5458 IEEE TRANSACTIONS ON INDUSTRIAL ELECTRONICS, VOL. 62, NO.
9, SEPTEMBER 2015
An Energy Stored Quasi-Z-Source Cascade
Multilevel Inverter-Based Photovoltaic
Power Generation System
Dongsen Sun, Baoming Ge, Member, IEEE, Weihua Liang, Haitham Abu-Rub, Senior Member, IEEE,
and Fang Zheng Peng, Fellow, IEEE
Abstract—The quasi-Z-source cascade multilevel in-
verter (qZS-CMI) presents many advantages over con-
ventional CMI when applied in photovoltaic (PV) power
systems. For example, the qZS-CMI provides the balanced
dc-link voltage and voltage boost ability, saves one-third
modules, etc. However, the qZS-CMI still cannot overcome
the intermittent and stochastic fluctuation of solar power
injected to the grid. This paper proposes an energy stored
qZS-CMI-based PV power generation system. The system
combines the qZS-CMI and energy storage by adding an en-
ergy stored battery in each module to balance the stochas-
tic fluctuations of PV power. This paper also proposes a
control scheme for the energy stored qZS-CMI-based PV
system. The proposed system can achieve the distributed
maximum power point track for PV panels, balance the
power between different modules, and provide the desired
power to the grid. A detailed design method of controller
parameters is disclosed. Simulation and experimental re-
sults verify the proposed system and the control scheme.
Index Terms—Cascade multilevel inverter (CMI), energy
storage, photovoltaic (PV) power generation, quasi-Z-
source inverter (qZSI).
I. I NTRODUCTION
N OWADAYS, applying multilevel inverters to photovoltaic
(PV) power systems is gaining more and more attention
[1]–[3]. Among the typical multilevel inverter topologies, the
cascade multilevel inverter (CMI) is more widely used due to its
attractive features, such as achieving the distributed maximum
power point tracking (MPPT) and high voltage/high power
grid tie without a transformer. However, each module is a
buck inverter in the conventional CMI-based PV power system,
Manuscript received January 31, 2014; revised May 9, 2014, July 22,
2014, and September 24, 2014; accepted October 22, 2014. Date of
publication February 27, 2015; date of current version August 7, 2015.
This work was supported by NPRP-EP under Grant X-033-2-007 from
the Qatar National Research Fund (a member of Qatar Foundation).
D. Sun and W. Liang are with the School of Electrical Engineering,
Beijing Jiaotong University, Beijing 100044, China.
B. Ge is with the School of Electrical Engineering, Beijing Jiaotong
University, Beijing 100044, China, and with the Department of Electrical
and Computer Engineering, Texas A&M University, College Station, TX
77843 USA (e-mail: bm-ge@263.net).
H. Abu-Rub is with the Department of Electrical and Computer
Engineering, Texas A&M University at Qatar, Doha 23874, Qatar, and
with Qatar Environment and Energy Research Institute (QEERI), Doha,
Qatar.
F. Z. Peng is with the Department of Electrical and Computer Engi-
neering, Michigan State University, East Lansing, MI 48824 USA.
Color versions of one or more of the figures in this paper are available
online at http://ieeexplore.ieee.org.
Digital Object Identifier 10.1109/TIE.2015.2407853
0278-0046 © 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.
and each module’s PV voltage variation will cause the whole
system’s dc-link voltages to be imbalanced. Considering the
unique features of the Z-source inverter (ZSI) and quasi-Z-
source inverter (qZSI), i.e., implementing voltage boost/buck
and inversion in a single stage [4]–[6], the Z-source/quasi-
Z-source cascade multilevel inverter (ZS/qZS-CMI)-based PV
power systems have been proposed in [7]–[12] to overcome the
aforementioned disadvantages of the conventional CMI-based
PV system. In particular, GaN devices were used in ZS/qZS-
CMI-based PV systems to achieve high efficiency in [7] and
[8]; three modulation methods, i.e., phase-shifted sinusoidal
pulsewidth modulation (PS-SPWM), phase-shifted pulsewidth-
amplitude modulation (PS-PWAM), and modular multilevel
space vector modulation (MMSVM), were proposed for the
quasi-Z-source CMI (qZS-CMI)-based PV system in [9], [10],
and [12], respectively, where PS-SPWM was a basic method
derived from the conventional CMI, PS-PWAM was proposed
to reduce the switching number and loss, and MMSVM was for
the three-phase qZS-CMI to simplify modulation implementa-
tion and enhance the voltage utilization ratio. In [11], the qZS-
CMI-based PV system was compared with the traditional CMI
system, and the conclusion was that the qZS-CMI PV system
saved one-third of the modules than the traditional CMI-based
PV system.
In addition, solar power presents intermittent and stochastic
characteristics, so energy storage is added in the PV systems to
obtain smooth power [13]. Energy stored qZSI was proposed
for PV system application in [14]–[17] to achieve the same pur-
pose in the simple way, where the focus was mainly on three-
phase two-level energy stored qZSI-based PV power generation
systems, and they presented system modeling, control schemes,
and controller design for the three-phase two-level system.
If the energy stored qZSI is combined with the CMI, the result-
ing system—the energy stored qZS-CMI-based PV system—
will inherit all of the merits of the energy storage and the
qZS-CMI-based system, where each module has a three-level
energy stored qZS H-bridge. When comparing to the ZS/qZS-
CMI-based PV system in [7]–[12], the additional energy stor-
age of each module in the new system will balance the power
difference between the grid and PV panels, while the whole
system provides the desired power to the grid.
The literature of the three-phase two-level energy stored
qZSI and the qZS-CMI provides effective references for the
energy stored qZS-CMI-based PV system, but some special
distinct features should be taken into account. For example, the
SUN et al.: ENERGY STORED QZS-CMI-BASED PHOTOVOLTAIC POWER GENERATION SYSTEM 5459

operating principle, the shoot-through state is added in the con-

ventional zero state in the modulation process. Therefore, each
module in the system still outputs a three-level voltage vHn .
For the whole system’s modulation, the PS-SPWM method
in [9] is employed, and the carriers of three modules are shifted
by 60◦ to each other. As a result, the energy stored qZS-CMI
provides a seven-level output voltage vH to feed the 50-Hz grid
through the L-filter.

