Sie sind auf Seite 1von 13

IEEE TRANSACTIONS ON POWER ELECTRONICS, VOL. 31, NO.

2, FEBRUARY 2016

915

Modeling and Control of a Multiport Power


Electronic Transformer (PET) for Electric
Traction Applications
Chunyang Gu, Student Member, IEEE, Zedong Zheng, Member, IEEE, Lie Xu, Member, IEEE,
Kui Wang, Member, IEEE, and Yongdong Li, Member, IEEE

AbstractThis paper proposes a multiport power electronic


transformer (PET) topology with multiwinding medium-frequency
transformer (MW-MFT) isolation along with the associated modeling analysis and control scheme. The power balance at different
ports can be controlled using the multiwinding transformers common flux linkage. The potential applications of the proposed multiport PET are high-power traction systems for locomotives and electric multiple units, marine propulsion, wind power generation, and
utility grid distribution applications. The complementary polygon
equivalent circuit modeling of an MW-MFT is presented. The current and power characteristics of the virtual circuit branches and
the multiports with general-phase-shift control are described. The
general current and power analysis for the multiple active bridge
(MAB) isolation units is investigated. Power decoupling methods,
including nonlinear solution for power balancing are proposed.
The zero-voltage-switching conditions for the MAB are discussed.
Control strategies including soft-switching-phase-shift control and
voltage balancing control based on the power decoupling calculations are described. Simulations and experiments are presented
to verify the performance of the proposed topology and control
algorithms.
Index TermsElectric traction, multiple active bridge (MAB),
multiwinding medium-frequency transformer (MW-MFT), power
decoupling, power electronic transformer (PET).

I. INTRODUCTION
INGLE-PHASE line frequency transformers with fourquadrant power converters are commonly used for traction applications on electric locomotives and electric multiple
units (EMUs) [1], [2] due to their simplicity, reliability, and low
price. However, they also have some obvious drawbacks including large size and weight because of low fundamental frequency
and second-order harmonics of the dc output voltage [3].
To increase the power density and improve the performance
of traction converter, as well as enhance the availability and flexibility of the electric traction system, medium-frequency transformer (MFT) isolated power converters can be used [4][6].

Manuscript received December 12, 2014; accepted March 14, 2015. Date
of publication March 24, 2015; date of current version September 29, 2015.
This work was supported by State Key Laboratory of Control and Simulation
of Power System and Generation Equipment (Tsinghua University, China).
This paper was partly presented at the IEEE Energy Conversion Congress and
Exposition, Pittsburgh, PA, USA, September 1418, 2014.
The authors are with the Department of Electrical Engineering, Tsinghua
University, Beijing 100084, China (e-mail: gucy10@mails.tsinghua.edu.cn;
zzd@tsinghua.edu.cn; xulie@tsinghua.edu.cn; wangkui@tsinghua.edu.cn;
liyd@tsinghua.edu.cn).
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/TPEL.2015.2416212

This type of power converter using an MFT for voltage matching and electrical isolation is also known as a power electronic
transformer (PET) or solid state transformer [6], [7]. Fig. 1(a)
shows the traditional traction converter configuration with line
frequency medium-voltage (MV) transformers. This configuration can be replaced by the PET-based configurations, as
shown in Fig. 1(b) and (c), with the increasing demand of lower
size/weight and better output voltage performance, especially
when the efficiency for the PET is increased with silicon carbide devices and nanocrystalline magnetic materials [8][15].
Traction-related industrial companies and research institutes are investigating PETs for traction use. ALSTOM developed their e-transformer in which a 1.5-MVA single-output
MFT was used for isolation [16]. In other early studies, cascaded single-phase matrix converters were used as the gridconnected converters [17], [18]. SIEMENS proposed a highvoltage direct ac/ac converter for line frequency to mediumfrequency conversion. A modular multilevel converter was also
used as the primary converter as proposed in [19]. ABB and
BOMBARDIER suggested configurations with primary-seriessecondary-parallel-connected dc-links as traction converters for
high-speed trains [20], [21], as shown in Fig. 1(b). This configuration is now considered as the most typical PET topology
for traction applications. This PET topology uses a cascaded Hbridge rectifier (CHBR) on the grid side and several dual active
bridge (DAB) circuits for isolation. The structure of secondary
side dc buses in parallel can drive a single load, however, it
is difficult to balance the dc voltages with two or more different loads [22][26]. Moreover, numbers of MV MFTs increase
the weight and volume because of duplicate insulation in each
transformer.
A few studies have considered a multiport PET with a multiwinding medium-frequency transformer (MW-MFT) [27][34].
The topology configuration, transformer modeling, voltage and
power balancing, as well as high efficiency control of switching
devices are key issues for multiport PETs. Some research works
for multiport PETs have been published on circuit topologies
[27] and dc-voltage balancing methods [34], [35] with singlephase shift (SPS) [31] control, but there has been little work on
multiwinding transformer modeling for PET, multiport power
analysis, and zero-voltage switching (ZVS) control.
The aim of this paper is to propose a multiport traction PET
topology to analyze the current and power of multiport isolation
units in detail and to put forward the ZVS control method and
voltage balancing strategy for the multiport PET.

0885-8993 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.

916

IEEE TRANSACTIONS ON POWER ELECTRONICS, VOL. 31, NO. 2, FEBRUARY 2016

Fig. 1. Transformer and converter configurations for electric trains. (a) Traditional configuration with a line frequency transformer. (b) Typical PET configuration.
(c) Multiport PET configuration.

II. MULTIPORT PET TOPOLOGY FOR ELECTRIC


TRACTION APPLICATIONS
The multiport PET topology for electric traction applications
proposed in this paper is shown in Fig. 1(c). The main idea of
this topology is to convert single-phase medium-voltage linefrequency ac voltage into several medium frequency voltages.
These medium frequency voltages are isolated using MW-MFT,
and rectified to produce several output dc links. The multiport
PET can be divided into three parts. The first part is CHBR.
The cell number of CHBR is determined by the ac grid voltage level and the rated voltage of power electronics devices.
The second part is MW-MFT isolation units (multiple active
bridge, MAB), in which a group of medium-frequency inverters,
medium-frequency rectifiers, and a multiwinding transformer
are involved. The number of secondary windings of multiwinding transformer is determined by the requirements of loads. The
third part is the load converters. The load-side converters can be
of different types, such as two-level inverters for motor drive,
inverters for the multiphase motor, and cascaded inverter for
higher voltage motors.
All windings of the transformer are linked by the same flux,
therefore, the load power can be balanced naturally by the energy delivery within different windings through the common
transformer flux. The major advantages of multiport PET with
MW-MFT are as follows:
1) The voltage and power balancing are realized by flux
linkage, so that the load of different windings of the transformer can be different from each other.
2) The power density of multiport PET is higher than
the typical PET shown in Fig. 1(b). For example, for
25-kV traction systems, the MFT is with medium frequency and MV insulation requirements (>35 kV), so
that multiwinding structure could decrease the size and
weight of insulation by sharing the oil tank of the transformer.
3) The efficiency of multiport PET is higher than the typical
PET configuration, for the magnetic current and loss is
smaller than that of the typical PET, which will be explained in the following section.
4) As shown in Fig. 2(a), for EMU train traction, a single
MW-MFT is needed. For higher power rating locomotives,

Fig. 2. Multiport PET topologies for electric traction. (a) Multiport PET topology for EMU trains. (b) Multiport PET topology for locomotives.

