Received April 24, 2019, accepted May 3, 2019, date of publication May 16, 2019, date of current version

June 4, 2019.
Digital Object Identifier 10.1109/ACCESS.2019.2917320

Interpreting the Individual Capacitor Voltage

Regulation Control of PSC-PWM
MMC via Consensus Theory
SHUGUANG SONG , (Student Member, IEEE), AND JINJUN LIU , (Fellow, IEEE)
State Key Laboratory of Electrical Insulation and Power Equipment, School of Electrical Engineering, Xi’an Jiaotong University, Xi’an 710049, China
Corresponding author: Jinjun Liu (jjliu@mail.xjtu.edu.cn)

ABSTRACT In computer science, consensus theory has been studied for decades. ‘‘Consensus’’ in a multi-
agent system means reaching an agreement regarding a certain quantity of interest. The modular multilevel
converter (MMC) is a promising topology for medium- and high-power applications owing to its salient
features. In MMC operation, one key point is to regulate the capacitor voltage of each submodule (SM).
The SMs in MMCs are interconnected both physically and via a control system, and these SMs can be
regarded as a multi-agent system, with the control objective of regulating their capacitor voltages, which
could formulate a consensus problem. In this paper, through mathematical derivation, it is found that
individual capacitor voltage regulation control of phase-shifted-carrier pulse-width modulation (PSC-PWM)
MMC is one special case of consensus theory. Furthermore, based on the consensus theory, the influence of
different communication structures on the MMC voltage regulation performance is discussed. The analysis
proves that the existing one has the highest rate of convergence. The impact of current-tracking dynamics on
consensus-based interpretation is also analyzed. In addition, the proportional relationship between the rate of
convergence and the P controller parameter is also justified. The experiments on a seven-level down-scaled
prototype are conducted to confirm the analysis.

INDEX TERMS Consensus theory, modular multilevel converter, multi-agent system, PSC-PWM, voltage
regulation.

I. INTRODUCTION regulation control. The first three level controls can be

The modular multilevel converter (MMC) has attracted con-
siderable interest recently. Compared with the traditional
neutral point clamp or flying-capacitor-based multilevel con-
verter, the MMC possesses inherent benefits such as high
redundancy, high modularity, high controllability, and excel-
lent output voltages and currents. It is regarded as a promising
topology for medium- and high-power applications [1]–[8].
To ensure stable operation of the MMC, submodule (SM)
capacitor voltages need to be regulated at its reference value,
i.e., Udc /N , where Udc denotes the DC-link voltage of the
MMC and N denotes the number of SMs in one arm. This can
be achieved by four-level voltage regulation control [6], [9],
i.e., 1. total capacitor voltage control (in rectifier mode),
2. phase capacitor voltage balance control, 3. arm capacitor
voltage balance control, and 4. individual capacitor voltage
The associate editor coordinating the review of this manuscript and
approving it for publication was Huiqing Wen.
realized by adjusting the output active current and the
MMC inner circulating current, while, for individual capac-
itor voltage regulation control, the control strategy depends
on the specific application.
In MMC-based high-voltage direct current (HVDC) sys-
tems, the number of SMs in one arm can be up to several
hundred. To make this feasible, nearest level modulation
(NLM) [10]–[12] combined with a capacitor voltage sorting
algorithm is adopted to follow the output voltage reference
and regulate all SM capacitor voltages. Because of the
large number of SMs in one arm, the output waveform is
a good approximation to the voltage reference. In addition,
improvements have also been made to reduce the switch-
ing frequency [13], [14] and to increase the output volt-
age levels [15]. In MMC-based medium-voltage systems,
the number of SMs in one arm is insufficient to support the
NLM method to guarantee output quality. Unneglectable
low-order harmonics will appear. In this case,