B. Modeling of Proposed System

In this paper, the two capacitors of each module have the
same capacitance, and the two inductors of each module have
the same inductance, i.e., C = C1 = C2 and L = L1 = L2 . The
reasons are as follows.
1) Different capacitances will cause different current behav-
iors of the two inductors. In particular, there will be larger
inductor current second harmonic (2ω) ripple (at least for
the inductor L2 ) when compared to the case with the same
capacitance. Large inductor current 2ω ripple will result in large
loss and high current stress of inductors.
2) Our design also aims to obtain constant inductor current
Fig. 1. Topology of the proposed system. without 2ω ripple. For this case, two capacitors will have the
same 2ω current ripple.
three-phase two-level energy stored qZSI in [14]–[17] only Two capacitors with equal capacitance will result in the same
has one inverter with one energy storage, while the energy 2ω voltage ripple, so the 2ω ripple of dc-link voltage is divided
stored qZS-CMI needs to manage the operation of several into two equal parts—half on the capacitor C1 and half on
modules and consists of many inverter modules with many the capacitor C2 . Otherwise, if the capacitor C2 has smaller
energy storages; the qZS-CMI in [7]–[12] did not have an en- capacitance than the capacitor C1 , the capacitor C2 will have
ergy storage, which achieved the balanced dc-link voltages and higher 2ω voltage ripple, which will result in higher 2ω ripple
resulted in all PV powers being injected into the grid even with of the battery current.
fluctuation. The energy stored qZS-CMI can have balanced 3) A similar situation will occur if the two inductors have
and smooth power injected into the grid, where each module’s differing inductances. The modeling and controller design will
energy storage will absorb redundant PV power and/or supply also be complicated if C1 = C2 and L1 = L2 .
absent power to the grid. Therefore, the energy stored qZS-CMI According to the operating principle of energy stored qZSI
requires different control methods from those existing in the [14]–[17], the state space equation of each module is
three-phase two-level energy stored qZSI and qZS-CMI. ⎡ Dn −1 Dn
⎤
This paper proposes an energy stored qZS-CMI applied in a 0 0 L L
⎢ Dn Dn −1 ⎥
PV power system. A control scheme is investigated, and the ⎢ 0 0 ⎥
ẋn = ⎢ L L ⎥ xn
controller parameters are designed by using the built small- ⎣ 1−Dn
C − DCn 0 0 ⎦
signal model and Bode plot method. With the proposed control − DCn 1−Dn
0 0
C
scheme, the energy stored qZS-CMI-based PV system can ⎡ ⎤
1
achieve the distributed MPPT for all PV panels; moreover, it L 0 0
can provide the desired power to the grid with the balanced ⎢ 0 0 0 ⎥
+⎢
⎣ 0
⎥u (1)
module power. Simulation and experimental results verify that Dn −1
C 0 ⎦ n
Dn −1 1
the proposed system, along with the control scheme, works 0 C C
properly in the desired manner.
where the subscript n denotes the nth module, n ∈ {1, 2, 3}; xn
II. P ROPOSED E NERGY S TORED QZS-CMI-PV is the state vector, xn = [x1n x2n x3n x4n ]T = [iL1n iL2n vC1n
P OWER S YSTEM vC2n ]T ; Dn is the shoot-through duty ratio of the nth module
and 0 < Dn < 0.5; un = [u1n u2n u3n ]T = [vpvn iPNn ibn ]T ;
A. Topology of Proposed System and the battery current is
A sample topology of the proposed system is shown in Fig. 1, vbn − vC2n
which consists of three modules with outputs connected in ibn = (2)
rbn
series. Each module includes a PV panel, a battery, and a qZS
H-bridge inverter. The unipolar PWM method is used in each where vbn is the battery open-circuit voltage, rbn is the battery
module to operate the H-bridge inverter. Considering the qZSI’s resistance, and vC2n is the voltage of capacitor C2 . For vbn at
5460 IEEE TRANSACTIONS ON INDUSTRIAL ELECTRONICS, VOL. 62, NO. 9, SEPTEMBER 2015

the determined battery state of charging (SOC), battery current

ibn depends on the voltage vC2n .
For the grid side, the dynamic equation is

dig
= [vH − vg − rf ig ]/Lf (3)
dt

where Lf and rf are the inductance and parasitic resistance of

the filter, respectively, vg and ig are the grid voltage and current,
respectively, and vH is the output voltage of the energy stored
qZS-CMI, i.e.,

vH = vH1 + vH2 + vH3 . (4)

