4234 IEEE TRANSACTIONS ON INDUSTRIAL ELECTRONICS, VOL. 60, NO.

10, OCTOBER 2013

Adjustable Proportional Hybrid SVPWM Strategy
for Neutral-Point-Clamped Three-Level Inverters
Changliang Xia, Senior Member, IEEE, Hongjun Shao, Yun Zhang, and Xiangning He, Fellow, IEEE
AbstractNeutral-point (NP)-clamped three-level inverter,
which is a widely used topology of multilevel converters, suffers
from the NP voltage drift as its main technical drawback. The
nearest-three-virtual-vector (NTV
2
) modulation method can con-
trol the NP voltage balance for any load over the full range of
inverter output voltage. However, compared with the nearest-
three-vector (NTV) modulation method, NTV
2
increases the
switching frequency. This paper combines NTV
2
with NTV, and
the two methods are used alternately in a fundamental cycle. The
duty-cycle coefcients of the redundant small vectors are analyzed
to explore the intervals where NTV can control the NP voltage
balance in a fundamental cycle. The proportional parameter,
which represents the combination of the two methods in hybrid
space vector pulsewidth modulation, can be selected to achieve
a high-performance NP balance control and low switching fre-
quency. The steady- and transient-state behaviors of the proposed
strategy are analyzed in this paper. The experimental results verify
the validity of the proposed strategy.
Index TermsHybrid modulation, multilevel converters,
nearest-three-virtual-vector (NTV
2
), neutral-point (NP)-clamped
(NPC) three-level inverters, NP voltage balance control, space
vector pulsewidth modulation (SVPWM).
I. INTRODUCTION
M
ULTILEVEL converters have attracted more and more
attention in high-power applications due to low voltage
across the power device, improved output voltage quality, and
low switching frequency of the power device compared with the
traditional two-level converters [1]. Among them, the neutral-
point (NP)-clamped (NPC) three-level inverter, in Fig. 1, is the
most widely used multilevel converter topology, which only
needs one dc-voltage source and has low hardware requirement
[2][8].
Manuscript received January 4, 2012; revised May 11, 2012 and July 20,
2012; accepted July 26, 2012. Date of publication August 16, 2012; date of
current version May 16, 2013. This work was supported in part by the Project
supported by the National Basic Research Program of China (973 Project)
under Grant 2013CB035602 and in part by the Key Program of the National
Natural Science Foundation of China under Grant 51037004.
C. Xia is with the School of Electrical Engineering and Automation, Tianjin
University, Tianjin 300072, China, and also with the Tianjin Key Laboratory of
Advanced Technology of Electrical Engineering and Energy, Tianjin Polytech-
nic University, Tianjin 300387, China (e-mail: motor@tju.edu.cn).
H. Shao and Y. Zhang are with the School of Electrical Engineering and
Automation, Tianjin University, Tianjin 300072, China (e-mail: shaohongjun@
tju.edu.cn; zhangy@tju.edu.cn).
X. He is with the Power Electronics Research Institute, Zhejiang University,
Hangzhou 310027, China (e-mail: hxn@zju.edu.cn).
Color versions of one or more of the gures in this paper are available online
at http://ieeexplore.ieee.org.
Digital Object Identier 10.1109/TIE.2012.2213558
Fig. 1. Topology of three-level NPC inverter.
The NP voltage drift is the main technical drawback of NPC
three-level inverters. At present, there are four main methods
for the control of the NP voltage balance.
1) Use two isolated dc-voltage sources at the dc side, which