2169-3536 2019 IEEE. Translations and content mining are permitted for academic research only.
VOLUME 7, 2019 Personal use is also permitted, but republication/redistribution requires IEEE permission. 66807
See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
 S. Song, J. Liu: Interpreting the Individual Capacitor Voltage Regulation Control of PSC-PWM MMC via Consensus Theory
phase-shifted-carrier pulse-width modulation (PSC-PWM)
is advantageous. The PSC-PWM method enables low-
frequency harmonics to be canceled out and increases
the equivalent switching frequency by properly assign-
ing the phase angle displacement of each triangular car-
rier [6], [16]–[20]. The corresponding individual voltage
regulation control is based on the closed-loop controller
for each SM. This controller compares the actual capacitor
voltage and average arm capacitor voltage, and the error is
sent to a proportional controller to generate the adjusting
signal.
In computer science, the consensus problem has been
studied for a long time [21]–[26]. In multi-agent systems,
the word ‘‘consensus’’ means that a certain quantity of inter-
est comes to an agreement, based on the states of all agents.
Recently, compelling attention has been paid to the con-
sensus problem, because of the wide application of multi-
agent systems in many important domains such as microgrids,
unmanned air vehicles, flocking, and so on [27]–[32].
SMs in MMCs are interconnected both physically and via
a control system, and these SMs can be regarded as a multi-
agent system, with the control objective of regulating their
capacitor voltages, which could formulate a consensus prob-
lem. Therefore, consensus theory may also be able to realize
SM individual capacitor voltage regulation control, or the
existing control strategy can be interpreted by consensus
theory. Therefore, presenting a clear discussion of this subject
will be very essential. To fulfill this need, this paper aims
to establish the relationship between the individual capacitor
voltage regulation control and consensus theory.
Through mathematical derivation, we deeply analyze
PSC-PWM closed-loop-based individual capacitor voltage
regulation control and find that individual capacitor volt-
age regulation control of PSC-PWM MMC is one special
case of consensus theory. This gives an interpretation of
MMC individual capacitor voltage regulation via a more
general theory, which, in other words, also provides a the-
oretical proof of its validity. In addition, based on consensus
theory, we demonstrate that the communication structure of
multi-agent systems will also influence system performance.
We compare three different communication structures and
discuss their impact on MMC control performance. This
proves that the existing communication structure in MMC
achieves the highest rate of convergence. The impact of
current-tracking dynamics on consensus-based interpretation
is also discussed. A second-order circulating current con-
troller is added to the MMC control system. Voltage reg-
ulation is tested in both current tracked with large errors
and current well-tracked situations, showing that the current-
tracking dynamics will not deteriorate the consensus-based
interpretation. Besides, the proportional relationship between
the rate of convergence and the P controller parameter is also
justified. Experiments on a seven-level down-scaled proto-
type are conducted in the laboratory. The experimental results
coincide well with the theoretical analysis, confirming the
validity of the theory.
66808
FIGURE 1. Graphical representation of a multi-agent system. Each vertex
indicates one agent, e.g., a converter, a submodule, etc.
The interpretation of the individual capacitor voltage reg-
ulation control of PSC-PWM MMC via consensus theory in
this paper provides a different viewpoint for researchers and
power electronics engineers to understand control of power
electronics systems. Through consensus theory, the conver-
gence of individual voltage regulation control is justified.
Moreover, the discussion of the P controller parameter could
be helpful in the design of system control parameters.
II. CONSENSUS THEORY AND CAPACITOR VOLTAGE
REGULATION CONTROL IN AN MMC
A. CONCEPT OF CONSENSUS THEORY
First, the concept of consensus theory is presented. A multi-
agent system can be represented by a graph S = (VS , ES ),
where VS is a finite nonempty set of agents named ver-
tices and ES is a set describing communications between
agents named edges. Given the features of power elec-
tronics systems, this discussed graph is considered to
be unweighted, with undirected edges, without loops and
without multi-edges.
The adjacency matrix AS of graph S is a symmetric matrix
depicting the connection of vertices of this graph, with entries
defined as
(
1, if ES [i, j] = 1,
aij = (1)
0, otherwise.
The adjacency matrix of the example system in Fig. 1 then
can be represented by