The equilibrium point of the system (1) can be solved as

⎧
⎪
⎪ (1 − D̄n ) · īPNn − D̄n · ībn
⎪
⎪
⎪
⎪ 1 − 2D̄n
⎪
⎪
⎪
⎪
Fig. 2. Control scheme for the proposed system.
⎪
⎪ (1 − D̄n ) · īPNn − (1 − D̄n ) · ībn
⎪
⎨ 1 − 2D̄n
x̄n = (5) III. C ONTROL S CHEME
⎪
⎪ 1 − D̄n
⎪
⎪ · v̄pvn
⎪
⎪ 1 − 2D̄n Fig. 2 shows the proposed control scheme for the energy
⎪
⎪
⎪
⎪ stored qZS-CMI-based PV power generation system. Each
⎪
⎪ D̄n
⎪
⎩ · v̄pvn module has its own MPPT and battery energy management to
1 − 2D̄n ensure: 1) Each module’s PV power is maximum, and 2) each
module’s battery works properly through controlling each
where the symbol “–” above the variables denotes their equilib- battery SOC within the safe range (i.e., SOCmin < SOC <
rium terms. SOCmax , where SOCmin and SOCmax are the allowable min-
The dc-link voltage envelope of the nth module is imum and maximum battery SOCs, respectively). The grid
power control is used to transmit the three modules’ power to
1
v̄ˆPNn = v̄C1n + v̄C2n = v̄pvn . (6) the grid and keep the whole system operating at unity power
1 − 2D̄n factor.
The currents and powers of the nth module meet
A. MPPT Control of Each Module PV Panel
īL1n − īL2n = ībn ; Pn − Ppvn = Pbn (7)
The perturb and observe algorithm is employed for the
PV panel’s MPPT control [18]. As shown in Fig. 2, the nth
where Ppvn , Pn , and Pbn are the PV power, the load power module has its own MPPT-n that outputs the voltage reference
(positive in going to the load), and the battery power (positive ∗
vpvn , where n denotes the nth module, n = 1, 2, 3. The nth
in discharging) of the nth module, respectively. proportional and integral (PI) regulator is applied to track the
From (7), the energy stored qZSI module will perform the ∗
voltage reference vpvn by regulating Dn . Here, Dn is not only
following. used to implement the MPPT for the nth PV panel but also used
1) If Ppvn < Pn , Pbn > 0, ibn > 0, and the battery is to boost the PV voltage to a high voltage level. With distributed
discharging. MPPT control, all PV panels in the system always provide the
2) If Ppvn > Pn , Pbn < 0, ibn < 0, and the battery is maximum power to the grid.
charging.
3) If Ppvn = Pn , Pbn = 0, ibn = 0, and the battery is nei- B. Battery Energy Management of Each Module
ther discharging nor charging. Sun et al. [19] presented a valuable battery energy manage-
ment method for the three-phase energy stored qZSI. The same
From (7), when two powers are controlled, the third power
method is applied in each module of the proposed system, as
is automatically managed. There are three ways to control ∗
shown in Fig. 2. If the total grid power reference is Pgrid , each
these three powers: 1) If PV and grid-tie powers are controlled ∗
module power reference can be set at Pgrid /3. For each module,
actively, the battery current/power is automatically fitted by
there are three operating cases.
(7); 2) if battery current/power and grid-tie power are actively
controlled, PV power is balanced by (7); and 3) if battery and Case 1) The nth module battery operates within a safe range,
PV powers are actively controlled, grid-tie power is balanced in i.e., SOCmin < SOC < SOCmax , and the nth mod-
(7). In this paper, the first control strategy is employed. ule output power can match its power reference,
SUN et al.: ENERGY STORED QZS-CMI-BASED PHOTOVOLTAIC POWER GENERATION SYSTEM 5461

∗
i.e., Pn = Pgrid /3, n = 1, 2, 3, no matter how much each module of the proposed system are produced to achieve
power is from the nth module’s PV panel. the desired performance.
Case 2) When the nth module battery SOC reaches SOCmin ,
this module battery cannot be discharged any further.
IV. C ONTROL PARAMETERS D ESIGN
The output power of the nth module depends on the
comparison of its PV power Ppvn and the module As shown in Fig. 2, the system includes three PV voltage
∗ ∗
power reference Pgrid /3. If Ppvn < Pgrid /3, the nth controllers and a grid current controller. Their parameters will
module battery will operate in neither charging nor be designed by using the built model and Bode plot as follows.
discharging state, and the nth module outputs entire Each module has the following parameters: carrier frequency
∗
PV power, i.e., Pn = Ppvn ; if Ppvn ≥ Pgrid /3, the fc = 10 kHz, L1 = L2 = 500 μH, C1 = C2 = 4400 μF, and
nth module battery will be charged, and the nth mod- the battery voltage vb = 36 V. The whole system’s rated power
∗
ule provides the desired power Pn = Pgrid /3. is Pgrid = 1500 W, the grid voltage is vg = 220 V, the grid
Case 3) When the nth module battery SOC reaches SOCmax , frequency is fg = 50 Hz, Lf = 500 μH, rf = 0.01 Ω, and each
this module battery cannot be charged any more. The module has Pn = 500 W (n = 1, 2, 3). The rated parameters
output power of the nth module depends on the com- of the PV panel are the following: PV voltage vpv = 68 V, PV
parison of its PV power Ppvn and the module power current ipv = 7.3 A, PV power Ppv = 496.4 W, the PV panel’s
∗ ∗
reference Pgrid /3. If Ppvn < Pgrid /3, the nth module resistance is Rn = 0.37 Ω, and the PV voltage range is 45–90 V.
battery will operate in discharging state, and the nth
∗
module outputs the desired power Pn = Pgrid /3; if
∗ A. PV Voltage Loop of the nth Module
Ppvn ≥ Pgrid /3, the nth module battery will be in
neither charging nor discharging state, and the nth Assume that small variations are present around their equi-
module outputs its entire PV power, i.e., Pn = Ppvn . librium states in (1), namely

