This article has been accepted for publication in a future issue of this journal, but has not been

fully edited. Content may change prior to final publication. Citation information: DOI 10.1109/TIE.2016.2535981, IEEE
Transactions on Industrial Electronics
IEEE TRANSACTIONS ON INDUSTRIAL ELECTRONICS 1

Model Predictive Control of Quasi-Z Source

Four-Leg Inverter
Sertac Bayhan, Member, IEEE, Haitham Abu-Rub, Senior Member, IEEE, and Robert S. Balog, Senior
Member, IEEE

Abstract—This paper presents a model predictive control The first drawback is the variation of the input voltage due to
(MPC) scheme for quasi-Z source (qZS) three-phase four-leg intermittent and stochastic nature of RES. The VSI must have
inverter. In order to cope with the drawbacks of traditional an input voltage that is greater than the maximum value of the
voltage source inverters (VSIs), a qZS three-phase four-leg
inverter topology is proposed. This topology features a wide range line-to-line output voltage in order to guarantee reliable and
of voltage gain which is suitable for applications in renewable uninterruptible power for the loads, which is a major challenge
energy based power systems,where the output of the renewable in such systems. To overcome this, DC/DC boost converter is
energy sources varies widely with operating conditions such often used as an input stage to create a well-regulated voltage
as wind speed, temperature and solar irradiation. To improve for the VSI. However, this solution results in complex power
capability of the controller, a MPC scheme is used which
implements a discrete-time model of the system. The controller circuit and multi-loop control structure which leads to lower
handles each phase current independently, which add flexibility reliability, and higher cost. The alternative topology to VSI
to the system. Simulation and experimental studies verify the is a current source inverter (CSI), which has the advantage
performances of the proposed control strategy under balanced of voltage boosting capability without using DC-DC boost
and unbalanced load conditions as well as single-phase open- converter [2]. This results in less complexity of the system
circuit fault condition.
Index Terms—Model predictive control, four-leg inverter,
and its control. However, CSI has some drawbacks such as: 1)
quasi-Z-source inverter (qZSI), DC−AC power conversion. its output voltage cannot be lower than the DC input voltage;
and 2) overlap time between phase legs is required to avoid
the open circuit of all upper switching devices or all lower
I. I NTRODUCTION devices. Otherwise, an open circuit of the DC inductor would

I NCREASING environmental awareness as a consequence

of climate change and the exhaustible nature of fossil fuels
have increased the importance of renewable energy sources
occur and destroy the devices. Furthermore, overlap time for
safe current commutation is needed in the CSI, which causes
waveform degradation [3]. Recently, a number of new power
(RES)s. However, interconnecting renewable energy gener- converter topologies have been proposed to cope with these
ation into the electrical distribution systems require power problems [4]. The quasi-impedance source inverter (qZSI)
electronics to generate and condition the electrical power to can overcome the aforementioned problems [5], [6]. They
meet voltage and frequency specifications. The performance advantageously utilize the shoot-through of the inverter bridge
of these renewable energy systems thus depends on the power to boost voltage in the VSI (or open circuit in the CSI to buck
converter topology and control method [1]. In most cases, voltage). Thus, buck - boost functionality is achieved with a
such systems use three-phase voltage source inverter (VSI) single-stage power conversion with a simple L-C network [7].
to supply the power either to the grid or the local loads. This A second drawback of three-phase VSI topology is that
approach is straightforward to implement, but, suffers from these inverters are usually designed for balanced three-phase
two major drawbacks: input voltage variation and unbalanced loads. However, unbalanced load conditions are common in
output current. distribution power generation system where power is delivered
to local loads. Unbalanced load condition creates unbalanced
Manuscript received May 7, 2015; revised October 18, 2015 and January
20, 2016; accepted February 13, 2016. This publication was made possible by current circulating in the power system, causing overheat of
NPRP-EP X-033-2-007 (Sections I and II) and NPRP 8-241-2-095 (Sections the neutral line, and harmonic distortion on the output voltage
III,IV, and V) from the Qatar National Research Fund (a member of Qatar [8], [9]. One way to alleviate this condition is to use a three-
Foundation). The statements made herein are solely the responsibility of the
authors. phase four-leg VSI. This topology can produce balanced three-
S. Bayhan is with the Department of Electronics and Automation, Gazi phase output even when the load is unbalanced and nonlinear,
University, Ankara, Turkey, and also with the Department of Electrical and making it a good choice for standalone power generation and
Computer Engineering, Texas A&M University at Qatar, Doha, 23874 Qatar
(e-mails: [email protected], [email protected]). three-phase UPS systems [10].
H. Abu-Rub is with the Department of Electrical and Computer Engineer- The control strategy of the power converter plays a crucial
ing, Texas A&M University at Qatar, Doha, 23874 Qatar, and also with Qatar role to ensure reliable and efficient operation of the renewable
Environment and Energy Research Institute, Hamad Bin Khalifa University,
Qatar Foundation, Doha, Qatar (e-mail: [email protected]). energy based power generation systems. There are a number
R. S. Balog is with the Renewable Energy and Advanced Power Elec- of proposed control strategies for the qZSI [11]–[17]. Methods
tronics Research Laboratory in the Department of Electrical and Computer for three-phase four-leg VSI have been also studied [18],
Engineering, Texas A&M University, College Station, TX 77843 USA and
hold a joint appointment with Texas A&M University at Qatar, Doha, Qatar. [19]. In most applications, proportional-integral (PI) based
(e-mail: [email protected]) cascade control structure has been employed to control current,

0278-0046 (c) 2015 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.
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI 10.1109/TIE.2016.2535981, IEEE
Transactions on Industrial Electronics
IEEE TRANSACTIONS ON INDUSTRIAL ELECTRONICS 2

voltage, etc. [20]. Although this control technique is easy to qZ-Source network Four-leg inverter RL filter Unknown loads
implement, it has some drawbacks. The major drawback is
that the performance of the entire system depends on the C2

Phase-a
performance of the inner control loop [21]. This controller L1 D L2

Phase-b
requires a modulator, such as 3D-SVPWM, to generate PWM Rfa Lfa ia
a

Phase-c
signals for the power switches to implement the desired control P Rfb Lfb