can be provided by two independent rectiers [9], [10].
2) Control the NP voltage balance through the front-end
circuit, such as back-to-back topology or three-level boost
circuit [11], [12].
3) Inject or extract current into or from the NP through the
extra converters [13].
4) Improve the modulation algorithm itself to control the NP
voltage balance [14][18].
Compared with the rst three approaches, the fourth method
can control the NP balance without extra hardware equipment
or control system; thus, it saves the cost and reduces the system
bulk.
The nearest-three-vector (NTV) method is one of the most
commonly used multilevel space vector pulsewidth modulation
(SVPWM) methods, which has the advantages of low switching
frequency and good output voltage quality. The disadvantage
of the NTV method is the presence of low-frequency voltage
oscillations on the NP under the conditions of high modulation
index and low power factor, which results in low-order har-
monics in the output voltage. In order to compensate for the
drawback, some researchers have presented the new-selected-
three-vector method [19] and the nearest-three-virtual-vector
(NTV
2
) method [20], [21], both of which can, on the one hand,
control the NP voltage balance over the full range of inverter
output voltage, but on the other hand, increase the switching
frequency of the power device.
0278-0046/$31.00 2012 IEEE
XIA et al.: ADJUSTABLE PROPORTIONAL HYBRID SVPWM STRATEGY FOR NPC THREE-LEVEL INVERTERS 4235
In [22], a hybrid modulation strategy combining NTV with
NTV
2
can control the NP voltage balance for all operation
conditions. In this hybrid strategy, NTV is adopted in the in-
tervals where NTV can control NP voltage balance, and NTV
2
is utilized in the intervals where NTV fails to control NP volt-
age balance. Compared with NTV
2
, this constant proportional
hybrid SVPWM (CPH-SVM) method not only can control NP
voltage balance but also can reduce the switching frequency.
This paper presents an adjustable proportional hybrid
SVPWM (APH-SVM) strategy, which combines NTV
2
with
NTV. The two methods are used alternately in a fundamental
cycle. Based on the analysis of the duty-cycle coefcients of
redundant small vectors, the intervals where NTV can control
the NP voltage balance in a fundamental cycle are explored.
The proportional parameter, which represents the combination
of the two methods in hybrid SVPWM, can be selected to min-
imize the switching frequency within the allowable NP ripple
amplitude range. In order to provide the basis for selecting
the proportional parameter, the performances of the proposed
method, which are dependent on the proportional parameter,
are analyzed. The steady- and transient-state behaviors of the
proposed strategy are studied by experiments to verify the
validity of the proposed technique.
II. NTV AND NTV
2
A. NTV
Fig. 2(a) shows the space-vector diagramof three-level topol-
ogy, which is divided into six sextants. P, O, and N rep-
resent the output voltages U
dc
/2, 0, and U
dc
/2, respectively,
where U
dc
is the dc-link voltage.
Fig. 2(b) shows the NTV space-vector diagram in the rst
sextant, and the sextant is divided into six triangles, where V
0
represents the zero vector,V
S1
and V
S2
are small vectors, V
M
represents the middle vector, V
L1
and V
L2
are large vectors,
and is the reference-vector rotating angle. The NP currents
produced by the switching states are shown in the square
brackets. When the reference vector is located in triangle 5,
the voltage-second balance expression is as follows:

V
ref
= V
S1
d
S1
+V
M
d
M
+V
L1
d
L1
d
S1
+ d
M
+ d
L1
= 1
(1)
where V
ref
represents the reference voltage vector and d
S1
,
d
M
, and d
L1
represent the duty cycles of the vectors V
S1
, V
M
,
and V
L1
, respectively. The voltage-second balance expressions
in other triangles are similar to that in triangle 5.
This paper adopts the three-vector four-state synthesis
method, i.e., each reference vector is composed of four switch-
ing states from three vectors. The rst vector of the switching
sequence is ONN in triangle 1, 3, or 5 and OON in triangle
2, 4, or 6. For example, the switching sequence is ONN
PNN PON POO when the reference vector is located in
triangle 5 and OON PON PPN PPO when in triangle
6. Moreover, the switching sequence of the next sample period
is the inverse sequence of the previous one.
The duty cycle of positive small vector V
S1(ONN)
is k
P1

d
S1
, and the duty cycle of negative small vector V
S1(POO)
Fig. 2. Space-vector diagram. (a) Three-level topology. (b) NTV in the rst
sextant. (c) NTV
2
in the rst sextant.
is (1 k
P1
) d
S1
, where k
P1
[0, 1] will be called the duty-
cycle coefcient of redundant small vectors in the following
text.
When the reference vector is located in the rst sextant, the
average NP current in one sample cycle is
i
NP
= (2k
P1
1) d
S1
i
a
+ d
M
i
b
+ (2k
P2
1) d
S2
i
c
(2)
where i
NP
represents the average NP current, i
a
, i
b
, and i
c
are
the three-phase load currents, and k
P1
and k
P2
are the duty-
cycle coefcients of redundant small vectors V
S1
and V
S2
,
respectively.
From (2), it can be seen that the NP current can be controlled
by selecting the values of k
P1
and k
P2
. In the case of high
modulation index and low power factor, the NP current cannot
4236 IEEE TRANSACTIONS ON INDUSTRIAL ELECTRONICS, VOL. 60, NO. 10, OCTOBER 2013
be controlled to be zero in part intervals of the fundamental
cycle. These intervals are called NTV uncontrollable balance
(NTV-UCB) intervals, and the other intervals are called NTV
controllable balance (NTV-CB) intervals.
B. NTV
2
The NTV
2
method was rst presented in [20], and then,
many researchers develop this method [22], [23]. The main idea
is to synthesize the new vectors by small vectors and middle
vectors. The synthesis vectors are named virtual space vectors,
which can produce zero average NP current. Thus, NTV
2
can
control NP voltage balance for any load.
Fig. 2(c) shows the NTV
2
space-vector diagram in the rst
sextant, where V
Z0
represents the virtual zero vector, V
ZS1
and V
ZS2
are virtual small vectors, V
ZM1
represents the
virtual middle vector, and V
ZL1
and V
ZL2
are virtual large
vectors. The sextant is divided into ve triangles. The rst
vector of the switching sequence in each triangle is ONN,
and the last is PPO. For example, when the reference vector
is located in triangle 3, the switching sequence is ONN
PNN PON POO PPO.
The virtual middle vector is dened as
V
ZM1
=
1
3
V
S1(ONN)
+
1
3
V
M(PON)
+
1
3
V
S2(PPO)
. (3)
From (3), it is known that the virtual middle vector V
ZM1
is composed of basic vectors V
S1(ONN)
, V
M(PON)
, and
V
S2(PPO)
. The durations of the three vectors are equal, which
are all one-third of the duration of the virtual middle vector.
In the three-phase system without midline, the relationship
of three-phase currents is
i
a
+ i
b
+ i
c
= 0. (4)
It is known from Fig. 2(b) that the NP currents produced
by switching states V
S1(ONN)
, V
M(PON)
, and V
S2(PPO)
are
i
a
, i
b
, and i
c
, respectively, so the NP current produced by the
virtual middle vector is 1/3(i
a
+ i
b
+ i
c
) = 0, i.e., the virtual
middle vector produces zero NP current.
The denitions of other virtual vectors are similar to that of
the virtual middle vector, and all virtual vectors produce zero
NP current. The vectors in other sextants can all be mapped into
the rst sextant through vector-rotating conversion. The deni-
tions of virtual vectors are the same as that of the rst sextant.
III. PROPOSED HYBRID SVPWM STRATEGY
This paper takes the rst sextant, for example, to explain the
APH-SVM strategy. The NP current i
NP
in (2) is set to zero,
and k

P1
and k

P2
, which are the needed values to control the
NP balance, can be solved from (2). The effective range of k
P1
and k
P2
is [0, 1], so when k

P1
is within [0, 1], k
P1
= k

P1
;
when k

P1
is greater than one, k
P1
= 1; when k

P1
is less than
zero, k
P1
= 0; and k
P2
has the same solving method. For
example, when the reference vector is located in triangle 5 in
Fig. 2(b), simplify (2) to obtain the following expression:
i
NP
= (2k
P1
1) d
S1
i
a
+ d
M
i
b
. (5)
Substituting i
NP
= 0 into (5), the following is obtained:
k

P1
=
d
S1
(i
a
) d
M
(i
b
)
2 d
S1
(i
a
)
. (6)
When the reference vector is located in triangle 6 in Fig. 2(b),
simplify (2) to obtain the following expression:
i
NP
= (2k
P2
1) d
S2
i
c
+ d
M
i
b
. (7)
Substituting i
NP
= 0 into (7), the following is obtained:
k

P2
=
d
S2
(i
c
) d
M
(i
b
)
2 d
S2
(i
c
)
(8)
where the expressions of d
S1
, d
M
, d
S2
, i
a
, i
b
, and i
c
are
explained in [15] in detail.
Fig. 3 shows the obtained k

P1
and k

P2
when modulation
index m = 0.95 and power-factor angle = [0

, 180

], where
m [0, 1] is the ratio of the amplitude of the reference vector to
the radius of the inscribed circle, i.e., m = |V
ref
|/(U
dc
/

3).
The duty-cycle coefcients k

P1
and k

P2
when m = 0.95
and = 45

are shown in Fig. 4. This paper takes this

operation condition, for example, to explain the APH-SVM
strategy.
It can be seen from Fig. 4 that, in intervals 1 and 4,
0 k