C. Grid Power Control xn = x̄n + x̃n , Dn = D̄n + D̃n , un = ūn + ũn

For the ac side, the grid power control is shown in Fig. 2. The where the symbols “∼” above the variables denote their small
phase angle of the grid voltage is measured by using the phase- variations around the equilibrium states. Thus, the small signal
locked loop and then is used to produce the phase reference of model of the system will be
the grid current.
⎡ Dn −1
⎤
The magnitude of the grid current reference is calculated by 0 0 Dn
L L
⎢ Dn Dn −1 ⎥
2Pgrid ⎢ 0 0 ⎥
îg = (8) x̃˙ n = ⎢ 1−Dn −Dn
L L ⎥ x̃n
v̂g ⎣ C C 0 0 ⎦
−Dn 1−Dn
C C 0 0
where Pgrid is the total grid-injected power of the three modules ⎡ 1 ⎤ ⎡ ⎤
v̄PNn
and v̂g is the magnitude of the grid voltage. L 0 0 L
Multiplying the magnitude reference by the phase reference, ⎢ 0 0 0 ⎥ ⎢ v̄PNn ⎥
+⎢
⎣ 0
⎥ ũ + ⎢ L ⎥ D̃n (10)
the grid current reference i∗g can be obtained, and it is in phase Dn −1
C 0 ⎦ n ⎣ ī11n
C
⎦
Dn −1 1 ī11n
with the grid voltage. 0 C C C
As shown in Fig. 2, the proportional and resonant (PR) con-
troller [20] assisted with a grid voltage feedforward is employed where v̄PNn = v̄C1n + v̄C2n , ī11n = −īL1n − īL2n + īPNn .
to enforce the actual current to track the grid current reference. The Laplace transformation is applied to (10), and when
∗
The output variable of the controller is the voltage vH , as shown defining k1 = sL and k2 = sC, the transfer functions from D̃n ,
in Fig. 2. The module power ratio an can be expressed by ṽpvn ,ĩPNn , and ĩbn to ĩL1n , ĩL2n , ṽC1n , and ṽC2n are written as
Pn (11), shown at the bottom of the page.
an = (9) Based on the PV panel’s V–I characteristic, the relationship
Pgrid
between PV voltage and current can be presented by
∗
vH and an are used to calculate the modulation signals mn for
each module in Fig. 2. With Dn and mn , the gate signals for vpvn = f (iL1n ). (12)

⎧
⎪
⎪ ĩL1n = (2D̄(2nD̄−1) ī11n +k2 v̄PNn
· D̃n + (kk21 [2 D̄n (D̄n −1)+k1 k2 +1]
· ṽpvn + (2 D̄n −1)(D̄n −1) n (k1 k2 +2D̄n −1)
· ĩPNn + (k1 k2D̄+1)[(2 · ĩbn
⎪
⎪ n −1) +k1 k2
2 k2 +1)[(2D̄n −1)2 +k1 k2 ] (2D̄n −1)2 +k1 k2 D̄n −1)2 +k1 k2 ]
⎪
⎨ĩL2n = (2D̄n −1)ī11n +k2 v̄PNn · D̃n − k2 [2D̄n (D̄n −1)] (2D̄n −1)(D̄n −1) (D̄n −1)(k1 k2 −2D̄n +1)
(2D̄n −1)2 +k1 k2 (k1 k2 +1)[(2D̄n −1)2 +k1 k2 ]
· ṽpvn + (2D̄n −1)2 +k1 k2 · ĩPNn + (k1 k2 +1)[(2D̄n −1)2 +k1 k2 ] · ĩbn
⎪ k1 ī11n −(2D̄n −1)v̄PNn (D̄n −1)[k1 k2 −(2D̄n −1)] D̄n −1) 2D̄n (D̄n −1)k1
⎪
⎪ ṽC1n = (2D̄n −1)2 +k1 k2 · D̃n − (k1 k2 +1)[(2D̄n −1)2 +k1 k2 ] · ṽpvn + (2D̄kn1 (−1) 2 +k k · ĩPNn − (k k +1)[(2D̄ −1)2 +k k ] · ĩbn
⎪
⎪
1 2 1 2 n 1 2
⎩ṽC2n = k1 ī11n −(2D̄2n −1)v̄PNn · D̃n − D̄n [k1 k2 +(2D̄n −1)]
· ṽpvn + k1 (D̄n2−1) · ĩPNn + k1 (2D̄n −2D̄n +k12k2 +1) · ĩbn
2

(2D̄n −1) +k1 k2 (k1 k2 +1)[(2D̄n −1) +k1 k2 ]

2 (2D̄n −1) +k1 k2 (k1 k2 +1)[(2D̄n −1) +k1 k2 ]
(11)
5462 IEEE TRANSACTIONS ON INDUSTRIAL ELECTRONICS, VOL. 62, NO. 9, SEPTEMBER 2015

Fig. 3. Block diagram of the nth module’s PV voltage loop.