0 1 0 0 1 1

1 0 0 1 0 0

0 0 0 0 1 0

0 1 0 0 0 0. (2)
 

1 0 1 0 0 0

1 0 0 0 0 0
Entries a21 and a12 are both equal to 1, which means that there
is an undirected edge between agent 1 and agent 2.
The degree matrix of DS of graph S is a diagonal matrix
with vertex degrees along its diagonal, where the degree of
the vertex stands for the number of vertices connected to it.
VOLUME 7, 2019
S. Song, J. Liu: Interpreting the Individual Capacitor Voltage Regulation Control of PSC-PWM MMC via Consensus Theory

Apparently, the degree matrix of the example system

in Fig. 1 is
 
3 0 0 0 0 0
0 2 0 0 0 0
 
0 0 1 0 0 0

0 0 0 1
 (3)
 0 0
0 0 0 0 2 0
0 0 0 0 0 1
Here, we define x as the quantity of interest. For different
applications of consensus theory, this quantity may be posi-
tion, speed, voltage, etc. To ensure that the quantity of interest
reaches an agreement, a linear consensus algorithm can be
defined as [22]
n
X
ẋi = C aij (xj − xi ), (4)
j

where C is the diffusion coefficient.

Consensus in a multi-agent system means the quantity of
interest x of each agent will finally be equal to each other:
n
1 X
lim xi (t) = lim xj (t) = xk (0).
t→+∞ t→+∞ n the graph.
k
Take agent 2 in Fig. 1 for example. As seen in (4), in its
control system, the difference in x between agent 2 and
connected agents, i.e. agent 1 and agent 4 in this system, will
be calculated. The sum of these differences will be added
together, generating a dynamic ẋ. This dynamic variable
makes agent 2 adjust its state.
From the lemma in [23], the linear consensus algorithm
defined in (4) will asymptotically solve a consensus problem
for all initial states if graph S is a connected undirected graph.
An undirected graph is a graph in which edges have no
orientation. In other words, the flow through edges is bidirec-
tional. A connected undirected graph is one such that there is
a path from any vertex to any other vertex in the graph.
This system can also be expressed in the compact form
ẋ = −C (DS − AS ) x = −CLS x, (6)
   
ẋ1 x1
 ẋ2   x2 
where ẋ =  .  , x =  . , and the matrix LS is called
   
 ..   .. 
ẋn xn
the Laplacian of the graph.
The Laplacian of graph LS in the analysis of the corre-
sponding linear consensus algorithm is crucial. LS will be
symmetric and possess real eigenvalues in undirected graphs.
The set of eigenvalues can be listed as
0 = λ1 ≤ λ2 ≤ · · · ≤ λn .
The second smallest eigenvalue of the Laplacian, λ2 , is
called the algebraic connectivity of the graph [33]. More
importantly, it is a measure of performance and speed of the
consensus algorithms [22]. From the definition of the graph
FIGURE 2. MMC circuit configuration.
Laplacian, it is clear that LS is determined by the structure of
(5)
For more details on consensus theory, see [22], [23],
and [34].
In power electronics, the agents can be power converters,
SMs in MMCs, etc. The quantity of interest can be voltage,
current, active or reactive power, etc. The differences of this
quantity of interest between different agents will generate a
dynamic ẋi . Apparently, this dynamic variable is associated
with the adjacency matrix, i.e., the system communication
structure. This dynamic variable will cause flow of the quan-
tity of interest among agents. If the system satisfies the
requirements of the lemma, a consensus will be reached, i.e.,
voltage and current will be balanced, and active and reactive
power will be equally shared, etc.
B. MMC BASIC OPERATING PRINCIPLES
A typical three-phase MMC system is shown in Fig. 2. In each
phase of the MMC, there is two arms, upper and lower arms.
Each arm consists of a series connection of N SMs, an arm
inductor, and a resistor indicating the loss. On the right-hand
side, the structure of the half-bridge SM is presented. The
SM consists of two switching devices (S1 and S2) and one
capacitor. The SM can output two voltage levels: 0 and Usm .
When the gate signal received is 1, switch S1 is turned on,
and S2 is off. The SM is inserted into the circuit, being
charged or discharged, according to the current direction.
When the gate signal received is 0, switch S1 is turned off,
and S2 is on. The SM is bypassed.
(7) Applying Kirchhoff’s voltage law to the MMC upper
and lower arms, respectively, gives mathematical model of
an MMC:
Udc diU ,x
ux − + uU .x + L + RiU ,x = 0 (8)
2 dt