Fig. 3.

MW-MFT isolation units (MAB).

PET topology with several MW-MFTs could be used due


to the limitation of a single MV MFT capacity, as shown
in Fig. 2(b).
III. MODELING AND ANALYSIS OF AN MAB AS THE
ISOLATION UNITS
The multiport isolation units consist of several single-phase
full H-bridges and a medium-frequency MW-MFT, as shown in
Fig. 3. Similar to the definition of the DAB circuit, we could
call this multiport isolated dc/dc converter as an MAB circuit.
The ac voltage outputs of isolation units are represented by u1 ,
u2 , . . . , un , the ac currents are i1 , i2 , . . . , in , and the dc voltages
are Udc,1 , Udc,2 , . . . , Udc,n , respectively.
A. Equivalent Circuit of an MW-MFT
The circuit diagram [36], [37] of a multiwinding transformer
is shown in Fig. 4.
The turning ratio of the transformer is considered to be 1. The
effect of magnetic saturation is neglected. Suppose the winding

GU et al.: MODELING AND CONTROL OF A MULTIPORT POWER ELECTRONIC TRANSFORMER

Fig. 4.

917

Circuit diagram of a multiwinding transformer.

resistances are R1 , R2 , . . . , Rn . The self-leakage inductance of


windings are L1,1 , L2,2 , . . . , Ln ,n , respectively. The mutual
leakage inductance between windings are L1,2 , . . . , L1,n ;
L2,1 , . . . , L2,n ; Ln ,1 , . . . , Ln ,n 1 , respectively. The current of excitation circuit is im = i1 + i2 + . . . + in . The magnetizing inductance is Lm . The core loss resistance is Rm . The
electromotive force (EMF) caused by main flux is expressed as
e = Lm dim /dt Rm im .
The voltage equations of the multiwinding transformer can
be written as

Fig. 5. Complementary polygon equivalent circuit of a multiwinding transformer with a magnetizing branch.

By solving (2), we could get the current of the jth winding


n

1
1
d
ij =
(uj + e) +
(u ui )
dt
Lj,j,j
Lj,j,i j

(3)

i=1,i= j


e
u1 i1 R1
..

..
.

uj ij Rj + e

..
..

.
e
un in Rn

L1,1

.
.
.

= Lj,1

.
.
.

Ln ,1


where 1/Lj,j,i is the element of Row j, and Column i in the


inversion matrix of Lj .
Set the virtual branch current
u j u i
Lj,j,i

...

L1,j

...

L1,n

..

..
.

..

..
.

...

Lj,j

...

Lj,n

..

..
.

..

..
.

...

Ln ,j

...

Ln ,n

i1

.
.
d .

ij (1)
dt .
.

in


. . . L1,n
...
...
where LTran = . . .
is the inductance maLn ,1 . . . Ln ,n
trix of the multiwinding transformer, Lj,k = Lk ,j .
No. j equation in (1) is subtracted in the rest of the equations
on the left and right sides, respectively, then (2) shown at the
bottom of the page, where uj = uj ij Rj . Lj is the jsubtracted inductance matrix. No. j equation in (1) is maintained.

Lj,1 L1,1

i1
..

i2
Lj,1 Lj 1,1

d ...
=
Lj,1
Lj

dt
i
j

Lj,1 Lj +1,1
...

..

.
in

Lj,1 Ln ,1

u j +e
Lj,j,j

u j +e di j , i
L j , m , dt

, then

d
d
ij =
ij,m +
ij,i .
dt
dt

L1,1

u j u i
L j , i

di j , m
dt

(4)

i=1,i= j

For the first, second, . . . , nth windings, take the same analysis
as from (2) to (4). By inverting L1 , L2 , . . . , Ln , we can get the
equations as follows:

u1 uj

di
u + e
u1 un

1 = 1 
+ ... +
+
...
+

L1,m
L1,j
L1,n

dt

...

di
uj u1
uj + e
uj un
j
=
+ ... + 
+ ... +
(5)


dt
Lj,1
Lj,m
Lj,n

...

un uj
din
un u1
u + e

+ ... +
+ ... + n
dt = L

Ln ,j
Ln ,m
n ,1
where Lj,i = Li,j can be verified.
From (5), the complementary polygon equivalent circuit of
multiwinding transformer is shown in Fig. 5. The advantages

...

Lj,j L1,j

...

..

..
.

..

...

Lj,j Lj 1,j

...

...

Lj,j

...

...

Lj,j Lj +1,j

...

..

..
.

..

...

Lj,j Ln ,j

...

uj u1
i
..

...
.
i

2 

Lj,n Lj 1,n uj uj 1


d ...

Lj,n
= uj + e
dt ij




Lj,n Lj +1,n
uj uj +1

...

..
..,

.
in



uj un
Lj,n Ln ,n
Lj,n L1,n

(2)

918

IEEE TRANSACTIONS ON POWER ELECTRONICS, VOL. 31, NO. 2, FEBRUARY 2016

Fig. 6. Polygon equivalent circuit of a multiwinding transformer without a


magnetizing branch.

Fig. 7.

AC voltages of one virtual branch.

of the modeling with the complementary polygon equivalent


circuit are: the physical meaning is clear, the back EMF information could be obtained intuitively, and the iron loss could be
estimated more accurately.
As shown in Fig. 5, the magnetic current is almost the same
as a two-winding transformer with the same Lm and Rm , so that
the total magnetic loss of the PET can be the same of a twowinding transformer. The virtual currents ij,i and ij,m are the
practical branch currents in Fig. 5. From (1) to (5), we could get
the parameters of the complementary polygon equivalent circuit
by self-inductance and mutual-inductance matrix (calculated by
multiwinding voltages and currents) with the magnetizing inductance (calculated by open-circuit test and short-circuit test).
The disadvantage of the complementary polygon equivalent circuit is that the equivalent circuit is complicated.
By further ignoring the excitation current and copper loss, we
could get the polygon equivalent circuit in Fig. 6

d
1
1

i1 =  (u1 u2 ) + ... +  (u1 un )

dt
L
L

1,2
1,n

...
.