P1
1 and 0 k

P2
1; in interval 2, k

P1
1 and
0 k

P2
1; and, in interval 3, 0 k

P1
1 and k

P2
0.
The requirements for the NP balance control are 0 k

P1
1
and 0 k

P2
1. Thus, in intervals 1 and 4, the NP balance
control can be achieved by NTV, and these intervals are NTV-
CB intervals; in intervals 2 and 3, the NP balance control
cannot be achieved by NTV, and these intervals are NTV-UCB
intervals.
The NTV-CB intervals in the rst sextant are [0

,
1
] and
[
2
, 60

], as shown in Fig. 5.
1
and
2
can be obtained by
judging the values of k

P1
and k

P2
.
The CPH-SVM strategy uses NTV in the NTV-CB intervals
and NTV
2
in the NTV-UCB intervals. Thus, the CPH-SVM
strategy can control the NP balance in the whole fundamental
cycle, and the combination of the two methods is constant.
The APH-SVM strategy, presented in this paper, extends the
NTV action intervals of CPH-SVM to make the NTV action in-
tervals of APH-SVM include the whole NTV-CB intervals and
part of the NTV-UCB intervals. A variable p
m
is proposed in
this paper, which represents the ratio of the extension intervals
to the whole NTV-UCB intervals.
The size of the NTV-UCB interval in the rst sextant is

3
=
2

1
. (9)
The extension intervals in the rst sextant are
41
and
42
,
respectively, as shown in Fig. 5.
When the proportional parameter is p
m
, the sum of the
extension intervals,
4
, is

4
=
41
+
42
= p
m

3
. (10)
As shown in Fig. 5, the NTV action intervals after extension
are [0

,
5
] and [
6
, 60

]. When the left and right NTV action

XIA et al.: ADJUSTABLE PROPORTIONAL HYBRID SVPWM STRATEGY FOR NPC THREE-LEVEL INVERTERS 4237
Fig. 3. Duty-cycle coefcients k

P1
and k

P2
when m = 0.95. (a) k

P1
( =
[0

, 90

]). (b) k

P1
( = [90

, 120

]). (c) k

P1
( = [120

, 180

]). (d) k

P2
( = [0

, 60

]). (e) k

P2
( = [60

, 90

]). (f) k

P2
( = [90

, 180

]).
Fig. 4. Duty-cycle coefcients k

P1
and k

P2
when m = 0.95 and = 45

.
Fig. 5. Realization of APH-SVM.
intervals are equal after extension,
5
and
6
can be obtained
from the following expression:

5
= 60

6
(
5

1
) + (
2

6
) =
4
.
(11)
In the APH-SVM strategy, the NTV method is used in the
intervals [0

,
5
] and [
6
, 60

], and the NTV

2
method is adopted
in the interval [
5
,
6
]. The proportional parameter p
m
can be
varied to adjust the size of NTV action interval. The larger the
value of p
m
is, the longer the NTV action interval is. p
m
can
take values within the interval [0, 1]. If p
m
takes the extreme
values of zero or one, the hybrid SVPWM becomes CPH-SVM
or NTV, respectively.
Fig. 6 shows the normalized ripple peak-to-peak amplitude
of NP voltage (be scaled by U
dc
) when p
m
ranges from zero to
unity. Fig. 7 shows the total harmonic distortion (THD) value
of line voltage when p
m
ranges from zero to unity.
4238 IEEE TRANSACTIONS ON INDUSTRIAL ELECTRONICS, VOL. 60, NO. 10, OCTOBER 2013
Fig. 6. Normalized ripple peak-to-peak amplitude dependent on p
m
.
Fig. 7. THD of line voltage dependent on p
m
.
From Fig. 6, it can be noted that the ripple amplitude in-
creases when p
m
approaches the unity value, and the ripple
amplitude is minimal when p
m
= 0 (CPH-SVM) and maximal
when p
m
= 1 (NTV). From Fig. 7, it can be seen that the
THD decreases when p
m
approaches the unity value, and the
THD value is maximal when p
m
= 0 and minimal when p
m
=
1. It can be concluded that the performances of the system
depend on the p
m
value, and in practical application, the p
m
value should be determined to achieve the optimal performance.
Figs. 6 and 7 provide the basis for selecting the proportional
parameter.
IV. COMPARISON OF SWITCHING FREQUENCY
The switching losses generated during the turn-on and turn-
off processes of power devices are directly related to the
switching frequency, i.e., the switching losses decrease with the
reduction of switching frequency.
Fig. 8 shows the waveform of phase voltage u
aO
when m =
0.95 and = 45