VOLUME 7, 2019 66809

 S. Song, J. Liu: Interpreting the Individual Capacitor Voltage Regulation Control of PSC-PWM MMC via Consensus Theory

and
Udc diL,x
ux + − uL,x − L − RiL,x = 0, (9)
2 dt
where ux is the MMC output voltage, Udc is the DC link volt-
age, uU .x and uL,x are upper and lower arm output voltages,
and iU ,x and iL,x are upper and lower arm currents. x = a, b,
and c denotes phases A, B, and C, respectively.
The circulating current in an MMC is defined as
iU ,x + iL,x
icir,x = . (10)
2
The AC side output current can be expressed as
ix = iU ,x − iL,x . (11)
By adding (8) and (9) together and subtracting (8) from (9),
we can get
uL ,x − uU ,x L dix R
ux = − − ix (12)
2 2 dt 2
and
Udc uL ,x + uU ,x dicir,x
− =L + Ricir,x . (13)
2 2 dt
Equations (12) and (13) show the MMC outer and inner char-
acteristics. We can find that the MMC output is determined by
the difference between upper and lower arm output voltages,
whereas the MMC inner circulating current is influenced
by the sum of upper and lower arm output voltages. Con-
sequently, the MMC system can be controlled by adjusting FIGURE 3. Control diagram for capacitor voltage regulation control
MMC upper and lower arm output voltages. system.

C. CAPACITOR VOLTAGE REGULATION CONTROL AND shown in Figs. 3(a) and 3(b), respectively. Fig. 3(c) depicts
PSC-PWM circulating current control.
In the operation of an MMC, capacitor voltage regulation In Fig. 3(d), the individual capacitor voltage is compared
is very important. SM capacitor voltage regulation control with the arm average value. The error passes a P controller,
consists of four levels: then is multiplied by the corresponding arm current, finally
1. total capacitor voltage control (in rectifier mode), giving the adjusting signal. The output 1u∗BU ,i,x will be part
2. phase capacitor voltage balance control, of the modulation signal for the SMs. Based on the voltage
3. arm capacitor voltage balance control, and regulation control system, the final modulation signals for
4. individual capacitor voltage regulation control. each SM can be synthesized by
Total capacitor voltage control is achieved by control- Udc u∗x

uU ,i,x = 2N − N + 1uA,x + 1uBU ,i,x ,
 ∗ ∗ ∗
ling the active current of the system. This level con- 
trol exists only under rectifier mode operation of the (14)
MMC when no stiff DC source is connected to the MMC
 Udc u∗x
u∗ + 1uA,x + 1uBL,i,x ,
∗ ∗

L,i,x = +
DC link. Phase capacitor voltage control balances the power 2N N
among the three phases by adjusting the DC circulating where Udc denotes the DC link voltage, u∗x is the output
current. Arm capacitor voltage control balances the capaci- voltage reference, subscripts U and L denote upper and lower
tor voltage between upper and lower arms by adjusting the arms, and i= 1−N denotes the ith SM in one arm.
fundamental frequency circulating current. If these controls Then the PSC-PWM will generate the gate signals for the
work well, the average capacitor voltage of one arm will switching devices. Fig. 4 shows the operation principle of
be well maintained to its reference value. Then, the fourth the PSC-PWM. For an MMC system with N SMs per arm,
level control, i.e., individual capacitor voltage regulation two sets of N triangular carriers (of frequency fc and, angular
control, will regulate the capacitor voltages of each SM to frequency ωc ) with phase shift 2π/N are needed. Carriers
be balanced. The corresponding control diagrams of phase, CU 1 to CUN are for the upper arm, and carriers CL1 to CLN are
arm, and individual voltage control are shown in Fig. 3. The for the lower arm. Each SM possesses one modulation signal
generation of DC and AC circulating current references is as synthesized in (14) and is assigned to a triangular carrier.