d
1
1

(un un 1 )
in =  (un u1 ) + ... + 
dt
Ln ,1
Ln ,n 1
(6)
Equation (6) shows that the derivative of current ij,i of a
virtual circuit branch (made up of two voltage sources uj , ui
and an inductor Lj,i ) is only influenced by the voltage difference
between uj and ui . By using the polygon equivalent circuit, we
could calculate the current and power of each virtual branch, and
then calculate the current and power of each port by summing
up these virtual branches. It is better to use a polygon equivalent
circuit instead of a simplified equivalent circuit [34] because the
mutual leakage inductance of a high-power MW-MFT cannot
be ignored.
B. Current and Power of Each Virtual Branch
Fig. 7 shows two ac voltages for one virtual branch, which is
actually a virtual DAB circuit. The frequency of voltage sources

uj and ui is f . The magnitudes of uj and ui are Udc,j and Udc,i ,


respectively. To realize the general-phase-shift (GPS) control of
the isolation units, variable duty cycle modulation is added by
the phase shift of two arms inside an H-bridge. Suppose the duty
cycle of uj is dj , and duty cycle of ui is di , where dj [0, 1],
di [0, 1]. Set the phase shift between uj and ui to be j,i ,
where j,i [, ). To facilitate analysis, set the inner phaseshift ratio DIN,j = 1 dj [0, 1] and DIN,i = 1 di [0, 1].
Set the outer phase-shift ratio Dj,i = j,i / [1, 1). In Fig. 7,
assuming that the switching period T = 1/f , when a switching
period starts at t = 0, then t1 = DIN,j T /4, t2 = DIN,i T /4 +
1
2 Dj,i T /2, t3 = T /4 + Dj,i T /2, t4 = T /2 DIN,j T /4, and
t5 = T /2 DIN,i T /2 + Dj,i T /2.
Assume dc voltages Udc,j Udc,i . The proportion of voltages is aj,i = Udc,i /Udc,j < 1. The superposition theorem is
used to simplify the analysis process. The subcircuit current ij,i can be considered as the superposition of the current isub1 j,i and isub2 j,i generated by uj and ui , respecdu
i
tively, where ij,i = isub1 j,i + isub2 j,i = Lj,i ( dtj + du
dt ). The
magnitudes of isub1 j,i and isub2 j,i are Im 1 j,i and Im 2 j,i ,
U
(1D
)
U
(1D
)
where Im 1 j,i = d c , j4f L  I N , j and Im 2 j,i = d c , i4f L  I N , i .
j,i

j,i

With DIN,j [0, 1], DIN,i [0, 1], and Dj,i [1, 1), 14 different positions of uj relative to ui are shown in Table I and
Fig. 8 as in conditions of (a) to (n). For example, (g) is in
condition of DIN,j > DIN,i and |Dj,i | < (DIN,j DIN,i )/2.
Descriptions of instantaneous currents at the time of t1 , t2 , t4 ,
and t5 in Fig. 7 are shown in Table I.
The average power flow Pj,i between uj and ui is
considered as


Pj,i =

uj (isub1

j,i

+ isub2

j,i )d(2f t)


=

uj isub2

j,i d(2f t).

(7)

Traversing the 14 possible situations, a piecewise analytic


function of average power Pj,i is summarized in Table II, where

= 0.5 ||Di,j | 0.5| [0, 0.5], DIN,j


= DIN,j /2,
ratios Di,j

and DIN,i = DIN,i /2. The interval division schematics of average power calculation are shown in Fig. 9.
Fig. 10(a) shows a series of 3-D surface of power,
where DIN,j [0, 1], Di,j [1, 1), and Pbase = Udc,j
Udc,i /(8f Lj,i ). It is apparent that when Di,j [0.5, 0.5], the
power function is monotonic related to Di,j , which is the value
of the outer phase shift ratio that we could use to control the
system, as shown in Fig. 10(b). Fig. 10(c), as an example, is the
average power surface where DIN,j [0, 1], Di,j [0.5, 0.5],
and DIN,i = 0.8, which is a piecewise function apparently different from the average power curve with SPS control.
C. Current and Power of Each Port
The MAB circuit, as shown in Fig. 3, is the combination of
virtual DAB circuits. The current of each winding of MW-MFT
ij is a function of inner phase-shift ratios and outer phase-shift
ratios.

GU et al.: MODELING AND CONTROL OF A MULTIPORT POWER ELECTRONIC TRANSFORMER

919

TABLE I
INSTANTANEOUS VALUE OF THE DAB CURRENT WITH GPS CONTROL IN FIG. 8 (DIVIDED BY U d c , j /4f L j, i )
i j , i (t 1 )

i j , i (t 2 )

i j , i (t 4 )

i j , i (t 5 )

a)

(1 D I N , j ) a j , i (1 D I N , i )

(1 D I N , j ) + a j , i (1 D I N , i )

b)

(1 D I N , j ) + a j , i (2D j , i + 1 +
DIN , j )
(1 D I N , j ) + a j , i (2D j , i + 1 +
DIN , j )
(1 D I N , j ) + a j , i (2D j , i + 1 +
DIN , j )
(1 D I N , j ) + a j , i (1 D I N , i )
(1 D I N , j ) + a j , i (1 D I N , i )

(2D j , i 3 + D 2 ) + a j , i (1
DIN , i )
(1 D I N , j ) + a j , i (1 D I N , i )

(2D j , i 1 D I N , i ) a j , i (1
DIN , i )
(1 D I N , j ) a j , i (1 D I N , i )

c)
d)
e)
f)
g)
h)
i)
j)
k)
l)
m)
n)

(1 D I N , j ) + a j , i (2D j , i + 1
DIN , j )
(1 D I N , j ) + a j , i (1 D I N , i )
(1 D I N , j ) + a j , i (2D j , i + 1
DIN , j )
(1 D I N , j ) a j , i (1 D I N , i )
(1 D I N , j ) + a j , i (2D j , i + 1
DIN , j )
(1 D I N , j ) a j , i (1 D I N , i )
(1 D I N , j ) + a j , i (2D j , i 3 +
DIN , j )
(1 D I N , j ) a j , i (1 D I N , i )

(1 D I N , j ) + a j , i (2D j , i +
3 DIN , j )
(1 D I N , j ) + a j , i (1 D I N , i )

(1 D I N , j ) + a j , i (1 D I N , i )
(2D j , i 1 + D 2 ) + a j , i (1
DIN , i )
(1 D I N , j ) + a j , i (1 D I N , i )
(2D j , i 1 + D 2 ) + a j , i (1
DIN , i )
(1 D I N , j ) + a j , i (1 D I N , i )
(2D j , i 1 + D 2 ) + a j , i (1
DIN , i )
(1 D I N , j ) + a j , i (1 D I N , i )
(1 D I N , j ) + a j , i (1 D I N , i )
(2D j , i + 1 + D 2 ) + a j , i (1
DIN , i )
(2D j , i + 1 + D 2 ) + a j , i (1
DIN , i )
(1 D I N , j ) + a j , i (1 D I N , i )
(2D j , i + 1 + D 2 ) + a j , i (1
DIN , i )

(1 D I N , j ) + a j , i (2D j , i 1 +
DIN , j )
(1 D I N , j ) + a j , i (1 D I N , i )
(1 D I N , j ) + a j , i (2D j , i 1 +
DIN , j )
(1 D I N , j ) + a j , i (2D j , i 1 +
DIN , j )
(1 D I N , j ) a j , i (1 D I N , i )
(1 D I N , j ) a j , i (1 D I N , i )
(1 D I N , j ) a j , i (1 D I N , i )
(1 D I N , j ) + a j , i (2D j , i 1
DIN , j )
(1 D I N , j ) + a j , i (2D j , i 1
DIN , j )
(1 D I N , j ) + a j , i (2D j , i 1
DIN , j )
(1 D I N , j ) + a j , i (1 D I N , i )

(2D j , i 1 D I N , i ) a j , i (1
DIN , i )
(2D j , i 1 D 2 ) a j , i (1
DIN , i )
(1 D I N , j ) a j , i (1 D I N , i )
(1 D I N , j ) a j , i (1 D I N , i )
(1 D I N , j ) a j , i (1 D I N , i )
(2D j , i + 1 D 2 ) a j , i (1
DIN , i )
(2D j , i + 1 D 2 ) a j , i (1
DIN , i )
(1 D I N , j ) a j , i (1 D I N , i )
(2D j , i + 1 D 2 ) a j , i (1
DIN , i )
(1 D I N , j ) a j , i (1 D I N , i )
(1 D I N , j ) a j , i (1 D I N , i )
(2D j , i + 3 D 2 ) a j , i (1
DIN , i )

D. Decoupling of Multiport Powers of the PET


For general control of the MAB, the inner phase-shift ratio is set to be a constant value or zero, then the phaseshift ratio between fundamental component of uj and ui is
Dj,i = DOUT,j DOUT,i , where DOUT,j is the phase shift
between uj and u1 (DOUT,1 = 0), then
Pj = Pj (Dj,1 , Dj,2 , ..., Dj,N )
= Pj (DOUT,1 , DOUT,2 , ..., DOUT,N ).