.
It can be seen from Fig. 8(a) that, in NTV, the phase voltage
commutates between P and O or between O and N
in each carrier cycle, so the switching state only changes one
time per carrier cycle; thus, the switching frequency is low. In
NTV
2
, as shown in Fig. 8(b), there are some intervals where
the phase voltage commutates among P, O, and N in a
carrier cycle, so in these intervals, the switching state changes
two times in a carrier cycle; thus, the switching frequency is
high. This phenomenon is called overlapping of level layers
Fig. 8. Phase voltage. (a) NTV. (b) NTV
2
. (c) CPH-SVM. (d) APH-SVM.
TABLE I
SWITCHING NUMBER IN ONE FUNDAMENTAL CYCLE
in [1]. In CPH-SVM, as shown in Fig. 8(c), the intervals of
overlap phenomenon reduce compared with those of NTV
2
.
In APH-SVM, as shown in Fig. 8(d), the intervals of overlap
phenomenon decrease compared with those of CPH-SVM, so
the switching frequency of the APH-SVMstrategy is lower than
those of NTV
2
and CPH-SVM.
The switching number is three in one carrier cycle when the
NTV method is used. For example, the switching sequence is
ONN PNN PON POO when the reference vector is
located in triangle 5 in Fig. 2(b), and the switching state of each
phase changes one time; thus, the sum of switching numbers
XIA et al.: ADJUSTABLE PROPORTIONAL HYBRID SVPWM STRATEGY FOR NPC THREE-LEVEL INVERTERS 4239
Fig. 9. Experimental waveforms when R = 6 . (a) NTV. (b) CPH-SVM. (c) APH-SVM.
for three phases is three. The switching number is four in one
carrier cycle when the NTV
2
method is used. For example, the
switching sequence is ONN PNN PON POO PPO
when the reference vector is located in triangle 3 in Fig. 2(c),
and the switching states of phases a and b change one time and
the switching state of phase c changes two times; thus, the sum
of switching numbers for three phases is four.
The switching number in one fundamental cycle for APH-
SVM is
N
switching
=
f
s
f

3 (60
3
+
4
) + 4 (
3

4
)
60
(12)
where N
switching
is the total switching number in a fundamental
cycle, f
s
represents the carrier frequency, and f is the funda-
mental frequency.
Equation (12) can be simplied as follows:
N
switching
=
f
s
f

3 +

3
60
(1 p
m
)