We can get the small signal model of the PV panel as [15]

ṽpvn = −Rn · ĩL1n (13)
where Rn is the PV panel’s resistance.
Battery current follows (7) and is considered as external input
in designing the PV voltage controller according to (11). From
(11) and (13), we get the transfer function from Dn to vpvn as

Gpvdn (s)
−Rn · (k1 k2 + 1)[(2D̄n − 1)ī11n + k2 v̄PNn ]
= .
(k1 k2+1)[(2D̄n −1)2+k1 k2+ k2 Rn ]+k2 Rn·2D̄n (D̄n−1)
(14)
The PV voltage control block diagram of the nth module
is shown in Fig. 3, where the PI regulator is employed. The
transfer function of the PI regulator is
kpn s + kin
GPIn (s) = . (15)
s
The battery current’s effect on PV voltage control will be
compensated by the PI regulator, where the battery current’s
variation is a disturbance in the PV voltage control system.
The cutoff angular frequency of Gpvdn (s) is around
250 rad/s, so the cutoff angular frequencies of GPIn and
Gpvdn (s)GPIn (s) are set at 250 and 25 rad/s, respectively.
Then, we can calculate the parameters of the PI regulator,
i.e., kpn = −0.0039 and kin = −0.975. The Bode diagrams
of Gpvdn (s), GPIn (s), and Gpvdn (s)GPIn (s) are shown in
Fig. 4(a), respectively. The compensated open-loop transfer
function indicates that the closed-loop system is stable, as
shown in Fig. 4(b). Fig. 4. (a) Bode diagram of Gpvdn (s), GPIn (s), and Gpvdn (s)
GPIn (s), respectively. (b) Bode diagram of PV voltage closed-loop
B. Grid Current Loop transfer function.

The grid-current control block diagram is shown in Fig. 5.

A PR regulator is employed to track the sinusoidal reference of
the grid current.
The transfer function of the PR regulator is
kp s2 + kr s + kp ω02
GPR (s) = . (16)
s2 + ω02
Fig. 5. Block diagram of the grid current loop.
Moreover, a grid voltage feedforward control is introduced
The transfer function of the nth qZSI module is
to compensate the disturbance of the grid voltage and achieve
the fast dynamic response, as shown in Fig. 5. Ginvn = v̂PNn . (19)
For the system shown in Fig. 1, each module’s modulation
signal mn will be Based on the control scheme in Fig. 5, there is
Pn · vH∗  3

mn = . (17) Pn 1
Pgrid · v̂PNn · · Ginvn = 1. (20)
n=1
Pgrid v̂PNn
From (3), the energy stored qZS-CMI-based grid-tie PV
Therefore, the block diagram in Fig. 5 can be simplified into
system has the transfer function
the block diagram shown in Fig. 6.
Ig (s) 1 As a tradeoff between the stability and dynamic response, the
GL (s) = = . (18)
VH (s) − Vg (s) Lf s + rf parameters of the PR regulator are selected as follows: kp = 20,
SUN et al.: ENERGY STORED QZS-CMI-BASED PHOTOVOLTAIC POWER GENERATION SYSTEM
Fig. 6. Simplified block diagram of the grid current loop.
Fig. 7. (a) Bode diagrams of GL (s), GPR (s), and GL (s)GPR (s), re-
spectively. (b) Bode diagram of grid current closed-loop transfer function.
kr = 20, and ω0 = 314 rad/s. The Bode diagrams of GL (s),
GPR (s), and GL (s)GPR (s) are shown in Fig. 7(a), respectively.
The Bode diagram of the grid-current closed-loop transfer
function is shown in Fig. 7(b).
V. L EAKAGE C URRENT S UPPRESSION
Leakage current suppression is a concern in transformer-less
PV power systems. For utility-scale PV farms, due to high
voltage level, PV panels have to be mounted on a structure
with insulation, which also avoids the leakage current to the
ground. For small-scale PV power generation systems, such as
residential PV systems, the leakage current may flow to the
ground because of the requirement of grounding PV panels,
so it will require a special design to limit the leakage current
within a specified range.
5463
The same situations exist for the proposed energy stored qZS-
CMI-based PV power system. For utility-scale application, the
PV panels’ high voltage insulation from the ground has to be
taken into account, and hence, there is no leakage current issue.
In residential applications, special approaches are necessary
to handle the leakage current issue. Zhou and Li [21] ana-
lyzed the leakage current model of the CMI-based PV power
system and proposed two methods used for high switching
frequency and low switching frequency systems, respectively.
Both methods are suitable for the qZS CMI-based PV power
system. Referring to [21], the filter-based solution will help
to suppress leakage current through employing extra dc-side
common mode (CM) chokes, CM capacitors, and ac-side CM
chokes, but further study is inevitable to minimize the size.
VI. S IMULATION R ESULTS
Before doing experiments, simulations are carried out to
verify the proposed energy stored qZS-CMI-based PV system
and control scheme. Simulation parameters are as follows: the
rated PV panel power is 496.4 W, PV voltage range is 45–90 V,
carrier frequency is fc = 10 kHz, L1 = L2 = 500 μH, C1 =
C2 = 4400 μF, Lf = 500 μH, rf = 0.01Ω, vbattery = 36 V,
∗
Pgrid = 1500 W, and each module’s power reference is 500 W.
The grid voltage is vg = 220 V, and the grid frequency is
fg = 50 Hz. The operational conditions of three PV panels
are as follows: During 0–1 s, all three PV panels are kept at
35 ◦ C and 1050 W/m2 , while at t = 1 s, the conditions of
PV panels 1 and 3 change (i.e., PV1: 8 ◦ C and 1300 W/m2 and
PV3: 60 ◦ C and 900 W/m2 ). This change is only for testing the
system; in practical systems, the temperature change cannot be
so large and so sudden.
The output voltages and powers of three PV panels are
shown in Fig. 8(a). During 0–1 s, all PV panels output the
same voltage and the same power, while during 1–2 s, they
are obviously different. However, each module still provides the
desired power of 500 W with the battery operating in different
states. Three battery currents are also shown in Fig. 8(a), where
the current is positive with battery discharging, negative with
battery charging, and zero for no charging/no discharging.
Figs. 8(b) and (c) and 9(a) show inductor currents, capacitor
voltages, and dc-link voltages of three modules, which verifies
the operating principle of the energy stored qZS-CMI-based PV
power system.
The grid voltage and current are shown in Fig. 9(b), from
which we can see that the grid current has the same phase with
the grid voltage and the total grid-injected power is kept at
the desired value of 1500 W. Fig. 9(b) also shows the seven-
level output voltage of energy stored qZS-CMI. Fig. 9(c) shows
three modulation signals of the system at the time interval
of 1.8–2 s. With the proposed control scheme in Section III,
three modulation signals mn present little difference due to the
different dc-link voltages of three modules.
VII. E XPERIMENTAL R ESULTS
A set of experiments is carried out to verify the proposed
energy stored qZS-CMI-based PV system and control scheme.
5464 IEEE TRANSACTIONS ON INDUSTRIAL ELECTRONICS, VOL. 62, NO. 9, SEPTEMBER 2015