Fig. 8.

Different related positions of u j and u i in conditions of (a) to (n).

When setting ij = ij (Dj,1 , . . . , Dj,n ; DIN,1 , . . . , DIN,n ),


then
n

ij,i .
(8)
ij =

For further control of the MAB, the decoupling of port power


Pj is needed. Two kinds of power decoupling methods are proposed in this paper: nonlinear solution method and simplified
estimation method.
1) Nonlinear Solution
 of Multiport Power Equations: For
the jth winding, Pj = ni=1,i= j Pj,i , where Pj,i is definite in
Table II. The inner phase-shift ratios DIN,1 , . . . , DIN,n are set
to be constant. Udc,1 , . . . , Ddc,n , Lj,i , and f are constant values
as well. As Pj,i is a function of Dj,i = DOUT,j DOUT,i , then
Pj = Pj (DOUT,1 , ...DOUT,n )

i=1,i= j

The average power of Port No. j is the sum of n 1 subcircuits, which is relatively complicated while using GPS control. When setting Pj = Pj (Dj,1 , . . . , Dj,n ; DIN,1 , . . . , DIN,n ),
then
n

Pj,i .
(9)
Pj =
i=1,i= j

(10)

Pj,i (DOUT,j DOUT,i ).

(11)

i=1,i= j

For the nonlinear equations with n variables Pj =


Pj (DOUT,1 , . . . DOUT,n ), the number of linearly independent
input variables is n 1 (where DOUT,1 =
0). The number of
output variables is n 1 (where P1 = ni=2 Pi ). Then, the

920

IEEE TRANSACTIONS ON POWER ELECTRONICS, VOL. 31, NO. 2, FEBRUARY 2016

TABLE II
AVERAGE POWER OF THE DAB WITH GPS CONTROL
Zone

Situations of Phase shifting

Average power

D IN , j > D IN , i and 0 D i, j < D IN , j D IN , i

sgn(D i , j )

Ud c, j Ud c, i
(1 2D IN , j )D i, j
2f L j , i

D IN , j D IN , i and 0 D i, j < D IN , i D IN , j

sgn(D i , j )







D I N , j D IN , i  D i, j < m in (D IN , j + D IN , i ), 1 (D IN , j + D IN , i )

Ud c, j Ud c, i
(1 2D IN , i )D i, j
2f L j , i

D
E

D IN , j + D IN , i
D IN , j + D IN , i >

1
1
and (D IN , j + D IN , i ) D i, j
2
2

sgn(D i , j )


Ud c, j Ud c, i 
2D i, j (1 2D IN , j ) (D i, j D IN , j + D I2N , i )
4f L j , i

sgn(D i , j )

1
1
and1 (D IN , j + D IN , i ) D i, j
2
2


Ud c, j Ud c, i 
2D i, j (1 D i, j ) 2D I2N , j 2D I2N , i
4f L j , i

sgn(D i , j )

Ud c, j Ud c, i
(1 2D IN , j )(1 2D IN , i )
4f L j , i

iteration method
k +1
k
0
k
= DOUT
[F  (DOUT
)]1 F (DOUT
)
DOUT

0
)=
F  (DOUT

Fig. 9. Five value ranges of inner phaseshift ratios and outer phaseshift ratio
in Table II. (a) Zone A. (b) Zone B. (c) Zone C. (d) Zone D. (e) Zone E.

n 1 dimensional nonlinear equations are

P2 = f2 (DOUT,2 , ...DOUT,n )

n


P2,i (DOUT,2 DOUT,i )


=

i=1,i= 2
...

Pn = fn (DOUT,2 , ...DOUT,n )

n


Pn ,i (DOUT,n DOUT,i )
=

(12)

i=1,i= n

where F (DOUT ) = [f2 (DOUT ), fn (DOUT )]T . The Jacobi


 D OfU2T , 2 . . . DOfU2T , n

... ...
...
matrix of (13) is F (DOUT ) =
,
fn
fn
.
.
.
DO U T ,2
DO U T ,n
which is a real symmetric matrix. F  (DOUT ) is also a
strictly
diagonally dominant matrix where |fj /DOUT,j | >
n
i=1,i= j |fj /DOUT,i | and fj /DOUT,j > 0. Therefore,
the use of the Newton iteration method and other methods like
the steepest descent method is convergence [38], [39].
For the Newton iteration method
k +1
k
k
k
= DOUT
[F  (DOUT
)]1 F (DOUT
).
DOUT

(13)

The Newton iteration method needs the calculation of partial


derivative for every step which is complicated to be realized in
DSP controllers. In this paper, we used the simplified Newton

n


(14)

1
2f

Udc,2 Udc,i
L2,i

i=1,i= 2

..

Udc,n Udc,2

Ln ,2

...
..

Udc,2 Udc,n
L2,n

...

n

i=1,i= n

..

. (15)
.

Udc,n Udc,i
Ln ,i

Only one inverse matrix needs to be solved offline, and


the appropriate look-up table needs to be established in the
DSP controller whose complexity is available for practical
application.
Table III shows the simulation results of the simplified Newton iteration method solution of multiport powers using one time
of linear solving and five times of iteration correction. The simulation results are shown in Fig. 11. When the phase shifting is
less than /6, a 2% accuracy solution is reached with one time
of linear solving and one time of iteration correction which is
applicative with DSP controllers.
2)Simplified Estimation of Multiport Power for SPS Control:
SPS control is the most popular control strategy in DAB as well
as in MAB circuits. For SPS control, DIN,1 = . . . = DIN,n = 0,
then from Table II, we can get the branch power
Pj,i =

sgn(Di,j )Udc,j Udc,i


4f Lj,i



2
2
2Di,j
(1 Di,j
) 2DIN,j
2DIN,i
Udc,j Udc,i
Di,j (1 |Di,j |).
2f Lj,i

(16)

Ignoring the mutual leakage inductance, excitation current,


and the winding
resistance, then Lj,i = Lj,j Li,i /LP , where
n
LP = 1/ i=1 (1/Li,i ). The total power of the source voltage

GU et al.: MODELING AND CONTROL OF A MULTIPORT POWER ELECTRONIC TRANSFORMER

921

Fig. 10. Three-dimensional surfaces of power of each virtual branch. (a) When D IN , j [0, 1],D i , j [1, 1). (b) When D IN , j [0, 1],D i , j [0.5, 0.5].
(c) When D IN , j [0, 1), D i , j [0.5, 0.5], and D IN , i = 0.8.

