel converter used as a shunt active filter,”
[23] K. K. Gupta and S. Jain, “A novelmultilevel inverter based on switched IEEE Trans. Ind. Electron., vol. 60, no. 11, pp. 5019–5028, Nov. 2013.
dc sources,” IEEE Trans. Ind. Electron., vol. 61, no. 7, pp. 3269–3278, [45] M. Depenbrock, “The FBD-method, a generally applicable tool for analyz-
Jul. 2014. ing power relations,” IEEE Trans. Power Syst., vol. 8, no. 2, pp. 381–387,
[24] H. Vahedi, P.-A. Labbé, H. Y. Kanaan, H. F. Blanchette, and K. Al-Haddad, May 1993.
“A new five-level buck-boost active rectifier,” in Proc. IEEE Int. Conf. Ind. [46] M. Karimi-Ghartemani, “Linear and pseudolinear enhanced phased-
Technol., Mar. 2015, pp. 2559–2564. locked loop (EPLL) structures,” IEEE Trans. Ind. Electron., vol. 61,
[25] R. A. Madhukar and K. Sivakumar, “A fault-tolerant single-phase five- no. 3, pp. 1464–1474, Mar. 2014.
level inverter for grid-independent PV systems,” IEEE Trans. Ind. Elec- [47] L. Tarisciotti, P. Zanchetta, A. Watson, S. Bifaretti, and J. C. Clare, “Mod-
tron., vol. 62, no. 12, pp. 7569–7577, Dec. 2015. ulated model predictive control for a seven-level cascaded h-bridge back-
[26] S.-J. Park, F.-S. Kang, M. H. Lee, and C.-U. Kim, “A new single-phase to-back converter,” IEEE Trans. Ind. Electron., vol. 61, no. 10, pp. 5375–
five-level PWM inverter employing a deadbeat control scheme,” IEEE 5383, Oct. 2014.
Trans. Power Electron., vol. 18, no. 3, pp. 831–843, May 2003. [48] M. D. P. Akter, S. Mekhilef, N. M. L. Tan, and H. Akagi, “Stability and
[27] L. A. Serpa, “Current control strategies for multilevel grid connected in- performance investigations of model predictive controlled active-front-
verters,” Ph.D. dissertation, Dept. Inf. Technol. Elect. Eng., Swiss Federal end (AFE) rectifiers for energy storage systems,” J. Power Electron.,
Inst. Technol., Zurich, Switzerland, 2007. vol. 15, no. 1, pp. 202–215, Jan. 2015.
[28] M. P. Kazmierkowski and L. Malesani, “Current control techniques for
three-phase voltage-source PWM converters: A survey,” IEEE Trans. Ind.
Electron., vol. 45, no. 5, pp. 691–703, Oct. 1998.
[29] P. Cortés, M. P. Kazmierkowski, R. M. Kennel, D. E. Quevedo, and
J. Rodrı́guez, “Predictive control in power electronics and drives,” IEEE
Trans. Ind. Electron., vol. 55, no. 12, pp. 4312–4324, Dec. 2008.
[30] V. Yaramasu, M. Rivera, B. Wu, and J. Rodriguez, “Model predictive
current control of two-level four-leg inverters—Part I: Concept, algorithm,
and simulation analysis,” IEEE Trans. Power Electron., vol. 28, no. 7, Vı́tor Monteiro (S’10) was born in Guimarães,
pp. 3459–3468, Jul. 2013. Portugal, in 1984. He received the M.Sc. de-
[31] M. Rivera, V. Yaramasu, J. Rodriguez, and B. Wu, “Model predictive gree in industrial electronics and computer
current control of two-level four-leg inverters—Part II: Experimental engineering from the School of Engineering,
implementation and validation,” IEEE Trans. Power Electron., vol. 28, University of Minho, Guimarães, Portugal, in
no. 7, pp. 3469–3478, Jul. 2013. 2012, where he is currently working toward the
[32] S. Kouro, P. Cortés, R. Vargas, U. Ammann, and J. Rodrı́guez, “Model Ph.D. degree supported by the doctoral scholar-
predictive control—A simple and powerful method to control power ship SFRH/BD/80155/2011 granted by the Por-
converters,” IEEE Trans. Ind. Electron., vol. 56, no. 6, pp. 1826–1838, tuguese FCT agency.
Jun. 2009. Since 2008, he has been a member of the
[33] J. Rodriguez et al., “State of the art of finite control set model predictive Group of Energy and Power Electronics (GEPE)
control in power electronics,” IEEE Trans. Ind. Informat., vol. 9, no. 2, of the Centro Algoritmi at the University of Minho. His research interests
pp. 1003–1016, May 2013. include power electronics converters, digital control techniques, smart
[34] P. Cortes, J. Rodriguez, C. Silva, and A. Flores, “Delay compensation grids, and electric vehicles.
inmodel predictive current control of a three-phase inverter,” IEEE Trans. Mr. Monteiro is a member of the IEEE Industrial Electronics and IEEE
Ind. Electron., vol. 59, no. 2, pp. 1323–1325, Feb. 2012. Vehicular Technology Societies.
5890 IEEE TRANSACTIONS ON INDUSTRIAL ELECTRONICS, VOL. 63, NO. 9, SEPTEMBER 2016