Fig. 8. Simulation results of the proposed system (to be continued Fig. 9. Simulation results of the proposed system (continued from
in Fig. 9). (a) PV voltages, PV currents, and battery currents of three Fig. 8). (a) DC-link voltages of three modules. (b) Grid voltage, grid
modules. (b) qZS network inductor currents of three modules. (c) qZS current, and seven-level voltage of the system. (c) Modulation indexes
network capacitor voltages of three modules. of three modules.

The experimental bench is shown in Fig. 10. Each module’s energy stored qZS-CMI-based PV system is connected to the
qZS network has the same parameters as the simulated system. 220-V grid through the L-filter and transformer.
Three PV panels with the type of KD135GH-2P are employed, A TMS320F28335-based control board is used to fulfill the
and 12-V lead–acid batteries are for the energy storage. The control scheme and algorithms. There are modules EPWMx
SUN et al.: ENERGY STORED QZS-CMI-BASED PHOTOVOLTAIC POWER GENERATION SYSTEM
Fig. 10. Experimental bench of the energy stored qZS-CMI-based
PV power system. (a) Whole system. (b) One module (battery will be
paralleled to capacitor C2 in tests).
Fig. 11. Experimental results when batteries are operating in neither
charging nor discharging mode. (a) PV voltages, PV currents, and bat-
tery currents of three modules. (b) DC-link voltage, seven-level voltage,
grid voltage, and grid current.
(x = 1, 2, . . . , 6) in TMS320F28335 to control the six bridge
legs of the qZS-CMI system, respectively. EPWM1 is set to
have a 180◦ phase shift compared to EPWM2 for the first qZS
H-bridge module, EPWM3 and EPWM4 have the 180◦ phase
shift for the second qZS H-bridge module, and EPWM5 and
5465
Fig. 12. Experimental results when batteries operating in charging
mode. (a) PV voltages, PV currents, and battery currents of three
modules. (b) DC-link voltage, seven-level voltage, grid voltage, and grid
current.
EPWM6 are set in the same way for the third qZS H-bridge
module in Fig. 1. As a result, the two legs of each qZS H-bridge
module have 180◦ -phase-shift carriers. In addition, EPWM1,
EPWM3, and EPWM5 should have a 60◦ phase shift from each
other, which are corresponding to three carriers for three qZS
H-bridge modules. Two compare registers CMA and CMB in
each EPWMx module are used for the sinusoidal signal mn and
shoot-through signal Dn , respectively. Finally, the produced
12 gate signals drive the six bridge legs of the qZS-CMI system,
respectively.
Fig. 11 shows the experimental result of the proposed system
operating in neither charging nor discharging mode. During
the test, the total grid-injected power reference is 550 W,
and the power reference of each module is 183.3 W. With
distributed MPPT control, three modules’ PV panels operate at
their maximum power points (MPPs), and three battery currents
are 0.25, −0.3, and −0.08 A, respectively.
5466 IEEE TRANSACTIONS ON INDUSTRIAL ELECTRONICS, VOL. 62, NO. 9, SEPTEMBER 2015
Fig. 13. Experimental results when batteries are operating in discharg-
ing mode. (a) PV voltages, PV currents, and battery currents of three
modules. (b) DC-link voltage, seven-level voltage, grid voltage, and grid
current.
Fig. 11(a) shows the PV voltages, PV currents, and battery
currents of three modules. Each module’s PV panel provides
the desired grid-injected power of 183.3 W and its circuit
power loss, so all three batteries operate in neither charging nor
discharging mode. Fig. 11(b) shows the dc-link voltage, seven-
level voltage, grid voltage, and grid current, where the dc-link
voltage is boosted to a higher voltage level when compared to
the PV voltage and the whole system operates in unity power
factor.
Fig. 12 shows the experimental results of the proposed
system operating in charging mode. During the test, the desired
grid power is 450 W, and each module needs to provide 150 W.
As shown in Fig. 12(a), the three PV panels all operate at their
MPPs and provide 225.8, 220.1, and 229.7 W, respectively. As
each module provides more power than desired, all of the bat-
teries of the three modules operate in charging states. The bat-
tery currents are −4.42, −4.69, and −3.94 A, respectively.
Fig. 12(b) shows the dc-link voltage, seven-level voltage, grid
voltage, and grid current of the system in charging state.
Fig. 13 shows the experimental result of the proposed system
operating in discharging mode. During the test, the desired grid
power is 630 W, and each module needs to provide 210 W. All
three PV panels operate at their MPPs, but the grid requires
more power than the power provided by PV panels, and three
batteries in the system discharge in 1.63, 1.61, and 1.32 A,