TABLE III
EXAMPLE OF THE NONLINEAR EQUATION SOLUTION OF POWERS

Set DOUT,1 = 0, then

Winding #1 Winding #2 Winding #3 Winding #4 Winding #5 Winding #6


Pre f
DO U T
E r rP 1
E r rP 2
E r rP 3
E r rP 4
E r rP 5
E r rP 6

-0.15
0
0.0537
0.0062
0.0008
0.0001
0.0000
0.0000

0.25
0.0718
0.0728
0.0114
0.0019
0.0003
0.0001
0.0000

0.05
0.0359
0.1056
0.0200
0.0038
0.0007
0.0001
0.0000

0.15
0.0537
0.0731
0.0118
0.0021
0.0004
0.0001
0.0000

0.1
0.0448
0.0806
0.0137
0.0025
0.0004
0.0001
0.0000

0.4
0.0471
0.0861
0.0152
0.0028
0.0005
0.0001
0.0000

DOUT,2

DOUT,3

...

DOUT,n

Udc,j Udc,i
(DOUT,j DOUT,i )
Lj,j Li,i

DOUT,j

i=1,i= j

[1 |DOUT,j DOUT,i |] = Pj .

(17)

Set Udc,1 = . . . = Udc,n = Udc and L1,1 = . . . = Ln ,n =


L , then LP = L /n. When DOUT,j is very small
2
Udc
2nf L

(DOUT,j DOUT,i ).

(18)

i=1,i= j

Then

2
Udc
2nf L

n1

...

DOUT,1

D
OUT,2

...

DOUT,n

2
n

2nf L n

=
2
...

Udc

1
n
2
n

...
...

...

...

1
n

...

n
P2

P
n
3

...
...

2
Pn
n

2nf L 2
Pj
=
Pi
2
Udc
n
=

Pj

(20)
The phase-shift ratio is solved using the subtraction of port
powers

uj meets the condition that


LP
2f

...

n1

...

...
1


2
1
2f L
Pj (P1 + Pj ) = 2 (Pj P1 ).
n
n
Udc
(21)

By using this simplified estimation method, the outer phaseshift ratio DOUT,j could be decided briefly. This method is
simple and accurate enough for practical realization with PIbased dc-voltage loop when the phaseshift angle is small.

IV. CONTROL OF AN MAB AS THE ISOLATION UNITS


P
2
= .

...

Pn

i=2,i= j

...
...

... n 1
P1

2nf L
2
Udc

(19)

In this section, two necessary control strategies of the MAB


are proposed. Soft-switching conditions of the DAB circuit as
well as the MAB circuit are discussed. Then, a soft-switchingphase-shift (SSPS) control soft switching control method is put
forward in which the inner phase shift ratios are set to be constant
values depending on dc voltages. The use of the SSPS method
will realize ZVS soft switching and reduce the peak current
of the MAB circuit, which could increase the efficiency of the
system. The other control strategy is the dc voltage balancing
method based on power decoupling calculation which could
increase the response speed of the voltage controller.

922

IEEE TRANSACTIONS ON POWER ELECTRONICS, VOL. 31, NO. 2, FEBRUARY 2016

Fig. 11. Iterative of solution of powers in Table III. (a) Error percentages of power. (b) Errors of power. (c) Calculation values of outer phaseshift ratio.
(d) Calculation values of power.

A. SSPS Control
ZVS soft switching is achieved in a condition when the current
flow upon the device is negative at the time the device is turned
ON. For the analysis of soft-switching conditions of each virtual
circuit branch (DAB), the instantaneous value of the current at
the switching moment is needed. The universal soft-switching
conditions of DAB are:
1) left-hand side arm of winding No. j: ij,i (1 DIN,j /2)
0;
2) right-hand side arm of winding No. j: ij,i (DIN,j /2) 0;
3) left-hand side arm of winding No. i: ij,i (Dj,i + 1
DIN,i /2) 0;
4) right-hand side arm of winding No. i: ij,i (Dj,i +
DIN,i /2) 0.
By analyzing the instantaneous current of the device (shown
in Table I), a sufficient but not necessary ZVS switching condition is noticed
Udc,i (1 DIN,i ) = Udc,j (1 DIN,j ).

(22)

For the MAB circuit, the ZVS condition of winding No. j is:
1) 
left-hand side arm of winding No. j: ij (1 DIN,j /2) =
n
i=1,i= j ij,i (1 DIN,j /2) 0;
2) 
right-hand side arm of winding No. j: ij (DIN,j /2) =
n
i=1,i= j ij,i (DIN,j /2) 0.
If every virtual branch circuit fits their own ZVS conditions,
the MAB circuit is in ZVS condition inevitably. We can get the
conclusions that:
1) When the voltages Udc,1 , Udc,2 , . . . , Udc,n are all the
same, a full range of ZVS soft-switching condition is
realized.
2) When the dc voltages are not the same, ZVS softswitching may not be realized under light load conditions.
3) When the dc voltages of the primary side are the same, and
the secondary side dc voltages are the same, respectively,
the extended-phase-shift control could be used for ZVS
soft-switching [40], [41].
4) Furthermore, when the dc voltages of each winding are not
the same, in the condition that the duty of the H-bridges
are set to be dj = Udc,m in /Udc,j , ZVS soft-switching is
realized.

Fig. 12. DC-voltage balancing control diagram of an MAB. (a) Power decoupling control. (b) Simplified PI-based control.

The inner phase shifting between two half-bridge squarewave gate signals of an H-bridge is independent of the outer
phaseshift ratio, which is proposed as the SSPS method of the
MAB. When Udc,1 Udc,2 . . . Udc,n

1 DIN,1 = 1

Udc,1
Udc,1

=
1 DIN,2 = (1 DIN,1 )

U
U
dc,2
dc,2

Udc,1
Udc,1 .
(23)
1 DIN,3 = (1 DIN,1 )
=

Udc,3
Udc,3

...

Udc,1
Udc,1

1 DIN,n = (1 DIN,1 )
=
Udc,n
Udc,n
B. DC Voltage Balancing Control
Based on the decoupling method of multiwinding powers,
a power balancing control method of the MAB is adopted, as
shown in Fig. 12(a). The ac powers of the MW-MFT isolation
units are controlled by the relative voltage phases. In order to
balance the dc voltages of each cell, the power can be controlled
by the feedback of the dc voltage. PI controllers are used to control dc voltages. The outputs of PI controllers are the reference

GU et al.: MODELING AND CONTROL OF A MULTIPORT POWER ELECTRONIC TRANSFORMER

923

TABLE IV
PARAMETERS OF SIMULATION
Parameter
Rated grid voltage
Grid frequency
Grid filter inductance
Switching frequency of CHBR
Number of CHBR modules

Value

Parameter

Value

25 kV
50 Hz
50 mH
150 Hz
26

Rated voltage of MW-MFT


Primary DC voltage
Secondary DC voltages
Frequency of MW-MFT
Simulation step

1800 V
1800 V/1671 V
1800 V/900 V
1 kHz
1 s

currents of dc capacitors IC,j


. Reference DC currents Idc,j
are

the sum of capacitor currents IC,j and load currents ILoad,j . The
reference powers Pj are the products of reference dc voltages

and reference dc currents Idc,j


.
Udc,j
Fig. 12(b) shows the simplified PI-based control method without power decoupling calculation, which is simple and easy to
be realized in the DSP controller. But its dynamic response
performance is worse than the voltage balancing control power
decoupling calculation.