. (13)
Assuming that f = 50 Hz and f
s
= 6 kHz, the switching
numbers when p
m
= 0, p
m
= 0.5, and p
m
= 1 are shown in
Table I.
It can be seen from Table I that the switching number of the
NTV method is minimal and that of the CPH-SVM strategy
is maximal; the switching number of the APH-SVM strategy
is between the aforementioned two cases. From (13), it can be
concluded that, the larger the proportional parameter p
m
is, the
less the switching action number is.
V. EXPERIMENTAL RESULTS
An experimental prototype of a three-level NPC in-
verter is set up in this paper. The inverter is controlled
by DSP (TMS320F2812) and eld-programmable gate array
(EP1C6Q240C8), and the dc side of the inverter is supplied by
a constant dc-voltage source. The experimental parameters are
as follows: dc-link voltage U
dc
= 100 V, dc-link capacitances
C
1
= C
2
= 235 F, inductance L = 40 mH, fundamental fre-
quency f = 50 Hz, and carrier frequency f
s
= 6 kHz.
Fig. 9 shows the steady-state waveforms when the load
resistance R = 6 . The waveforms of the NTV method are
shown in Fig. 9(a), in which the phase voltage u
aO
, line voltage
u
ab
, line current i
a
, capacitor voltage U
c2
, and harmonic spectra
of u
ab
are included. Fig. 9(b) shows the waveforms of the CPH-
SVM strategy, and Fig. 9(c) shows the waveforms of the APH-
SVM strategy (when p
m
= 0.5).
It can be seen from the waveforms of U
c2
in Fig. 9 that
the NP ripple amplitude in APH-SVM is smaller than that in
NTV. Furthermore, from the waveforms of u
aO
in Fig. 9, it
can be known that the intervals of overlap phenomenon in
APH-SVM are shorter than those in CPH-SVM. The harmonic
content of the line voltage in APH-SVM is less than that in
CPH-SVM. Thus, the APH-SVM strategy achieves a high-
performance NP voltage balance and low switching frequency.
Fig. 10 shows the steady-state waveforms when R = 12.5 .
Under this load condition, the NP ripple amplitude in APH-
SVM is smaller than that in NTV, and the intervals of overlap
phenomenon in APH-SVM are less than those in CPH-SVM.
Thus, the APH-SVM strategy also achieves a small NP ripple
4240 IEEE TRANSACTIONS ON INDUSTRIAL ELECTRONICS, VOL. 60, NO. 10, OCTOBER 2013
Fig. 10. Experimental waveforms when R = 12.5 . (a) NTV. (b) CPH-SVM. (c) APH-SVM.
amplitude and low switching frequency. The obtained results
are similar to those when R = 6 . Compared with the load
condition in Fig. 9, the power factor in Fig. 10 is high, so the
NP ripple amplitudes in NTV and APH-SVM are smaller than
those in Fig. 9. From these two steady-state results, it can be
concluded that the proposed APH-SVM strategy can achieve a
high-performance NP balance and low switching frequency and
is applicable to any power-factor load.
Fig. 11 shows the transient behavior when the load varies.
Fig. 11(a) shows the case when the resistance changes from 6 to
12.5 ; Fig. 11(b) shows the case when the resistance changes
from 12.5 to 6 . It can be seen from Fig. 11(a) that the load
current reduces and the capacitor voltage increases when the
load varies, and then, they return to the normal values soon. It
can be seen from Fig. 11(b) that the load current increases and
the capacitor voltage reduces when the load varies, and then,
they return to the normal values soon. It can be concluded that
the proposed APH-SVM strategy has a good transient behavior
when the load varies.
The APH-SVM strategy is also applicable to nonlinear load.
The amplitude of NP voltage ripple increases and the NP volt-
age shifts when the NTVmethod is applied to the nonlinear load
compared with the balanced linear load. The size and position
of the NTV-CBinterval change in the case of nonlinear load. On
the other hand, the NTV
2
method can completely control the
NP voltage balance over the full range of inverter output voltage
when the nonlinear load is used, because the NTV
2
method can
guarantee the NP balance for any load and the only requirement
is i
a
+ i
b
+ i
c
= 0. The realization of the APH-SVM strategy
in the case of nonlinear load is similar to that of balanced linear
load, and the NTV-CB intervals need to be analyzed to extend
the NTV action intervals. The ripple amplitude increases and
the NP voltage shifts when the nonlinear load is used, so the
APH-SVM strategy becomes a more efcient method to reduce
the switching frequency and limit the ripple amplitude.
VI. CONCLUSION
This paper has presented a new hybrid SVPWM strategy,
which can adjust the combination of NTV and NTV
2
in the
hybrid SVPWM strategy to achieve a compromise between the
NP ripple amplitude and the switching frequency. Based on
the analysis of the duty-cycle coefcients of redundant small