respectively, to compensate the power difference.
VIII. C ONCLUSION
This paper has proposed an energy stored qZS-CMI-based
PV power generation system. Its operating principle was pre-
sented in detail, and its control scheme was proposed. The con-
troller parameters were well designed by employing the built
small-signal model and Bode plots. A seven-level energy stored
qZS-CMI-based PV system prototype was built. The simulation
and experimental results verified the proposed energy stored
qZS-CMI-based PV system and the proposed control method.
Leakage current suppression is an important topic that will
be researched next for the proposed system.
R EFERENCES
[1] G. Buticchi, D. Barater, E. Lorenzani, C. Concari, and G. Franceschini,
“A nine-level grid-connected converter topology for single-phase trans-
formerless PV systems,” IEEE Trans. Ind. Electron., vol. 61, no. 8,
pp. 3951–3960, Aug. 2014.
[2] H. Abu-Rub, M. Malinowski, and K. Al-Haddad, Power Electronics for
Renewable Energy Systems, Transportation and Industrial Applications.
Hoboken, NJ, USA: Wiley, Jul. 2014.
[3] S. Rivera et al., “Cascaded H-bridge multilevel converter multistring
topology for large scale photovoltaic systems,” in Proc. IEEE Int. Symp.
Ind. Electron., 2011, pp. 1837–1844.
[4] F. Peng, M. S. Shen, and K. Holland, “Application of Z-source inverter for
traction drive of fuel cell—Battery hybrid electric vehicles,” IEEE Trans.
Power Electron., vol. 22, no. 3, pp. 1054–1061, May 2007.
[5] M. Shen et al., “Constant boost control of the Z-source inverter to mini-
mize current ripple and voltage stress,” IEEE Trans. Ind. Appl., vol. 42,
no. 3, pp. 770–778, May/Jun. 2006.
[6] E. Karaman, M. Farasat, and A. M. Trzynadlowski, “A comparative study
of series and cascaded Z-source matrix converters,” IEEE Trans. Ind.
Electron., vol. 61, no. 10, pp. 5164–5173, Oct. 2014.
[7] L. Liu, H. Li, Y. Zhao, X. He, and Z. Shen, “1 MHz cascaded Z-source
inverters for scalable grid-interactive photovoltaic (PV) applications us-
ing GaN device,” in Proc. IEEE Energy Convers. Congr. Expo., 2011,
pp. 2738–2745.
[8] Y. Zhou, L. Liu, and H. Li, “A high performance photovoltaic module-
integrated converter (MIC) based on cascaded quasi-z-source inverters
(qZSI) using eGaN FETs,” IEEE Trans. Power Electron., vol. 28, no. 6,
pp. 2727–2738, Jun. 2013.
[9] D. Sun et al., “A new grid-connected PV system based on cascaded
H-bridge quasi-Z source inverter,” in Proc. IEEE Int. Symp. Ind. Electron.,
2012, pp. 951–956.
[10] Y. Liu, B. Ge, H. Abu-Rub, and F. Peng, “Phase-shifted pulse-width-
amplitude modulation for quasi-Z-source cascade multilevel inverter
based photovoltaic power system,” IET Power Electron., vol. 7, no. 6,
pp. 1444–1456, Jun. 2014.
[11] Y. Xue, B. Ge, and F. Z. Peng, “Reliability, efficiency, and cost compar-
isons of MW-scale photovoltaic inverters,” in Proc. IEEE Energy Convers.
Congress Expo., 2012, pp. 1627–1634.
[12] Y. Liu, B. Ge, H. Abu-Rub, and F. Z. Peng, “A modular multilevel space
vector modulation for photovoltaic quasi-Z-source cascade multilevel in-
verter,” in Proc. 28th Annu. IEEE Appl. Power Electron. Conf. Expo.,
2013, pp. 714–718.
[13] D. Velasco de la Fuente, C. L. T. Rodríguez, G. Garcerá, E. Figueres, and
R. O. Gonzalez, “Photovoltaic power system with battery backup with
grid-connection and islanded operation capabilities,” IEEE Trans. Ind.
Electron., vol. 60, no. 4, pp. 1571–1581, Apr. 2013.
SUN et al.: ENERGY STORED QZS-CMI-BASED PHOTOVOLTAIC POWER GENERATION SYSTEM
[14] D. Sun, B. Ge, D. Bi, and F. Peng, “Analysis and control of quasi-Z source
inverter with battery for grid-connected PV system,” Int. J. Electr. Power
Energy Syst., vol. 46, pp. 234–240, Mar. 2013.
[15] B. Ge et al., “An energy stored quasi-Z-source inverter for application to
photovoltaic power system,” IEEE Trans. Ind. Electron., vol. 60, no. 10,
pp. 4468–4481, Oct. 2013.
[16] J. G. Cintron, Y. Li, S. Jiang, and F. Peng, “Quasi-Z-source inverter with
energy storage for photovoltaic power generation systems,” in Proc. 26th
Annu. IEEE Appl. Power Electron. Conf. Expo., 2011, pp. 401–406.
[17] Y. Liu, B. Ge, H. Abu-Rub, and F. Peng, “Control system design of
battery-assisted quasi-Z-source inverter for grid-tie photovoltaic power
generation,” IEEE Trans. Sustain. Energy, vol. 4, no. 4, pp. 994–1001,
Oct. 2013.
[18] M. A. G. de Brito, L. Galotto, L. P. Sampaio, G. de Azevedo e Melo, and
C. A. Canesin, “Evaluation of the main MPPT techniques for photovoltaic
applications,” IEEE Trans. Ind. Electron., vol. 60, no. 3, pp. 1156–1167,
Mar. 2013.
[19] D. Sun, B. Ge, H. A Rub, F. Peng, and A. T. de Almeida, “Power
flow control for quasi-Z source inverter with battery based PV power