João C. Ferreira (M’01–SM’15) received the João Luiz Afonso (M’00) was born in Rio de
Graduate degree in physics, the M.Sc. degree Janeiro, Brazil, in 1963. He received the B.S.
in telecommunications, and the Ph.D. degree in and M.Sc. degrees in electrical engineering from
computer science engineering from the Techni- the Federal University of Rio de Janeiro, Rio de
cal University of Lisbon (UTL/IST), Lisbon, Por- Janeiro, Brazil, in 1986 and 1991, respectively,
tugal, in 1991, 1995, and 2006, respectively. and the Ph.D. degree in industrial electronics
He is a Professor at the ISCTE–Instituto Uni- from the University of Minho, Guimarães, Portu-
versitário de Lisboa and a Consultant to various gal, in 2000.
companies and institutions. His professional and Since 1993, he has been with the Department
research interests include geographic and multi- of Industrial Electronics, University of Minho,
media retrieval, electric vehicles, intelligent sys- where he is an Associate Professor. He teaches
tems, intelligent transportation systems (ITS), and sustainable mobility courses electrical machines, electrical energy systems, complements of
systems. He is the author of more than 140 scientific papers presented power electronics, electrical power quality, active power filters, and re-
at international conferences and workshops on different areas of com- newable energy. He is a Researcher with the Group of Energy and Power
puter science. Electronics, and he coordinates the thematic strand of Sustainable and
Smart Cities of the Centro Algoritmi. His research interests include power
quality, active power filters, renewable energy, electric vehicles, energy
efficiency, energy storage systems, innovative railway systems, smart
grids, and smart cities.

Andrés A. Nogueiras Meléndez (M’99–SM’12)

was born in Rosario, Argentina, in 1967. He re-
ceived the Graduate degree and the Ph.D. de-
gree (cum laude) from the University of Vigo,
Vigo, Spain, in 1994 and 2003, respectively, both
in industrial engineering.
He was a Research Assistant at the Applied
Electronics Institute, Pedro Barrie de la Maza
Foundation, Vigo, in 1994. He became an Assis-
tant Professor in the Electronic Technology De-
partment, University of Vigo, in 1995 and an As-
sociate Professor, in 2008. His current research interests include power
electronics for wireless energy transfer and management, switched con-
verter nonlinear modeling, and applied reliability, availability, mainte-
nance, and safety (RAMS) technologies.
Mr. Nogueiras Meléndez is a member of the IEEE Industrial Electron-
ics Society.