vectors, the intervals where NTV can control the NP voltage
balance are obtained, and then, the NTV action intervals are
extended. The study result indicates that, with the increase
of NTV action intervals, the switching frequency gradually
decreases and the amplitude of NP voltage oscillation gradually
increases. In practical application, the combination of NTV
and NTV
2
should be determined to minimize the switching
frequency within the allowable NP ripple amplitude range. The
performance of the proposed technique, dependent on the pro-
portional parameter, has been analyzed to provide the basis for
selecting the combination of the two modulation methods. The
steady- and transient-state behaviors have been studied by ex-
periments. The steady-state results state that the proposed APH-
SVM strategy can achieve a high-performance NP balance and
low switching frequency and is applicable to any power-factor
load. The transient-state results indicate that the proposed APH-
SVM strategy has a good transient behavior.
XIA et al.: ADJUSTABLE PROPORTIONAL HYBRID SVPWM STRATEGY FOR NPC THREE-LEVEL INVERTERS 4241
Fig. 11. Experimental waveforms of APH-SVM. (a) R from 6 to 12.5 .
(b) R from 12.5 to 6 .
REFERENCES
[1] Y. Zhang and L. Sun, An efcient control strategy for a ve-level in-
verter comprising ying-capacitor asymmetric H-bridge, IEEE Trans.
Ind. Electron., vol. 58, no. 9, pp. 40004009, Sep. 2011.
[2] S. Kouro, M. Malinowski, K. Gopakumar, J. Pou, L. G. Franquelo, B. Wu,
J. Rodriguez, M. A. Perez, and J. I. Leon, Recent advances and industrial
applications of multilevel converters, IEEE Trans. Ind. Electron., vol. 57,
no. 8, pp. 25532580, Aug. 2010.
[3] L. G. Franquelo, J. Rodriguez, J. I. Leon, S. Kouro, R. Portillo, and
M. A. M. Prats, The age of multilevel converters arrives, IEEE Ind.
Electron. Mag., vol. 2, no. 2, pp. 2839, Jun. 2008.
[4] J. Rodriguez, S. Bernet, P. K. Steimer, and I. E. Lizama, A survey
on neutral-point-clamped inverters, IEEE Trans. Ind. Electron., vol. 57,
no. 7, pp. 22192230, Jul. 2010.
[5] J. Zaragoza, J. Pou, S. Ceballos, E. Robles, P. Ibaez, and J. L. Villate, A
comprehensive study of a hybrid modulation technique for the neutral-
point-clamped converter, IEEE Trans. Ind. Electron., vol. 56, no. 2,
pp. 294304, Feb. 2009.
[6] Y. C. Zhang, Z. M. Zhao, and J. G. Zhu, A hybrid PWM applied to high-
power three-level inverter-fed induction-motor drives, IEEE Trans. Ind.
Electron., vol. 58, no. 8, pp. 34093420, Aug. 2011.
[7] S. R. Pulikanti and V. G. Agelidis, Hybrid ying-capacitor-based active-
neutral-point-clamped ve-level converter operated with SHE-PWM,
IEEE Trans. Ind. Electron., vol. 58, no. 10, pp. 46434653, Oct. 2011.
[8] M. C. Cavalcanti, A. M. Farias, K. C. Oliveira, F. A. S. Neves, and
J. L. Afonso, Eliminating leakage currents in neutral point clamped
inverters for photovoltaic systems, IEEE Trans. Ind. Electron., vol. 59,
no. 1, pp. 435443, Jan. 2012.
[9] R. Sommer, A. Mertens, C. Brunotte, and G. Trauth, Medium voltage
drive system with NPC three-level inverter using IGBTs, in Proc. IEEE
PWM Medium Voltage Drives Semin., May 11, 2000, pp. 3/13/5.
[10] R. W. Menzies, P. Steimer, and J. K. Steinke, Five-level GTO inverters
for large induction motor drives, IEEE Trans. Ind. Appl., vol. 30, no. 4,
pp. 938944, Jul./Aug. 1994.
[11] J. Pou, R. Pindado, D. Boroyevich, and P. Rodriguez, Limits of the
neutral-point balance in back-to-back-connected three-level converters,
IEEE Trans. Power Electron., vol. 19, no. 3, pp. 722731, May 2004.
[12] C. L. Xia, X. Gu, T. N. Shi, and Y. Yan, Neutral-point potential
balancing of three-level inverters in direct-driven wind energy conver-
sion system, IEEE Trans. Energy Convers., vol. 26, no. 1, pp. 1829,
Mar. 2011.
[13] M. K. Mishra, A. Joshi, and A. Ghosh, Control schemes for equalization
of capacitor voltages in neutral clamped shunt compensator, IEEE Trans.
Power Del., vol. 18, no. 2, pp. 538544, Apr. 2003.
[14] A. K. Gupta and A. M. Khambadkone, A simple space vector PWM
scheme to operate a three-level NPC inverter at high modulation index
including overmodulation region, with neutral point balancing, IEEE
Trans. Ind. Appl., vol. 43, no. 3, pp. 751760, May/Jun. 2007.
[15] N. Celanovic and D. Boroyevich, Acomprehensive study of neutral-point
voltage balancing problem in three-level neutral-point-clamped voltage
source PWM inverters, IEEE Trans. Power Electron., vol. 15, no. 2,
pp. 242249, Mar. 2000.
[16] J. Pou, R. Pindado, D. Boroyevich, and P. Rodriguez, Evaluation of
the low-frequency neutral-point voltage oscillations in the three-level in-
verter, IEEE Trans. Ind. Electron., vol. 52, no. 6, pp. 15821588, Dec.
2005.