V. EXPERIMENTAL VERIFICATION
A. Simulation Results
A full-scale simulation of the multiport PET system for
a 1.8-MVA EMU train traction system was done using the
MATLAB/Simulink software. The simulation parameters are
shown in Table IV.
1) SPS and SSPS Modulation: Fig. 13 shows the simulation
results of SPS and SSPS control with different dc voltages and
loads. For SPS control with the same dc voltage, the ac current
of MW-MFT was with a maximum value of about 70 A, as
shown in Fig. 13(a). Different H-bridges were operated with
different outer phaseshift ratio with different power. With no
load condition of Fig. 13(b), the ac current was very small.
But for SPS with different dc voltage (1671 V for the primary
side with 26 + 2 modules working for 46.8-kV dc total voltage,
1800 and 900V for the secondary side traction and auxiliary
power), the peak current increased to about 140 A, even in no
load situation. This phenomenon could be explained by using
Table I, when the outer phase shift ratio Dj,i = 0, the inner
phase shift ratio DIN,j = 0, DIN,i = 0, and the peak current is
still very large: IM = (Udc,j Udc,i )/4f Lj,i , because of the
difference of dc voltages.
For SSPS control of Fig. 13(e) and (f), the peak current was
reduced to 100 A with full load. The peak current with no load
condition reduced to half (about 70 A).
Every switching device works under the ZVS condition in
Fig. 13(a) and (b). But for Fig. 13(c) and (d), the IGBTs of
the second secondary H-bridge are not in ZVS condition. The
switching might be much higher than in Fig. 13(a) and (b). By
using SSPS control in Fig. 13(e) and (f), the IGBTs of the second
secondary H-bridge are in ZVS condition, even with no load,
which could decrease the switching loss of the system.
2) Voltage Balancing Process: Simulation results of the
MAB voltage balancing process is shown in Fig. 14. There are
second-order harmonics in the primary-side dc voltage waveforms because of the single-phase input power for electric

Fig. 13. Simulation results of MAB voltages and currents with SPS or SSPS
control. (a) SPS with same dc voltage and different load. (b) SPS with same
dc voltage and no load. (c) SPS with different dc voltage and different load.
(d) SPS with different dc voltage and no load. (e) SSPS with different dc voltage
and different load. (f) SSPS with different dc voltage and no load.

traction. But with an appropriate voltage balancing method,


there is no second-order harmonics in the secondary-side dc
voltage waveforms, which will improve the output performance
compared to the traditional traction converters.
At the time of 0.4 s, the voltage balancing strategy is enabled.
The dynamic response of the voltage balancing control method
with power decoupling is better than the PI-based control.
3) System Operation: Simulation results of CHBR and
precharging operation are shown in Fig. 15(a). For time range

924

Fig. 14. Simulation results of dc-voltage balancing process. (a) PI-based control with different load and same voltage. (b) PI-based control with different
load and different voltages. (c) Power decoupling control with different load
and same voltage. (d) Power decoupling control with different load and different voltages.

0 to 0.12 s, a precharge resistor is put in series. At time 0.12 s,


the resistor is cut off. CHBR starts working. For time range 0.13
to 0.15s, the multiport isolation units on the primary side are
working, with a ramp duty ratio from 0 to 1. At time 0.15 s, the
precharging procedure is completed. Then, the system works
regularly with power balancing control. Fig. 15(b) shows the
primary and secondary dc voltages with full load and regeneration modes. At the time of 0.25 s, the load changed from rated
load to regeneration in the rated power. The dc voltages controls
well and in good voltage balancing conditions. Fig. 15(c) and
(d) shows the detailed waveforms of the grid voltage, CHBR
voltage, and grid current with rated load under regeneration
condition. The output voltage of CHBR is a 53-level waveform.
Unity power factor is achieved.
B. Experimental Verification
To verify the topology and control strategies, a 6-kVA experimental prototype was built up, as shown in Fig. 16. An MWMFT with six primary and six secondary windings is adopted.
TI DSP TMS320F28335 and Altera FPGA EP2C20Q240C8 are
used as central controllers. Altera CPLDs EPM1270T144C5 is
used as distributed controllers for modulation.

IEEE TRANSACTIONS ON POWER ELECTRONICS, VOL. 31, NO. 2, FEBRUARY 2016

Fig. 15. Simulation results of the whole system operation. (a) Grid voltage,
PWM voltage, grid current. (b) Primary-side dc voltages and secondary-side
dc voltages. (c) Detailed waveforms of U g and Ig of powering (full load).
(d) Detailed waveforms of U g and Ig of regeneration.

Fig. 16.

Photograph of the experimental prototype.

1) SPS and SSPS Modulation: Fig. 17 shows the experimental results of SPS and SSPS control. For SPS control with the
same DC voltage, different H bridges operated with different
outer phase-shift ratio with different output power. With no load
condition, the ac current is very small. But for SPS with different dc voltage (primary-side dc voltage 20 V, first secondary
dc voltage 25 V and second secondary DC voltage 16 V), the
peak currents increases to about 12 A. For SSPS control of

GU et al.: MODELING AND CONTROL OF A MULTIPORT POWER ELECTRONIC TRANSFORMER

925

Fig. 18. Experimental results of the voltage-balancing process. (a) PI control


with different load and same voltage. (b) PI control with different load and
different voltage. (c) Power decoupling control with different load and same
voltage. (d) Power decoupling control with different load and different voltage.

Fig. 17. Experimental results of MAB voltages and currents with SPS or
SSPS control. (a) SPS with same dc voltage and full load. (b) SPS with same
dc voltage and no load. (c) SPS with different dc voltage and full load. (d) SPS
with different dc voltage and no load. (e) SSPS with different dc voltage and
full load. (f) SSPS with different dc voltage and no load.

Fig. 17 (e) and (f), the peak current reduced to 2 A, compared


to that of SPS control, with full load. The peak current with no
load condition reduced to half (about 1.8 A).
Every IGBT works under ZVS condition in Fig. 17(a) and (b).
But for Fig. 17(c) and (d), the IGBTs of the second secondary
H-bridge are not in ZVS condition. By using SSPS control in
Fig. 17(e) and (f), the IGBTs of the second secondary H-bridge
are in ZVS condition, even with no load, which could decrease
the switching loss of the system.

Fig. 19. Experimental results of the whole system operation. (a) Waveforms of
U g and Ig of powering (full load). (b) Waveforms of U g and Ig of regeneration.
(c) Primary-side dc voltages and secondary-side dc voltages.

2) Voltage Balancing Process: Experimental results of the


MAB voltage balancing process is shown in Fig. 18. At the
time of 0 s, the voltage balancing strategy starts working. The
dynamic response with power decoupling control is better than
the PI-based control.

926

IEEE TRANSACTIONS ON POWER ELECTRONICS, VOL. 31, NO. 2, FEBRUARY 2016

3) System Operation: Experimental results of CHBR voltage


and grid current with full load and in regeneration condition are
shown in Fig. 19(a) and (b). The output voltage of CHBR is a
13-level waveform. Unity power factor is achieved. Fig. 19(c)
shows the primary and secondary dc voltages with full load.
The dc voltages controls well and under good voltage balancing
conditions. There is second-order harmonics in primary-side dc
voltage waveforms. But there is no second-order harmonics in
the secondary-side dc voltage waveforms, which corresponds to
the simulation results.
Experiment results show that the control strategies have a
good performance. The faithfulness of the multiport topology,
SSPS control strategy, and dc-voltage balancing method are
verified.