Phase-n
Sn b ib
action. Vin
C1 Rfc Lfc ic
c
The model predictive control (MPC) is an attractive alter-
n
native to the classical control methods due to its fast dynamic
response, simple concept, and ability to include different N Rfn Lfn in
Sn n
nonlinearities and constraints [22]. The major advantage of
MPC lies in the direct application of the control action to
the converter, without requiring a modulation stage. Several Fig. 1. quasi-Z source three-phase four-leg inverter topology.
studies have been presented under the name of MPC for cur-
rent control of traditional three-phase inverter [23], multilevel inverter switching cycle. In the shoot-through zero state, two
inverters [24]–[27], qZSI [28], and several electrical machine semiconductor switches in the same leg are simultaneously
drives [29]–[32]. It is concluded that MPC is currently one switched-on to create short-circuit across the DC link. During
of the most attractive control techniques for power converters this state energy is transferred in the qZS network from the
and machine drives. capacitors to the inductors, and this state is used to boost the
This paper presents MPC strategy of quasi-Z source (qZS) DC voltage.
three-phase four-leg inverter. As a response to the gaps in this
In the second stage of this topology, the four-leg inverter is
research area, the contributions of this study are summarized
used. As shown in Fig. 1, the load neutral point is connected to
as follows:
the mid-point of the inverter fourth phase leg to allow for zero
1) A qZS network has been used instead of a DC-DC+DC- sequence current/voltage. However, the addition of an extra leg
AC converter to overcome the drawbacks of traditional makes the switching schemes more complicated compared to a
three-phase VSI topology and two-stage power conver- three-leg VSI. Nevertheless, using the extra phase leg improves
sion, inverter capability and reliability. The four-leg inverter can be
2) A three-phase four-leg inverter has been employed to used under balanced/unbalanced and/or linear/nonlinear load
ensure reliable operation of renewable energy based conditions.
power generation system under balanced and unbalanced
load conditions,
3) MPC is used to control load current and qZS network B. Mathematical Model of the qZS Network
capacitor voltage with high accuracy and fast response,
4) The proposed controller handles each phase current The equivalent circuits of the qZS network in non-shoot-
independently. As a result of this, the proposed qZS through and shoot-through states are illustrated in Fig. 2 (a)
four-leg inverter provides fault tolerant capability, for and (b), respectively [33]. All voltages and currents are defined
example if one leg fails the others can work normally. in these figures and the polarities are shown with arrows.
1) Non-shoot-through state: During the non-shoot-through
In order to verify the steady-state and transient-state per- state four-leg inverter model is represented by a constant
formances of the proposed control method, simulation and current source; it can be seen from Fig. 2 (a). By applying
experimental studies were carried out with different operating Kirchhoff’s voltage law to Fig. 2 (a), inductors voltages (vL1
conditions. The paper is organized as follows: an overview of and vL2 ), DC link voltage (vP N ), and diode voltage (vdiode )
the proposed topology and the mathematical model are pre- are written as
sented in Section II. The proposed MPC strategy is described
in Section III. In Section IV simulation and experimental vL1 = Vin − VC1 , vL2 = −VC2 (1)
results are presented for different operating conditions. Finally,
the conclusion is provided in Section V. vP N = VC1 − vL2 = VC1 + VC2 , vdiode = 0 (2)

2) Shoot-through state: During the shoot-through state

II. Q UASI -Z S OURCE F OUR L EG I NVERTER M ODEL four-leg inverter model is represented by short-circuit, it can
A. Topology be seen from Fig. 2 (b). By applying Kirchhoff’s voltage law
The qZS four-leg inverter topology with R − L output filter to Fig. 2 (b), inductors voltages (vL1 and vL2 ), DC link voltage
is shown in Fig. 1. This topology can be investigated as two (vP N ), and diode voltage (vdiode ) are written as
stages: the quasi-Z source network and four-leg inverter with vL1 = VC2 + Vin , vL2 = VC1 (3)
R − L output filter and load.
In the first stage of this topology, the qZS is made of an vP N = 0 vdiode = VC1 + VC2 (4)
L − C impedance network, which can boost the DC voltage
in response to the so-called shoot-through zero state of the At steady state, the average voltage of the capacitors over

0278-0046 (c) 2015 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.
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI 10.1109/TIE.2016.2535981, IEEE
Transactions on Industrial Electronics
IEEE TRANSACTIONS ON INDUSTRIAL ELECTRONICS 3

VC2 iC2 For a three-phase four-leg inverter, the addition of the fourth
- + leg makes the switching states 16 (24 ). The valid switching
IL1 IL2 states with the corresponding phase and line voltages for the
vdiode
- + traditional four-leg inverter are presented in [34]. In addition to
+ vL1 - iD + vL2 - + these switching states, for this application, one extra switching
+
Vin + state is required in order to ensure shoot-through state. There-
Iin VC1 iC1 vPN fore, a total of 17 switching states are used in this application.
- - iPN
The voltage in each leg of the four-leg inverter can be
- expressed as
(a) vaN = Sa vdc , 

C 2 i
C 2 
- vbN = Sb vdc , 

+
IL1 IL2
(10)
vcN = Sc vdc , 
v d od



- + vnN = Sn vdc .

+ + vL1 - + vL2 - +
+ where Sa , Sb , Sc , and Sn are the switching states, vdc and
in
vPN iPN vnN are DC link and load neutral voltages, respectively.
Iin C 1 iC 1
- - The output voltage of this inverter can be written in terms
-
of the previous inverter voltages

(b) van = (Sa − Sn )vdc , 

Fig. 2. Equivalent circuit of the Quasi-Z source network. (a) in non-shoot- vbn = (Sb − Sn )vdc , (11)
through state, (b) in shoot-through state. 
vcn = (Sc − Sn )vdc .


one switching cycle are By applying Kirchhoff’s voltage law to Fig. 3, the inverter
T1
) voltages can be expressed in terms of load-neutral voltages
VC1 = T1 −T0 Vin ,
(5) and load currents as, in the following:
T0
VC2 = T1 −T0 Vin .
vaN = (Rf a + Ra ) ia + Lf a didta + vnN , 


where T0 is the duration of the shoot-through state and T1 vbN = (Rf b + Rb ) ib + Lf b dib
+ v , (12)
dt nN
is the duration of the non-shoot-through state, and Vin is the dic

vcN = (Rf c + Rc ) ic + Lf c dt + vnN .