[17] J. Pou, J. Zaragoza, P. Rodriguez, S. Ceballos, V. M. Sala, R. P. Burgos,
and D. Boroyevich, Fast-processing modulation strategy for the neutral-
point-clamped converter with total elimination of low-frequency voltage
oscillations in the neutral point, IEEE Trans. Ind. Electron., vol. 54, no. 4,
pp. 22882294, Aug. 2007.
[18] H. du Toit Mouton, Natural balancing of three-level neutral-point-
clamped PWM inverters, IEEE Trans. Ind. Electron., vol. 49, no. 5,
pp. 10171025, Oct. 2002.
[19] F. B. Grigoletto and H. Pinheiro, A space vector PWM modulation
scheme for back-to-back three-level diode-clamped converters, in Proc.
COBEP, Oct. 2009, pp. 10581065.
[20] S. Busquets-Monge, J. Bordonau, D. Boroyevich, and S. Somavilla, The
nearest three virtual space vector PWMA modulation for the compre-
hensive neutral-point balancing in the three-level NPC inverter, IEEE
Power Electron. Lett., vol. 2, no. 1, pp. 1115, Mar. 2004.
[21] S. Busquets-Monge, S. Somavilla, J. Bordonau, and D. Boroyevich,
Capacitor voltage balance for the neutral-point-clamped converter us-
ing the virtual space vector concept with optimized spectral perfor-
mance, IEEE Trans. Power Electron., vol. 22, no. 4, pp. 11281135,
Jul. 2007.
4242 IEEE TRANSACTIONS ON INDUSTRIAL ELECTRONICS, VOL. 60, NO. 10, OCTOBER 2013
[22] W. D. Jiang, S. W. Du, L. C. Chang, Y. Zhang, and Q. Zhao, Hy-
brid PWM strategy of SVPWM and VSVPWM for NPC three-level
voltage-source inverter, IEEE Trans. Power Electron., vol. 25, no. 10,
pp. 26072619, Oct. 2010.
[23] J. Y. Zheng, Z. L. Shen, J. Mei, and L. F. Wang, An improved
neutral-point voltage balancing algorithm for the NPC three-level inverter
based on virtual space vector PWM, in Proc. ICECE, Jun. 2010,
pp. 32833287.
Changliang Xia (M08SM12) was born in
Tianjin, China, in 1968. He received the B.S. degree
in electrical engineering from Tianjin University,
Tianjin, in 1990, and the M.S. and Ph.D. degrees
in electrical engineering from Zhejiang University,
Hangzhou, China, in 1993 and 1995, respectively.
He is currently a Professor with the School of
Electrical Engineering and Automation, Tianjin Uni-
versity, and also with Tianjin Key Laboratory of
Advanced Technology of Electrical Engineering and
Energy, Tianjin Polytechnic University, Tianjin. His
research interests include electrical machines and their control systems, power
electronics, and control of wind generators.
Dr. Xia was recognized as a Yangtze Fund Scholar Distinguished Professor
in 2008 and is currently a recipient of the National Science Fund for Distin-
guished Young Scholars.
Hongjun Shao was born in Liaoning, China, in
1983. She received the B.S. degree in electrical engi-
neering from Tianjin Polytechnic University, Tianjin,
China, in 2005 and the M.S. degree in electrical en-
gineering from Tianjin University, Tianjin, in 2009,
where she is currently working toward the Ph.D.
degree in electrical engineering at the School of
Electrical Engineering and Automation.
Her main research interests include multilevel
power conversion and motor control.
Yun Zhang was born in Jiangsu, China, in 1980.
He received the B.S. and M.S. degrees in electrical
engineering from the Harbin University of Science
and Technology, Harbin, China, in 2003 and 2006,
respectively, and the Ph.D. degree in electrical en-
gineering from the Harbin Institute of Technology,
Harbin, in 2010.
He is currently a Lecturer with the School of
Electrical Engineering and Automation, Tianjin Uni-
versity, Tianjin, China. His current research inter-
ests include multilevel power conversion and motor
control.
Xiangning He (M95SM96F10) received the
B.Sc. and M.Sc. degrees fromthe Nanjing University
of Aeronautics and Astronautics, Nanjing, China,
in 1982 and 1985, respectively, and the Ph.D. de-
gree from Zhejiang University, Hangzhou, China, in
1989.
From 1985 to 1986, he was an Assistant Engineer
with the 608 Institute of Aeronautical Industrial Gen-
eral Company, Zhuzhou, China. From 1989 to 1991,
he was a Lecturer with Zhejiang University. In 1991,
for two years, he was a Postdoctoral Research Fellow
with the Department of Computing and Electrical Engineering, HeriotWatt
University, Edinburgh, U.K., where he conducted research under a fellowship
from the Royal Society of U.K. In 1994, he joined Zhejiang University as
an Associate Professor, where he was the Director of the Power Electronics
Research Institute, was the Head of the Department of Applied Electronics, has
been a Full Professor with the College of Electrical Engineering since 1996,
and is currently the Vice Dean of the College of Electrical Engineering. His
research interests include power electronics and their industrial applications.