66810 VOLUME 7, 2019

S. Song, J. Liu: Interpreting the Individual Capacitor Voltage Regulation Control of PSC-PWM MMC via Consensus Theory

the average upper arm voltage ūsmU ,x can be calculated as

N
1 X
ūsmU ,x = usmU ,j,x . (15)
N
j=1

The corresponding error signal for the ith SM is then obtained

as

errorU ,i,x = ūsmU ,x − usmU ,i,x

N
1 X
= usmU ,j,x − usmU ,i,x . (16)
N
j=1

The equation above can be transformed as

PN
j=1 usmU ,j,x − NusmU ,i,x
errorU ,i,x =
N
N
1 X
usmU ,j,x − usmU ,i,x ,


= (17)
N
j=1

and then the final adjusting signal of individual voltage regu-

lation control will be
N
KP iU ,x X
1u∗BU ,i,x = usmU ,j,x − usmU ,i,x .


(18)
N
j=1

According to the final adjusting signal in (18), the corre-

FIGURE 4. PSC-PWM of an MMC. (a) Upper arm. (b) Lower arm.
sponding power flow generated can be expressed as

1PU ,i,x = iU ,x · 1u∗BU ,i,x

Then the modulation signal compares with the corresponding
triangular carrier; if the modulation signal is greater than the
triangular carrier, the gate signal is 1, and the SM will be in
the on state (and vice versa).
Figs. 4(a) and 4(b) show the upper and lower arm triangular
carriers, respectively. The modulation signals of the first SM
in the upper and lower arms are also depicted. The corre-
sponding gate signals are given as well.
The phase shift of triangular carriers results in a high equiv-
alent switching frequency (Nf c ) and low harmonic distortion.
The phase displacement between the upper arm carriers and
lower arm carriers θ will also influence the harmonic features
of the MMC [18].
III. RELATIONSHIP BETWEEN CONSENSUS THEORY AND
MMC CONTROL
A. EQUIVALENCE
In the following analysis, the assumption has been made that
all other level controls operate normally. Therefore, only the
individual capacitor voltage regulation control in one arm is
considered. Under this assumption, the SMs in one arm can
be regarded as a multi-agent system, and the capacitor voltage
of each SM is the quantity of interest for which we want to
reach a consensus.
Because the MMC is symmetrical, we take one upper arm
for an example in the analysis. From the control diagram,
N
KP i2U ,x X

= usmU ,j,x − usmU ,i,x (19)
N
j=1
When the individual capacitor voltage regulation control
works, the control system will give the adjusting signal and
generate the power flow to adjust the capacitor voltage. The
power flow will finally be zero when the individual capacitor
voltage regulation control achieved. Comparing (19) with
the linear consensus algorithm defined in (4), we find that
these two equations have a similar form. To gain a deeper
understanding, here, as in consensus theory, we define a
simple graph C, where the SMs are the vertices and the
communications among SMs are the edges.
Typically, centralized control strategies are used in an
MMC, which means that all the measured quantities (capac-
itor voltages, arm currents, output currents, etc.) are sent
to the digital signal processor (DSP). Then the regulation
control is also executed in the DSP. Therefore, the capacitor
voltages of each SM are all stored in the DSP and these are
used to calculate the adjusting signal in (18). In this case,
we can say that there is communication among all SMs in one
arm. Correspondingly in the graph, all vertices are connected
together. Therefore, the simple graph C is a complete graph,
as shown in Fig. 5(a). In this paper, we take an MMC system
with six SMs in one arm as an example. The corresponding

VOLUME 7, 2019 66811

 S. Song, J. Liu: Interpreting the Individual Capacitor Voltage Regulation Control of PSC-PWM MMC via Consensus Theory

FIGURE 5. Graphical representation of MMC control systems: (a) graph C , controller taking the average of
all SM voltages, (b) graph C1 , controller taking the average of a portion of SM voltages, and (c) graph C2 ,
controller taking the average of neighboring SM voltages.

   
adjacency matrix, degree matrix, and graph Laplacian are errorU ,1,x usmU ,1,x
 errorU ,2,x   usmU ,2,x 
errorU ,x = .. , usmU ,x = .. .
     