input DC voltage.
From (2), (4), and (5) the peak DC link voltage across the
inverter bridge in Fig. 1 is From (11) and (12), the output voltages can be expressed
as
T
vP N = VC1 + VC2 = Vin = BVin (6) van = (Rf a + Ra ) ia + Lf a didta , 

T1 − T0 
vbn = (Rf b + Rb ) ib + Lf b dib
, (13)
where T is switching cycle (T0 + T1 ), B is the boost factor dt 
vcn = (Rf c + Rc ) ic + Lf c di
dt .
c

of the qZSI.
The average current of the inductors L1 , and L2 can be which is simplified to
calculated from the system power P ;
dij
vj = (Rf j + Rj ) ij + Lf j , j = a, b, c. (14)
IL1 = IL2 = Iin = P/Vin (7) dt
Applying Kirchhoff’s current law and (7) results in
ic1 = ic2 = iP N − IL1 (8)
Lf a
f 
The voltage gain (G) of the qZSI can be expressed as i a


G = v̂ln /0.5vP N = M B (9) Lf b

f
i b

in
where M is the modulation index, v̂ln is the peak AC phase

n
voltage. Lfc fc c
ic
c

C. Mathematical Model of the Four-Leg Inverter v a v b vcN vnN

The equivalent circuit of the four-leg inverter with the output ~ ~ ~ ~
R − L filter is shown in Fig. 3, where the Lf j is the filter N
inductance, Rf j is the filter resistance, and Rj is the load
resistance for each of the phase j = a, b, c. Fig. 3. Equivalent circuit of the three-phase four-leg inverter.

0278-0046 (c) 2015 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.
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI 10.1109/TIE.2016.2535981, IEEE
Transactions on Industrial Electronics
IEEE TRANSACTIONS ON INDUSTRIAL ELECTRONICS 4

and neutral current in can be written as Model Predictive Controller

* (k)
iabc qZS four-leg inverter Rfa Lfa ia Ra
in = ia + ib + ic (15) References
Sa(k)
a b c n
* (k)
vc1
Rfb Lfb Rb
The expression for output current, derived from (14) is Cost Function
Sb(k) ib
n
Optimization
Sc(k) Rfc Lfc ic Rc
dij 1 iabc(k+1) g(k+1)
= [vj − (Rf j + Rj ) ij ] , j = a, b, c. (16) Predictive
dt Lf Model vc1(k+1)
Sn(k) in

iabc(k),vc1(k), iL1(k)
III. T HE P ROPOSED M ODEL P REDICTIVE C ONTROL
The proposed MPC scheme is shown in Fig. 4. It has two Fig. 4. Block diagram of the proposed model predictive control scheme.
main layers, consisting of a predictive model and cost function
optimization. The discrete-time model of the system is used where ij (k + 1) is the predicted output current vector at the
to predict future behavior of the control variables. The cost next sampling time and Av and Ai are constant as defined by
function is used to minimize the error between the reference
and the predicted control variables in the next sampling time. Ts
)
Av = Lf +(R+Rf )Ts
This control technique has several advantages, such as: easy (20)
Lf
to implement in both linear and nonlinear systems, it shows Ai = Lf +(R+Rf )Ts
high accuracy and fast dynamic response, and it has very small
2) Predictive Model II: This model is used to predict future
steady state error throughout different operating points. More
behavior of the capacitor voltage (VC1 ). The continuous-time
detailed analysis of the MPC technique and its characteristics
model of the capacitor current can be expressed as
can be found in [35]. Here, the proposed MPC scheme is
described in the following steps.
d (VC1 − iC1 rc )
• Determination of references. iC1 = C1 (21)
dt
• Build discrete-time models of the system.
where C1 and rc are the capacitance and the equivalent series
• Define a cost function g.
resistance (ESR) of the capacitor, respectively. Based on (21),
• Prepare control algorithm.
the capacitor voltage is derived as,
A. Determination of references
DC link voltage and output currents references are normally dVC1 diC1 1
= rc + iC1 (22)
obtained through maximum power point tracking algorithm for dt dt C1
renewable energy sources. However, the objective of this paper By substituting (18) into (22), the discrete-time model of the
is the control capability of the qZS four-leg inverter. For this VC1 can be obtained as
reason, without loss of generality these references are left to
be defined by the user. 
Ts

VC1 (k + 1) = VC1 (k) + iC1 (k + 1) rc + iC1 (k) − rc
B. Discrete-time models of the system C
(23)
The control of the qZS four-leg inverter output currents
(ia , ib , ic ) and capacitor voltage (VC1 ) required two discrete- where VC1 (k +1) is the predicted capacitor voltage at the next
time models be created from the continuous-time equations. sampling time, and iC1 (k) is capacitor current that depends
To do that, the general structure of the forward-difference on the states of the qZSI topology. According to operational
Euler equation (17) is used so as to compute the differential principle of qZS network explained in Section II-B, for non-
equations of the output current and the capacitor voltage: shoot-through and shoot-through states, capacitor current can
be defined as follows:
df f (x0 + h) − f (x0 )
≈ (17)
dt h i) During non-shoot-through state:
To estimate the value in the next sample time, for a suitably
small time step (17) becomes the discretization equation iC1 = IL1 − (Sa ia + Sb ib + Sc ic ) (24)
∆f (k) f (k + 1) − f (k) ii) During shoot-through state:
≈ (18)
∆t Ts iC1 = −IL1 (25)
where Ts is the sampling time.
1) Predictive Model I: This model is used to predict
C. Cost Function Optimization
future behavior of each of the output currents (ia , ib , ic ). The
continuous-time expression for each phase current is given in The selection of the cost function is a key part of the MPC
(16). By substituting (18) into (16), the discrete-time model scheme. The proposed MPC scheme has two cost functions
for each output phase current is which are used to minimize output current and capacitor
voltage errors in the next sampling time. The output current
ij (k + 1) = Av vj (k + 1) + Ai ij (k), j = a, b, c. (19) cost function is defined as