[10]

[11]
[12]
[13]

[14]
[15]

VI. CONCLUSION
A multiport PET with MW-MFT isolation was analyzed in
this paper. The original contribution of above works included
modeling of MAB circuits as isolation units, complementary
polygon equivalent circuit of multiwinding transformer with
magnetizing branch, GPS control of current and power of each
virtual branch, as well as current and power of each port. Furthermore, the multiport power decoupling methods including
nonlinear solution of power equations and simplified estimation
method were devised. The following control strategies of the
proposed topology were adopted: the SSPS control method was
created to decrease the peak current and realize ZVS condition;
power-decoupling-based and PI-based strategy was proposed to
achieve voltage balancing. Simulations of multiport PET with
26-cascaded modules on the primary side were given. Experiments with a 6-kVA prototype were presented. The feasibility of
the proposed topology and voltage balancing control strategies
as well as ZVS soft-switching method have been verified by
both simulation and experimental results.

[16]
[17]
[18]

[19]
[20]

[21]
[22]

[23]
[24]

REFERENCES
[1] A. Steimel, Electric railway traction in Europe, IEEE Ind. Appl. Mag.,
vol. 2, no. 6, pp. 617, Nov. 1996.
[2] Z. Yang, X. Huang, S. Wu, and H. Peng, Traction technology for Chinese
railways, in Proc. Int. Power Energy Conf., 2010, pp. 28422848.
[3] H. Ouyang, K. Zhang, P. Zhang, Y. Kang, and J. Xiong, Repetitive
compensation of fluctuating DC link voltage for railway traction drives,
IEEE Trans. Power Electron., vol. 26, no. 8, pp. 21602171, Aug. 2011.
[4] S. Inoue and H. Akagi, A bidirectional isolated DC-DC converter as
a core circuit of the next-generation medium-voltage power conversion
system, IEEE Trans. Power Electron., vol. 22, no. 2, pp. 535542, Feb.
2007.
[5] X. Liu, H. Li, and Z. Wang, A start-up scheme for a three-stage solid state
transformer with minimized transformer current response, IEEE Trans.
Power Electron., vol. 27, no. 12, pp. 48324836, Dec. 2012.
[6] M. Sabahi, A. Y. Goharrizi, S. H. Hosseini, M. B. B. Sharifian, and
G. B. Gharehpetian, Flexible power electronic transformer, IEEE Trans.
Power Electron., vol. 25, no. 8, pp. 21592169, Aug. 2010.
[7] H. Fan and H. Li, High-frequency transformer isolated bidirectional DCDC converter modules with high efficiency over wide load range for
20 kVA solid-state transformer, IEEE Trans. Power Electron., vol. 26,
no. 12, pp. 35993608, Dec. 2011.
[8] P. Ladoux, M. Mermet, J. Casarin, and J. Fabre, Outlook for SiC devices
in traction converters, in Proc. Elect. Syst. Aircraft, Railway Ship Propul.,
2012, pp. 16.
[9] J. Casarin, P. Ladoux, B. Chauchat, D. Dedecius, and E. Laugt, Evaluation
of high voltage SiC diodes in a medium frequency AC/DC converter

[25]
[26]

[27]
[28]

[29]

[30]

[31]

for railway traction, in Proc. Int. Symp. Power Electron., Elect. Drives,
Autom. Motion, pp. 11821186.
A. Merkert, T. Krone, and A. Mertens, Characterization and scalable
modeling of power semiconductors for optimized design of traction inverters with si- and SiC-Devices, IEEE Trans. Power Electron., vol. 29,
no. 5, pp. 22382245, May 2014.
J. Fabre, P. Ladoux, and M. Piton, Characterization and implementation
of dual-SiC MOSFET modules for future use in traction converters, IEEE
Trans. Power Electron., vol. 30, no. 8, pp. 40794090, Aug. 2014.
H. Hoffmann and B. Piepenbreier, Medium frequency transformer for
rail application using new materials, in Proc. Elect. Drives Prod. Conf.,
2011, pp. 192197.
G. Ortiz, M. Leibl, J. W. Kolar, and O. Apeldoorn, Medium frequency
transformers for solid-state-transformer applications: Design and experimental verification, in Proc. IEEE Int. Conf. Power Electron. Drive Syst.,
2013, pp. 12851290.
G. T. Nikolov and V. C. Valchev, Nanocrystalline magnetic materials
versus ferrites in power electronics, Procedia Earth Plan. Sci., vol. 1,
no. 1, pp. 13571361, Jan. 2009.
J. Fuzer, P. Kollar, J. Fuzerova, and S. Roth, Soft magnetic properties
of nanostructured vitroperm alloy powder cores, IEEE Trans. Magn.,
vol. 46, no. 2, pp. 471474, Feb. 2010.
J. Taufiq, Power electronics technologies for railway vehicles, in Proc.
Power Convers. Conf., 2007, pp. 13881393.
M. Carpita, M. Marchesoni, M. Pellerin, and D. Moser, Multilevel converter for traction applications: small-scale prototype tests results, IEEE
Trans. Ind. Electron., vol. 55, no. 5, pp. 22032212, May 2008.
P. Drabek, Z. Peroutka, M. Pittermann, X. Ce, and M. Dl, New configuration of traction converter with medium-frequency transformer using
matrix converters, IEEE Trans. Ind. Electron., vol. 58, no. 11, pp. 5041
5048, Nov. 2011.
M. Glinka and R. Marquardt, A new AC/AC multilevel converter family,
IEEE Trans. Ind. Electron., vol. 52, no. 3, pp. 662669, Mar. 2005.
C. Zhao, D. Dujic, A. Mester, J. K. Steinke, M. Weiss, S. Lewdeni-Schmid,
T. Chaudhuri, and P. Stefanutti, Power electronic traction transformer
Medium voltage prototype, IEEE Trans. Ind. Electron., vol. 61, no. 7,
pp. 32573268, Jul. 2014.
M. Steiner and H. Reinold, Medium frequency topology in railway applications, in Proc. Eur. Conf. Power Electron. Appl., 2007, pp. 110.
X. She, A. Q. Huang, and G. Wang, 3-D space modulation with voltage
balancing capability for a cascaded seven-level converter in a solid-state
transformer, IEEE Trans. Power Electron., vol. 26, no. 12, pp. 3778
3789, Dec. 2011.
H. Iman-Eini, J. L. Schanen, S. Farhangi, and J. Roudet, A modular
strategy for control and voltage balancing of cascaded H-bridge rectifiers,
IEEE Trans. Power Electron., vol. 23, no. 5, pp. 24282442, May 2008.
C. Cecati, A. DellAquila, M. Liserre, and V. G. Monopoli, Design of
H-bridge multilevel active rectifier for traction systems, IEEE Trans. Ind.
Appl., vol. 39, no. 5, pp. 15411550, Sep. 2003.
J. Shi, W. Gou, H. Yuan, T. Zhao, and A. Q. Huang, Research on voltage
and power balance control for cascaded modular solid-state transformer,
IEEE Trans. Power Electron., vol. 26, no. 4, pp. 11541166, Apr. 2011.
T. Zhao, G. Wang, S. Bhattacharya, and A. Q. Huang, Voltage and power
balance control for a cascaded H-bridge converter-based solid-state transformer, IEEE Trans. Power Electron., vol. 28, no. 4, pp. 15231532, Apr.
2013.
H. Tao, A. Kotsopoulos, J. L. Duarte, and M. A. M. Hendrix, Family
of multiport bidirectional DC-DC converters, IEE Proc. Electr. Power
Appl., vol. 153, no. 3, pp. 451458, May 2006.
C. Gu, H. S. Krishnamoorthy, P. N. Enjeti, Z. Zheng, and Y. Li, A
medium-voltage matrix converter topology for wind power conversion
with medium frequency transformers, J. Power Electron., vol. 14, no. 6,
pp. 11661177, Nov. 2014.
A. Sankala, J. Korhonen, J. Strom, J. Luukko, P. Silventoinen, R. Komulainen, H. Saren, N. Sodo, and D. Isaksson, Modular double-cascade
converter, in Proc. IEEE Appl. Power Electron. Conf. Expo., 2012,
pp. 647652.
C. Gu, H. S. Krishnamoorthy, P. N. Enjeti, and Y. Li, A novel mediumfrequency-transformer isolated matrix converter for wind power conversion applications, in Proc. IEEE Energy Convers. Congr. Expo., 2014,
pp. 10701077.
S. Falcones, R. Ayyanar, and X. Mao, A DC-DC multiport-converterbased solid-state transformer integrating distributed generation and
storage, IEEE Trans. Power Electron., vol. 28, no. 5, pp. 21922203,
May 2013.