0 1 1 1 1 1
1 0 1 1 1 1
 .   . 

1 1 0 1 1 1
 errorU ,N ,x usmU ,N ,x
AC =   , (20)
1 1 1 0 1 1

KP i2
1 1 1 1 0 1 It is now clear that (25) is identical to (6), where NU ,x
1 1 1 1 1 0 is the diffusion coefficient. From the analysis above, we can

5 0 0 0 0 0
 find that the existing individual capacitor voltage regulation
0 5 0 0 0 0 control in PSC-PWM MMC is one special case of consensus

0 0 5 0 0 0
 theory. In addition, the capacitor voltage will definitely con-
DC =   , (21) verge to its average value (under the assumption that other
0 0 0 5 0 0

0 0 0 0 5 0 level controls are achieved), because graph C is a connected
0 0 0 0 0 5 graph, based on the lemma in [23]. It should be noted that the
diffusion coefficient here is time varying for the arm current
and is periodic. However, this coefficient will only influence the
  rate of convergence [22], and it will not impact the consen-
5 −1 −1 −1 −1 −1
sus. The individual capacitor voltage regulation control of
 −1 5 −1 −1 −1 −1 
  PSC-PWM MMC is now interpreted via consensus the-
 −1 −1 5 −1 −1 −1 
LC =   (22) ory. In addition, stability of MMC capacitor voltage reg-
 −1 −1 −1 5 −1 −1 
  ulation control has also been proved in [35] by using the
 −1 −1 −1 −1 5 −1 
‘‘Routh–Hurwitz stability criterion.’’
−1 −1 −1 −1 −1 5
It needs to be pointed out that the diffusion coefficient
The following compact forms of (17), (18) and (19) can contains the MMC arm current, iU ,x , for the upper arm
also be obtained: (iL,x for the lower arm), which is related to the MMC
operation state. As mentioned in the beginning, we assume
1 that other level controls are well achieved, so iU ,x tracks
errorU ,x = − LC usmU ,x , (23)
N its reference i∗U ,x well. Thus, no impact of current-tracking
KP iU ,x dynamics will be introduced to the individual capacitor volt-
1u∗BU ,x = KP iU ,x errorU ,x = − LC usmU ,x , (24)
N age regulation control. In practice, tracking dynamics always
KP i2U ,x exist. However, for SMs in one arm, the arm current flowing
1PU ,x = iU ,x ·1u∗BU ,x =− LC usmU ,x , (25)
N into them is the same. In addition, arm current in consensus
theory is only part of the diffusion coefficient and it may
where
only influence the rate of convergence. Therefore, current-
1PU ,1,x 1u∗BU ,1,x tracking dynamics will not deteriorate the consensus-based
   
 1PU ,2,x  1u∗ interpretation.
BU ,2,x 
 
1PU ,x = .. , 1u∗BU ,x = .. , To verify this, a second-order circulating current sup-
  
 .   .  pression controller is added to the MMC control system.
1PU ,N ,x 1u∗BU ,N ,x In the circulating current, the second-order harmonic is one

66812 VOLUME 7, 2019

S. Song, J. Liu: Interpreting the Individual Capacitor Voltage Regulation Control of PSC-PWM MMC via Consensus Theory

The Laplacians of graphs C1 and C2 are

 
3 −1 0 −1 0 −1
 −1
 3 −1 0 −1 0 
 0 −1 3 −1 0 −1 
LC 1 =   (28)
 −1
 0 −1 3 −1 0 
 0 −1 0 −1 3 −1 
−1 0 −1 0 −1 3
FIGURE 6. Control diagram of the second-order harmonic elimination
controller. and
 