0278-0046 (c) 2015 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.
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI 10.1109/TIE.2016.2535981, IEEE
Transactions on Industrial Electronics
IEEE TRANSACTIONS ON INDUSTRIAL ELECTRONICS 5

2
gi = i∗j (k + 1) − ij (k + 1)
2
= [i∗a (k + 1) − ia (k + 1)]
2
(26)
+ [i∗b (k + 1) − ib (k + 1)]
+ [i∗c (k + 1) − ic (k + 1)]2
where i∗j (k + 1) is the reference output current vector and
ij (k + 1) is the predicted output current vector in the next
step (j = a, b, c).
The cost function of capacitor voltage can also be defined
as
gv = λ ∗ |vC1
∗
(k + 1) − vC1 (k + 1)| (27)
where vC1∗
(k + 1) and vC1 (k + 1) are the reference and
predicted capacitor voltages, respectively. The weighting factor
(λ) was determined by using cost function classification tech-
nique that was detailed in [36]. The complete cost function
is
g(k + 1) = gi (k + 1) + gv (k + 1) (28)

D. Control Algorithm
The flowchart for the proposed control algorithm is given
in Fig. 5. Cost function minimization is implemented as a
repeated loop for each voltage vector to predict the values,
evaluate the cost function, and store the minimum value and
the index value of the corresponding switching state. The
control algorithm can be summarized in the next steps:
1) Sampling the output phase currents (iabc ), inductor cur-
rent (iL1 ), and capacitor voltage (vC1 ).
2) These are used to predict output currents and capacitor Fig. 5. Flowchart of the proposed MPC algorithm for qZS four-leg inverters.
voltage using the predictive model I, and II, respectively.
A. Buck&Boost Conversion Modes Analysis
3) All predictions are evaluated using the cost function.
4) The optimal switching state that corresponds to the The proposed qZS four-leg inverter can operate both buck
optimal voltage vector that minimizes the cost function or boost conversion modes according to the input voltage and
is selected to be applied at the next sampling time. the desired output voltage. To test voltage gain and boost
factor performances of the four-leg qZSI, the input voltage is
changed from 180 V to 80 V. To simplify analysis, we assume
IV. S IMULATION AND E XPERIMENTAL R ESULTS that output voltage (vln ) is equal to load voltage (vR ) and the
To verify the theoretical analysis and confirm the proposed voltage on R − L filter √ is neglected. In order to ensure same
MPC technique of qZS four-leg inverter, simulation and ex- load voltage (v̂ln = 50. 2) at wide range input voltage, the
periments have been conducted with the configuration shown reference output currents are (i∗a = i∗b = i∗c ) set to 7 A and
in Fig. 1. The parameters for both simulation and experiments loads are balanced (Ra =Rb =Rc =10 Ω).
are given in Table I. The minimum √ input voltage must be Vin = 2v̂ln /M = 123
V (with M = 2/ 3) to maintain 50 V rms output voltage. If
the input voltage is above 123 V, the qZSI can operate in buck
TABLE I conversion mode, vice versa, if the input voltage is below 123
QZS FOUR-LEG INVERTER AND LOAD PARAMETERS V, the qZSI can operate in boost conversion mode.
Parameter Value Case – A1: Vin =180 V, M =0.8;
Input DC voltage (Vin ) 80-180 V Experimental results of this case are shown in Fig. 6 (a).
qZS network inductances (L1 , L2 ) 2.5 mH Here, Vin >123 V, so, qZSI works in buck conversion mode.
qZS network capacitors (C1 , C2 ) 1000 µF
Thus, the boost factor B =1 and the voltage gain is G =
Load resistance 5-10 Ω
Filter inductance, Lf 10 mH B.M = 0.8. The maximum output line-to-line voltage is
Filter resistance Rf 0.05 Ω
Nominal frequency (fo ) 50 Hz √
Nominal output voltage (vln ) 50 V rms v̂ab = v̂bc = v̂ca = 3.G.Vin /2 ∼
=123 V
Sampling time (Ts ) 40 µs

0278-0046 (c) 2015 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.
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI 10.1109/TIE.2016.2535981, IEEE
Transactions on Industrial Electronics
IEEE TRANSACTIONS ON INDUSTRIAL ELECTRONICS 6

It can be observed from Fig. 6 (a), the voltage on C1 is inductor due to the voltage on L1 is zero.
equal to the input voltage 180 V and the voltage on C2 is 0 Case – A2: Vin =100 V, M =1;
V. It can be noted that a pure DC current flows through an
In order to maintain constant output voltage, qZSI works
in boost conversion mode because of Vin <123 V. From (9),
vRa vRb vRc 50 V/div one can get the boost factor B ∼ =1.42 and the voltage gain
G∼ =1.42. The experimental results of this test are shown in
Fig. 6 (b). The voltage on the DC link (vP N ) is boosted from
100 V to 142 V. In this case, the maximum output line-to-line
Vin 50 V/div voltage is
√
v̂ab = v̂bc = v̂ca = 3.G.Vin /2 ∼
=123 V
5 ms/div It can be seen from Fig. 6 (b), the voltage on C1 and C2 is
iL1 5 A/div 120 V and 22 V, respectively. Notice that the inductor current
(iL1 ) is continuous that reduce the input stress.
vP N , VC1 50 V/div Case – A3: Vin =80 V, M =0.85;
Fig. 6 (c) shows the experimental results for this case. From
(9), the boost factor B =2.08 and the voltage gain G =1.768
VC2 is obtained. Thus the DC link voltage (vP N ) is boosted from
80 V to 166 V. In this case, the maximum output line-to-line
voltage is
(a) √
v̂ab = v̂bc = v̂ca = 3.G.Vin /2 ∼=123 V
vRa vRb vRc 50 V/div
It can be seen from Fig. 6 (c), the voltage on C1 and C2 is
124 V and 42 V, respectively. Experimental results show that,
the qZS four-leg inverter can provide constant output voltage
under various input voltages without using DC/DC converter
Vin 50 V/div or transformer.