GU et al.: MODELING AND CONTROL OF A MULTIPORT POWER ELECTRONIC TRANSFORMER

[32] C. Gu, Z. Zheng, and Y. Li, Control strategy of a multi-level converter


with multi-winding MFT/HFT isolation, in Proc. IEEE Energy Convers.
Congr. Expo., 2014, pp. 37173724.
[33] J. Korhonen, A. Sankala, J. Strom, J. Luukko, P. Silventoinen, R. Komulainen, H. Saren, N. Sodo, and D. Isaksson, Power direction control of
medium frequency isolation DC/DC converter for modular double cascade converter, in Proc. IEEE Appl. Power Electron. Conf. Expo., 2013,
pp. 22402246.
[34] Z. Zheng, Z. Gao, C. Gu, L. Xu, K. Wang, and Y. Li, Stability and voltage
balance control of a modular converter with multiwinding high-frequency
transformer, IEEE Trans. Power Electron., vol. 29, no. 8, pp. 41834194,
Aug. 2014.
[35] T. Sarikurt, C. Sezenoglu, and A. Bahkci, A multi-level inverter system
design with multi-winding transformer, in Proc. Int. Aegean Conf. Elect.
Mach. Power Electron., 2011, pp. 605610.
[36] J. Wang, A. F. Witulski, J. L. Vollin, T. K. Phelps, and G. I. Cardwell,
Derivation, calculation and measurement of parameters for a multiwinding transformer electrical model, in Proc. IEEE Appl. Power Electron. Conf. Expo., 1999, pp. 220226.
[37] C. Sun, N. H. Kutkut, D. W. Novotny, and D. M. Divan, General equivalent circuit of a multi-winding co-axial winding transformer, in Proc.
IEEE IAS Annual Meeting, 1995, pp. 25072514.
[38] A. Frommer, Generalized nonlinear diagonal dominance and applications to asynchronous iterative methods, J. Comput. Appl. Math, vol. 38,
no. 1, pp. 105124, Dec. 1991.
[39] H. Liu and Q. Ni, Incomplete Jacobian Newton method for nonlinear
equations, Comput. Math Appl., vol. 56, no. 1, pp. 218227, Jul. 2008.
[40] B. Zhao, Q. Yu, and W. Sun, Extended-phase-shift control of isolated
bidirectional DC-DC converter for power distribution in microgrid, IEEE
Trans. Power Electron., vol. 27, no. 11, pp. 46674680, Nov. 2012.
[41] F. Krismer and J. W. Kolar, Accurate power loss model derivation of a
high-current dual active bridge converter for an automotive application,
IEEE Trans. Ind. Electron., vol. 57, no. 3, pp. 881891, Mar. 2010.

Chunyang Gu (S12) was born in Heilongjiang,


China, in 1988. She received the B.S. degree in electrical engineering from the Department of Electrical
Engineering, School of Electrical Engineering and
Automation, Harbin Institute of Technology, Harbin,
China, in 2010. She is currently working toward the
Ph.D. degree in electrical engineering at Tsinghua
University, Beijing, China.
Her current research interests include multilevel
converters, power electronic transformers, and the
control of power converters.

Zedong Zheng (M09) was born in Shandong, China,


in 1980. He received the B.S. and Ph.D. degrees
in electrical engineering from the Department of
Electrical Engineering, Tsinghua University, Beijing,
China, in 2003 and 2008, respectively.
He is currently an Associate Professor with the
Department of Electrical Engineering, Tsinghua University. His current research interests include power
electronics converters and high-performance motor
control systems.

927

Lie Xu (M11) was born in Beijing, China, in 1980.


He received the B.S. degree in electrical and electronic engineering from the Beijing University of
Aeronautics and Astronautics, Beijing, China, in
2003, and the M.S. and Ph.D. degrees from the University of Nottingham, Nottingham, U.K., in 2004
and 2008, respectively.
From 2008 to 2010, he was a Research Fellow
with the Department of Electrical and Electronic Engineering, University of Nottingham. He is currently
a Faculty Member with the Department of Electrical Engineering, Tsinghua University, Beijing. His current research interests
include multilevel techniques, multilevel converters, direct acac power conversion, and multilevel matrix converter.

Kui Wang (M11) was born in Hubei, China, in 1984.


He received the B.S. and Ph.D. degrees in electrical
engineering from the Department of Electrical Engineering, Tsinghua University, Beijing, China, in 2006
and 2011, respectively.
He is currently a Faculty Member with the Department of Electrical Engineering, Tsinghua University.
His current research interests include multilevel converters, control of power converters, and adjustablespeed drives.

Yongdong Li (M08) was born in Hebei, China, in


1962. He received the B.S. degree from the Harbin Institute of Technology, Harbin, China, in 1982, and the
M.S. and Ph.D. degrees from the Department of Electrical Engineering, Institut National Polytechnique
de Toulouse, Toulouse, France, in 1984 and 1987,
respectively.
Since 1996, he has been a Professor with the Department of Electrical Engineering, Tsinghua University, Beijing, China. From 2011 to 2014, he was the
Dean of the School of Electrical Engineering, Xinjiang University, Urumqi, China. He was also an Invited Professor with the
Institut National Polytechnique de Toulouse. His current research interests include power electronics, machine control, and wind power generation.
Dr. Li is a Senior Member of the China Electro-Technique Society, the Vice
Chairman of the China Power Electronics Society, and the Vice Chairman of
the Electrical Automation Committee of China Automation Association.

Das könnte Ihnen auch gefallen