generation system,” in Proc. IEEE Energy Convers. Congr. Expo., 2011,
pp. 1051–1056.
[20] L. F. A. Pereira, J. V. Flores, G. Bonan, D. F. Coutinho, and
J. M. G. da Silva, “Multiple resonant controllers for uninterruptible power
supplies—A systematic robust control design approach,” IEEE Trans. Ind.
Electron., vol. 61, no. 3, pp. 1528–1538, Mar. 2014.
[21] Y. Zhou and H. Li, “Analysis and suppression of leakage current
in cascaded-multilevel-inverter-based PV systems,” IEEE Trans. Power
Electron., vol. 29, no. 10, pp. 5265–5277, Oct. 2014.
Dongsen Sun is working toward the Ph.D. de-
gree in electrical engineering at Beijing Jiaotong
University, Beijing, China.
Her research interests include electrical ma-
chine and drive, renewable energy generation,
Z-source converter/inverter, etc.
Baoming Ge (M’11) received the Ph.D. degree
in electrical engineering from Zhejiang Univer-
sity, Hangzhou, China, in 2000.
Since 2002, he has been with Beijing
Jiaotong University, Beijing, China, as a Pro-
fessor in the School of Electrical Engineering.
He is also with the Department of Electrical
and Computer Engineering (ECE), Texas A&M
University, College Station, TX, USA. He was a
Visiting Scholar with the University of Coimbra,
Coimbra, Portugal, from 2004 to 2005, and a
Visiting Professor with ECE, Michigan State University, East Lansing,
MI, USA, from 2007 to 2008 and 2010 to 2014. His main research in-
terests are the renewable energy generation, electrical machine drives,
and power electronics.
5467
Weihua Liang is working toward the Ph.D. de-
gree in electrical engineering at Beijing Jiaotong
University, Beijing, China.
His research interests include electrical ma-
chines and drives, renewable energy genera-
tion, Z-source converter/inverter, etc.
Haitham Abu-Rub (M’99–SM’07) received two
Ph.D. degrees from Gdańsk University of Tech-
nology, Gdańsk, Poland, in electrical engineer-
ing and humanities.
For eight years, he was with Birzeit Univer-
sity, Birzeit, Palestine, where he was first an
Assistant Professor and then an Associate Pro-
fessor and was the Chairman of the Department
of Electrical Engineering for four years. He is
currently a Full Professor with the Department
of Electrical and Computer Engineering, Texas
A&M University at Qatar, Doha, Qatar, and a Chief Scientist with Qatar
Environment and Energy Research Institute (QEERI), Qatar Founda-
tion, Doha, Qatar. His main research interests are in energy conver-
sion systems, including renewable and electromechanical systems. His
work relates to the high performance electric drives, renewable energy
systems, multilevel inverters, impedance source inverters and modifica-
tions, pulse width modulation techniques, model predictive control, and
other related topics. He is the author of more than 200 conference and
journal papers and coauthor of four books.
Prof. Abu-Rub is an Editor of the IEEE T RANSACTIONS ON S USTAIN -
ABLE E NERGY and the IEEE J OURNAL OF E MERGING AND S ELECTED
TOPICS IN P OWER E LECTRONICS. He was a recipient of many interna-
tional prestigious awards such as the American Fulbright Scholarship
(at Texas A&M University, College Station, TX. USA), the German
Alexander von Humboldt Fellowship (at University of Wuppertal, Wup-
pertal, Germany), the German Academic Exchange Service (DAAD)
Scholarship (at Ruhr University Bochum, Bochum, Germany), and the
British Royal Society Scholarship (at the University of Southampton,
Southampton, U.K.).
Fang Zheng Peng (M’92–SM’96–F’05) re-
ceived the Ph.D. degree from Nagaoka Univer-
sity of Technology, Nagaoka, Japan, in 1990.
From 1990 to 1992, he was a Research
Scientist with Toyo Electric Manufacturing Com-
pany, Ltd., Tokyo, Japan. From 1992 to 1994,
he was a Research Assistant Professor with
Tokyo Institute of Technology, Tokyo. From 1994
to 1997, he was a Research Assistant Professor
with The University of Tennessee, Knoxville, TN,
USA, where he was also a Staff Member. From
1994 to 2000, he was with the Oak Ridge National Laboratory, where,
from 1997 to 2000, he was the Lead (Principal) Scientist with the Power
Electronics and Electric Machinery Research Center. Since 2000, he
has been with Michigan State University, East Lansing, MI, USA, where
he is currently a Professor with the Department of Electrical and Com-
puter Engineering. His research interests include power electronics,
renewable energy, and energy storage.
Dr. Peng has served the IEEE Power Electronics Society in many
capacities such as the Chair of the Technical Committee for Rectifiers
and Inverters, an Associate Editor of the IEEE Transactions on Power
Electronics, Region 1-6 Liaison, and Member-at-Large. He was a recip-
ient of many awards.