5 ms/div B. Steady-State Analysis

iL1 5 A/div
To perform steady-state analysis, the reference of qZS
∗
network capacitor voltage (Vc1 ) is set to 150 V and the input
50 µs/div voltage is Vin =100 V which results in the qZS four-leg inverter
vP N 50 V/div operating in boost mode in the following experimental studies.
VC1
The following three cases are considered to show the
VC2 effectiveness of the proposed controller under steady-state
operation.
1) Case–B1: Balanced reference currents (i∗a = i∗b = i∗c =
(b)
10 A ) and balanced loads (Ra =Rb =Rc =7.5 Ω).
vRa vRb vRc 50 V/div 2) Case–B2: Balanced reference currents (i∗a = i∗b = i∗c =
10 A ) and unbalanced loads (Ra =5 Ω, Rb =Rc =7.5 Ω).
3) Case–B3: Unbalanced reference currents (i∗a = 10 A,
i∗b = 5 A, i∗c = 5 A) and balanced loads (Ra =Rb =Rc =10
Ω).
Vin 50 V/div Steady-state experimental results of these cases are shown
in Figs. 7 (a)–(c). All experimental results show that the output
iL1 5 A/div 5 ms/div currents (ia , ib , ic ) and the capacitor voltage (VC1 ) track their
references (i∗a , i∗b , i∗c , VC1
∗
) with high accuracy while the DC
link voltage (VP N ) is kept constant. The neutral current, which
50 µs/div is the sum of the three-phase load currents, is zero in cases
vP N 50 V/div
of balanced reference current (see Figs. 7 (a), and (b)). On
VC1 the other hand, the neutral current flows through the fourth
VC2 leg of the inverter in case of the unbalanced reference current
(see Fig. 7 (c)). It can be observed that the inductor current
(iL1 ) remains continuous, which significantly reduce the input
(c) stress. It is important to note that double-line frequency (2*fo)
Fig. 6. Experimental results with the same output voltage at (a) Vin =180
ripple exist on the qZS inductor current, capacitor voltages,
V, M =0.8; (b) Vin =100 V, M =1; (c) Vin =80 V, M =0.85.

0278-0046 (c) 2015 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.
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI 10.1109/TIE.2016.2535981, IEEE
Transactions on Industrial Electronics
IEEE TRANSACTIONS ON INDUSTRIAL ELECTRONICS 7

and DC-link voltage due to the unbalanced currents. However, are very low and these are dependent on the design of qZS
it can be observed from Figs. 7 (b), and (c) that these ripple network.
The simulation result of the load currents trajectories under
various reference values is shown in Fig. 8. The blue trace
(c) is balanced current condition, while the black (b) and
ia ib ic 5 A/div red (a) are unbalanced. The trajectory under balanced current
condition shows a circle shape and the zero-sequence current
is zero. On the other hand, the trajectories under unbalanced
current condition show an oval shape and the zero-sequence
current travels alongside the γ axis.
in 10 A/div
5 ms/div C. Transient-State Analysis
iL1 10 A/div
The simulation results of transient-state analysis with bal-
vP N anced and unbalanced reference currents are shown in Figs.
50 V/div
VC1 9 (a) and (b), respectively. The reference output currents step
from 5 to 10 A are shown in Fig. 9 (a). For this test, reference
VC2 loads (Ra =Rb =Rc =6 Ω) are balanced. It can be seen that
the output currents track to the references with fast rise time
and no overshoot. The qZS network voltage (vP N ) is also
kept constant by the proposed controller at the current step-up
(a)
instant.
ia ib ic 5 A/div In Fig. 9 (b), the results are presented with unbalanced ref-
erence current step change and balanced loads (Ra =Rb =Rc =6
Ω). For this test, all reference output currents are set to 5 A
at the beginning. Then, reference currents are set to i∗a = 7 A,
i∗b = 10 A, and i∗c = 12 A. Results of this test show that, the
in 10 A/div proposed controller handles each phase current independently
and the output currents (ia , ib , ic ) and the capacitor voltage
5 ms/div
iL1 10 A/div (VC1 ) track their references (i∗a , i∗b , i∗c , VC1
∗
) with high accuracy
while the DC link voltage (VP N ) is kept constant. However,
the double-line frequency (2*fo) ripple exist on the qZS
vP N 50 V/div
network due to unbalanced current.
VC1
The experimental results with the same operating conditions
VC2 are shown in Figs. 10 (a) and (b), respectively, which totally
agree with the simulation results. It can be clearly seen
from results, the transient time is very short, and there is
no overshoot. Furthermore, the output currents are controlled
(b)
independently by the proposed controller.
ia 5 A/div
ib ic

(a)
in 10 A/div (b) ✻
5 ms/div (c) ✻
iL1 10 A/div ❵✻
❛γ
❄
❄
vP N 50 V/div ❄
VC1

VC2

(c)
Fig. 8. Trajectories of output current under various reference currents (a)
Fig. 7. Experimental results of steady-state analysis with (a) balanced
ia =10, ib =0 A, ic =10 A; (b) ia =10, ib =5 A, ic =5 A; (c) ia = ib =
reference currents and balanced loads; (b) balanced reference currents and
ic =10 A.
unbalanced loads; (c) unbalanced reference currents and balanced loads.

0278-0046 (c) 2015 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.
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI 10.1109/TIE.2016.2535981, IEEE
Transactions on Industrial Electronics
IEEE TRANSACTIONS ON INDUSTRIAL ELECTRONICS 8

D. Analysis of Fault-Tolerant Capability 5 A/div

The proposed system is also tested for a fault condition, ia ib ic
which may occur during the operation of the three-phase

in 10 A/div
5 ms/div
iL1 10 A/div

vP N 50 V/div
VC1
VC2

(a)
ib ic 5 A/div
ia

10 A/div
in
5 ms/div
iL1 10 A/div

vP N 50 V/div
(a) VC1
VC2

(b)
Fig. 10. Experimental results of transient-state analysis with (a) balanced
reference currents and balanced loads; (b) unbalanced reference currents and
balanced loads.

inverters. To do that, one-phase is made an open-circuit and

the corresponding reference current (i∗b ) is set to 0 A. The
transient-state experimental results are shown in Fig. 11 (a).
It is clear that, even with single-phase open-circuit fault
condition, the output currents are controlled independently
by the proposed controller and the circulating current flows
through the fourth leg of the inverter. Although there is single-
phase open-circuit fault, inverter can supply other two phases
loads with low (2.8%) total harmonic distortion (THD), as
shown in Fig. 12 (a).
To compare fault-tolerant capability of the proposed system
with a traditional MPC qZS three-leg inverter, the system is
tested under a single-phase open-circuit fault condition. In this
test, the neutral point of the load is connected to the star
point of the secondary side of the ∆/Y transformer, which
is connected to the output of qZSI, to observe the circulating
(b) current. The ’b’ phase is open-circuited and the corresponding
reference current (i∗b ) is set to 0 A similar to previous case in
Fig. 9. Simulation results of transient-state analysis with (a) balanced
reference currents and balanced loads; (b) unbalanced reference currents and
Fig. 11 (a). Experimental results of this test are shown in Fig.
balanced loads. 11 (b). The stability of the qZS network deteriorated after

0278-0046 (c) 2015 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.
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI 10.1109/TIE.2016.2535981, IEEE
Transactions on Industrial Electronics
IEEE TRANSACTIONS ON INDUSTRIAL ELECTRONICS 9

ic ia ib 5 A/div

in
10 A/div
5 ms/div
iL1 10 A/div
vP N 50 V/div
VC1
VC2 (a)

(a)
ic ia ib 5 A/div

in 10 A/div
5 ms/div
(b)
iL1 10 A/div
Fig. 12. Total harmonic distortion under single-phase open-circuit fault. (a)
vP N 50 V/div proposed qZS four-leg inverter, (b) traditional qZS three-leg inverter with
MPC control.
VC1
VC2 where ip , and in are positive, and negative sequence com-
ponent, respectively. According to the Fortescue method and
definition given in (29), Table II shows some of the case
studies and their symmetrical components, and the negative
(b) sequence current unbalance factor. Phase angles of the currents
Fig. 11. Transient-state experimental results under single-phase open-circuit are θa = 0◦ , θb = −120◦, and θc = −240◦.
fault. (a) proposed qZS four-leg inverter, (b) traditional qZS three-leg inverter
with MPC control. TABLE II
SYMMETRICAL COMPONENTS AND UNBALANCE FACTOR

single-phase fault. The inductor current is discontinuous that Case Studies i0 ip in ρi

increases the input stress, and the output current waveforms Fig. 7 (a) :
ia = ib = ic = 10 A 0A 10 A 0A 0%
affected from the fault. Furthermore, the output current of Fig. 7 (c) :
THD is higher than previous case, as shown in Fig. 12 (b). ia = 10 A, ib = ic = 5 A 1.67 A 6.67 A 1.67 A 25%
Fig. 8 (a) :
As a result of this test, the proposed qZS four-leg inverter ia = ic = 10 A, ib = 0 A 3.33 A 6.67 A 3.33 A 50%
offers much better fault-tolerant capability than traditional Fig. 10 (b):
ia = 7 A, ib = 10 A, ic = 12 A 1.45 A 9.67 A 1.45 A 15%
three-phase inverter.

E. Analysis of Symmetrical Components

V. C ONCLUSION
To analyze the symmetrical components (zero, positive, and
negative sequences) of the three-phase current signals under This paper presents model predictive control (MPC) scheme
unbalanced conditions, Fortescue method is used [37]. The for quasi-Z source (qZS) three-phase four-leg inverter. The
level of unbalance is also described by the negative sequence main aim of this paper is to achieve single-stage power con-
current unbalance factor (ρi ), which is given as the modulus verter topology for renewable energy based power generation
of the ratio of negative to positive sequence currents (same for systems under balanced and unbalanced conditions with high
the negative sequence voltage unbalance factor (ρv )) [38]. control capability. To do that, qZS three-phase four-leg inverter
topology was proposed in this study. To improve control
in capability of the controller, the MPC scheme was employed
ρi = ∗ 100 (%) (29)
ip in the controller stage. Simulation and experimental studies

0278-0046 (c) 2015 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.
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI 10.1109/TIE.2016.2535981, IEEE
Transactions on Industrial Electronics
IEEE TRANSACTIONS ON INDUSTRIAL ELECTRONICS 10

were performed to verify the performance of the proposed [20] D.-K. Choi and K.-B. Lee, “Dynamic performance improvement of ac/dc
inverter topology and its control strategy. The results show converter using model predictive direct power control with finite control
set,” IEEE Trans. Ind. Electron., vol. 62, no. 2, pp. 757–767, Feb 2015.
that the proposed technique not only has an excellent steady- [21] J. Bocker, B. Freudenberg, A. The, and S. Dieckerhoff, “Experimental
state and transient performances, but also it is robust against comparison of model predictive control and cascaded control of the
fault conditions. modular multilevel converter,” IEEE Trans. Power Electron., vol. 30,
no. 1, pp. 422–430, Jan 2015.
[22] M. Rivera, V. Yaramasu, A. Llor, J. Rodriguez, B. Wu, and M. Fadel,
“Digital predictive current control of a three-phase four-leg inverter,”
R EFERENCES IEEE Trans. Ind. Electron., vol. 60, no. 11, pp. 4903–4912, Nov 2013.
[23] P. Cortes, G. Ortiz, J. Yuz, J. Rodriguez, S. Vazquez, and L. Franquelo,
[1] H. Abu-Rub, M. Malinowski, and K. Al-Haddad, Power Electronics for
“Model predictive control of an inverter with output lc filter for ups
Renewable Energy Systems, Transportation and Industrial Applications.
applications,” IEEE Trans. Ind. Electron., vol. 56, no. 6, pp. 1875–1883,
A John Wiley&Sons Ltd., 2014.
June 2009.
[2] X. Guo, D. Xu, and B. Wu, “Four-leg current-source inverter with a new
[24] P. Cortes, A. Wilson, S. Kouro, J. Rodriguez, and H. Abu-Rub, “Model
space vector modulation for common-mode voltage suppression,” IEEE
predictive control of multilevel cascaded h-bridge inverters,” IEEE
Trans. Ind. Electron., vol. 62, no. 10, pp. 6003–6007, 2015.
Trans. Ind. Electron., vol. 57, no. 8, pp. 2691–2699, Aug 2010.
[3] F. Z. Peng, “Z-source inverter,” IEEE Trans. Ind. Appl., vol. 39, no. 2, [25] M. Narimani, B. Wu, V. Yaramasu, Z. Cheng, and N. Zargari, “Finite
pp. 504–510, 2003. control-set model predictive control (fcs-mpc) of nested neutral point
[4] S. Kouro, J. Leon, D. Vinnikov, and L. Franquelo, “Grid-connected clamped (nnpc) converter,” IEEE Trans. Power Electron., vol. PP, no. 99,
photovoltaic systems: An overview of recent research and emerging pv pp. 1–1, 2015.
converter technology,” IEEE Ind. Electron. Mag., vol. 9, no. 1, pp. 47– [26] B. Riar, T. Geyer, and U. Madawala, “Model predictive direct current
61, March 2015. control of modular multilevel converters: Modeling, analysis, and ex-
[5] Y. Li, S. Jiang, J. Cintron-Rivera, and F. Z. Peng, “Modeling and control perimental evaluation,” IEEE Trans. Power Electron., vol. 30, no. 1, pp.
of quasi-z-source inverter for distributed generation applications,” IEEE 431–439, Jan 2015.
Trans. Ind. Electron., vol. 60, no. 4, pp. 1532–1541, April 2013. [27] V. Yaramasu and B. Wu, “Model predictive decoupled active and reactive
[6] B. Ge, H. Abu-Rub, F. Z. Peng, Q. Lei, A. De Almeida, F. Ferreira, power control for high-power grid-connected four-level diode-clamped
D. Sun, and Y. Liu, “An energy-stored quasi-z-source inverter for inverters,” IEEE Trans. Ind. Electron., vol. 61, no. 7, pp. 3407–3416,
application to photovoltaic power system,” IEEE Trans. Ind. Electron., July 2014.
vol. 60, no. 10, pp. 4468–4481, Oct 2013. [28] S. Bayhan and H. Abu-Rub, “Model predictive control of quasi-z source
[7] W. Qian, F. Z. Peng, and H. Cha, “Trans-z-source inverters,” IEEE Trans. three-phase four-leg inverter,” in IEEE 41st Annual Conference of the
Power Electron., vol. 26, no. 12, pp. 3453–3463, 2011. IEEE Industrial Electronics Society, IECON 2015, 2015, pp. 362–367.
[8] H. Jin, X. Rui, Z. Yunping, W. Zhi, and S. Haixia, “The study of spwm [29] J. Guzinski and H. Abu-Rub, “Speed sensorless induction motor drive
control strategy to reduce common-mode interferences in three-phase with predictive current controller,” IEEE Trans. Ind. Electron., vol. 60,
four-leg inverters,” in IEEE 3rd Conference on Industrial Electronics no. 2, pp. 699–709, Feb 2013.
and Applications, 2008. ICIEA 2008., June 2008, pp. 924–928. [30] S. Bayhan and H. Abu-Rub, “Model predictive sensorless control
[9] S. Bifaretti, A. Lidozzi, L. Solero, and F. Crescimbini, “Comparison of standalone doubly fed induction generator,” in IEEE 40th Annual
of modulation techniques for active split dc-bus three-phase four- Conference of the IEEE Industrial Electronics Society, IECON 2014,
leg inverters,” in IEEE Energy Conversion Congress and Exposition Oct 2014, pp. 2166–2172.
(ECCE), Sept 2014, pp. 5631–5638. [31] C. S. Lim, E. Levi, M. Jones, N. Rahim, and W. Hew, “Fcs-mpc-based
[10] E. Demirkutlu and A. Hava, “A scalar resonant-filter-bank-based output- current control of a five-phase induction motor and its comparison with
voltage control method and a scalar minimum-switching-loss discontin- pi-pwm control,” IEEE Trans. Ind. Electron., vol. 61, no. 1, pp. 149–163,
uous pwm method for the four-leg-inverter-based three-phase four-wire Jan 2014.
power supply,” IEEE Trans. Ind. Electron., vol. 45, no. 3, pp. 982–991, [32] V. Yaramasu, B. Wu, S. Alepuz, and S. Kouro, “Predictive control
May 2009. for low-voltage ride-through enhancement of three-level-boost and npc-
[11] Y. Liu, B. Ge, H. Abu-Rub, and F. Z. Peng, “An effective control method converter-based pmsg wind turbine,” IEEE Trans. Ind. Electron., vol. 61,
for three-phase quasi-z-source cascaded multilevel inverter based grid-tie no. 12, pp. 6832–6843, Dec 2014.
photovoltaic power system,” IEEE Trans. Ind. Electron., vol. 61, no. 12, [33] Y. Li, J. Anderson, F. Peng, and D. Liu, “Quasi-z-source inverter for
pp. 6794–6802, Dec 2014. photovoltaic power generation systems,” in IEEE 24th Annual Applied
[12] L. Yushan, G. Baoming, H. Abu-Rub, and Z. P. Fang, “Modelling Power Electronics Conference and Exposition, APEC 2009., Feb 2009,
and controller design of quasi-z-source inverter with battery-based pp. 918–924.
photovoltaic power system,” IET Power Electronics, vol. 7, no. 7, pp. [34] V. Yaramasu, M. Rivera, B. Wu, and J. Rodriguez, “Model predictive
1665–1674, July 2014. current control of two-level four-leg inverters;part i: Concept, algorithm,
[13] L. Yushan, G. Baoming, H. Abu-Rub, and F. Z. Peng, “Control system and simulation analysis,” IEEE Trans. Power Electron., vol. 28, no. 7,
design of battery-assisted quasi-z-source inverter for grid-tie photo- pp. 3459–3468, July 2013.
voltaic power generation,” IEEE Trans. Sustain. Energy, vol. 4, no. 4, [35] J. Rodriguez, M. Kazmierkowski, J. Espinoza, P. Zanchetta, H. Abu-Rub,
pp. 994–1001, Oct 2013. H. Young, and C. Rojas, “State of the art of finite control set model
[14] H. Abu-Rub, A. Iqbal, S. Moin Ahmed, F. Peng, Y. Li, and G. Baom- predictive control in power electronics,” IEEE Trans. Ind. Informat.,
ing, “Quasi-z-source inverter-based photovoltaic generation system with vol. 9, no. 2, pp. 1003–1016, May 2013.
maximum power tracking control using anfis,” IEEE Trans. Sustain. [36] J. Rodriguez and P. Cortes., Predictive Control of Power Converters and
Energy, vol. 4, no. 1, pp. 11–20, Jan 2013. Electrical Drives. John Wiley & Sons, Ltd., 2012.
[15] I. Roasto, D. Vinnikov, J. Zakis, and O. Husev, “New shoot-through [37] R. Kumar, B. Singh, and D. Shahani, “Symmetrical components based
control methods for qzsi-based dc/dc converters,” IEEE Trans. Ind. technique for power quality event detection and classification,” in IEEE
Electron., vol. 9, no. 2, pp. 640–647, May 2013. International Conference on Power Electronics, Drives and Energy
[16] Y. Liu, B. Ge, F. Ferreira, A. de Almeida, and H. Abu-Rub, “Modeling Systems (PEDES), 2014, pp. 1–6.
and svpwm control of quasi-z-source inverter,” in 11th International [38] J. Miret, M. Castilla, A. Camacho, L. Garcia de Vicuna, and J. Matas,
Conference on Electrical Power Quality and Utilisation (EPQU), 2011, “Control scheme for photovoltaic three-phase inverters to minimize
pp. 1–7. peak currents during unbalanced grid-voltage sags,” IEEE Trans. Power
[17] Y. Liu, H. Abu-Rub, B. Ge, F. Blaabjerg, O. Ellabban, and P. Loh, Electron., vol. 27, no. 10, pp. 4262–4271, 2012.
Impedance Source Power Electronic Converters. Hoboken, NJ: John
Wiley & Sons Ltd., 2016.
[18] X. Li, Z. Deng, Z. Chen, and Q. Fei, “Analysis and simplification of
three-dimensional space vector pwm for three-phase four-leg inverters,”
IEEE Trans. Ind. Electron., vol. 58, no. 2, pp. 450–464, Feb 2011.
[19] D. Patel, R. Sawant, and M. Chandorkar, “Three-dimensional flux vector
modulation of four-leg sine-wave output inverters,” IEEE Trans. Ind.
Electron., vol. 57, no. 4, pp. 1261–1269, April 2010.

0278-0046 (c) 2015 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.
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI 10.1109/TIE.2016.2535981, IEEE
Transactions on Industrial Electronics
IEEE TRANSACTIONS ON INDUSTRIAL ELECTRONICS 11

Sertac Bayhan (M’14) is with the Department of

Electronic and Automation at the University of Gazi
in Turkey where he has been a faculty member since
2009. Dr. Bayhan is currently a Assistant Research
Scientist in Texas A&M University at Qatar. Dr.
Bayhan completed his Ph.D. at Gazi University and
his undergraduate studies also at the same univer-
sity. He graduated with valedictorian. His research
interests lie in the areas of power electronic, renew-
able energy conversion for PV and wind systems,
condition monitoring and power quality detection in
power systems, microgrid, and smart grid applications. He has published more
than 50 high impact journal and conference papers. He is a co-author of
one book and two book chapters. Dr. Bayhan was awarded best presentation
recognition in the 41st Annual Conference of the IEEE Industrial Electronics
Society in 2015. Dr. Bayhan also was a recipient of many awards, such
as research excellence travel awards in 2014 and 2015 (at Texas A&M
University at Qatar), paper support awards in 2009, 2014, 2015, and 2016
(at the Scientific and Technological Research Council of Turkey).

Haitham Abu-Rub (M’99, SM’07) holds two PhDs,

one in electrical engineering and another in human-
ities. Since 2006, Dr. Abu-Rub has been associ-
ated with Texas A&M University, where he was
promoted to professor. Currently he is the chair of
Electrical and Computer Engineering Program at the
same university as well as the managing director of
Smart Grid Center Extension in Qatar. He is also a
Chief Scientist with Qatar Environment and Energy
Research Institute (QEERI), Qatar Foundation. His
main research interests are energy conversion sys-
tems, including electric drives, power electronic converters, renewable energy
and smart grid. Dr. Abu-Rub is the recipient of many prestigious international
awards, such as the American Fulbright Scholarship, the German Alexander
von Humboldt Fellowship, the German DAAD Scholarship, and the British
Royal Society Scholarship. Dr. Abu-Rub has published more than 250 journal
and conference papers, and has earned and supervised many research projects.
Currently he is leading many potential projects on photovoltaic and hybrid
renewable power generation systems with different types of converters and on
electric drives. He is co-author of four books, two of which are with Wiley.
He is also an author and co-author of five book chapters. Dr. Abu-Rub is an
active IEEE senior member and is an editor in many IEEE journals.

Robert S. Balog (S’92 M’96 SM’07) received the

B.S. degree in electrical engineering from Rutgers,
The State University of New Jersey, New Brunswick,
NJ, USA in 1996, and the M.S. and Ph.D. degrees in
electrical engineering from the University of Illinois
at Urbana-Champaign, Urbana, IL, USA, in 2002
and 2006, respectively.
From 1996 to 1999, he was an Engineer with
Lutron Electronics, Coopersburg, PA, USA. From
2005 to 2006, he was a Researcher with the U.S.
Army Corp of Engineers, Engineering Research and
Development Center, Construction Engineering Research Laboratory, Cham-
paign, IL, USA. From 2006 to 2009, he was a Senior Engineer at SolarBridge
Technologies, Champaign, IL, USA (aquired by Sunpower Corp in 2014).
He then joined Texas A&M University, College Station, TX, USA, where
he is currently an Associate Professor with tenure in the Department of
Electrical and Computer Engineering. He simultaneously holds a joint faculty
apppointment with Texas A&M University at Qatar and is the Director of the
Renewable Energy and Advanced Power Electronics Research Laboratory.
Dr. Balog holds 17 issued and pending U.S. patents. His current research
interests include power converters and balance-of-systems technologies for
solar photovoltaic energy, particularly microinverters for ac photovoltaic
modules, and highly reliable electrical power and energy systems including
dc microgrids. He is co-author of the book Microgrids and other Local Area
Power and Energy Systems published by Cambridge Univeristy Press.

0278-0046 (c) 2015 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.