2 −1 0 0 0 −1
 −1 2 −1 0 0 0 
important component. Its existence increases the RMS value 
 0 −1 2 −1 0

0 
of arm currents, causing extra loss in the system. Conse- LC 2 =  . (29)
 0 0 −1 2 −1 0 
quently, many suppression methods have been proposed [13], 
 0 0 0 −1 2

−1 
[36]–[39]. In this paper, the proportional resonant (PR) con- −1 0 0 0 −1 2
troller is adopted. The control diagram is shown in Fig. 6. The
second-order harmonic is first extracted from the circulating As mentioned earlier, the performance can be measured
current. Then, through the PR controller, the corresponding by the second smallest eigenvalue of the Laplacian, λ2 .
control voltage 1u∗cir,x is generated. By using MATLAB, the eigenvalues of LC , LC 1 , and LC 2
The transfer function of the PR controller can be written as can be calculated, giving λ2,C = 6.0000, λ2,C1 = 3.0000,
and λ2,C2 = 1.0000. Apparently, the algebraic connectivity
KR,PR s of graph C is the highest. Theoretically, the rate of con-
FPR (s) = KP,PR + , (26)
s2 + (2ωo )2 vergence will also be the highest (using the same control
parameters). A mathematical model in MATLAB has been
where KP,PR is the P parameter, KR,PR is the R parameter, and built. This system has six agents with different initial states.
ωo is the angular frequency (100π). Figs. 7(a)–7(c) show the results of the corresponding commu-
This control voltage is added to the modulation signals nication structures C, C1 , and C2 , respectively. The conver-
in (14). The new modulation signals for SMs are gence times of these three cases are in the ratio of 55:110:310,
which are very close to the ratio of λ2,C :λ2,C1 :λ2,

Udc u∗x C2 = 16 : 31 : 11 .
uU ,i,x = 2N − N + 1uA,x + 1uBU ,i,x − 1ucir,x ,

 ∗ ∗ ∗ ∗

(27) IV. EXPERIMENTAL VERIFICATION

 Udc u∗x
u∗ + 1uA,x + 1uBL,i,x − 1ucir,x .
∗ ∗ ∗

L,i,x = +
2N N phase MMC prototype is built and tested in the laboratory.
Voltage regulation is tested under different situations: cur-
rent tracked with large errors and current well-tracked. The
experimental results are given in Section IV.
B. PERFORMANCE
Consensus will be reached as long as the graph is connected.
Therefore, MMC voltage regulation control systems with dif-
ferent communication structures are also capable of achieving
capacitor voltage regulation. To discuss the impact on control
performance, we here compare different SM communication
structures. We define two new graphs C1 and C2 . These are
represented in Figs. 5(b) and 5(c). Obviously, they are con-
nected graphs. In graph C1 , the controller takes the average
of a portion of the SM voltages, and in graph C2 , the controller
takes the average of neighboring SM voltages.
It should be noted that these two SM communication struc-
tures are not actually used in real MMC control systems.
The discussion of these two structures aims to understand
MMC voltage regulation control system within the aspect of
consensus theory.
To verify the analysis in this paper, a seven-level single-
Fig. 8 shows the whole system setup. The DC link voltage is
supported by two cascaded Chroma programmable DC power
supplies (62150H-100S), which also create a neutral point
on the DC side. Thus, the DC link capacitors are clamped
by these two DC voltage sources, and no charging or dis-
charging control is needed in the experiment. A Myway
PE-Expert4 digital control system is used. PE-Expert4 is
equipped with a TI TMS320C6657 1.25-GHz dual-core DSP,
which achieves the control algorithm, with an FPGA and
corresponding optical output, allowing reliable communica-
tions, and with several ADC modules.
The circuit parameters are listed in Table 1. Since this paper
mainly focuses on the MMC capacitor voltage regulation
control, the MMC system in the experiment is designed to
drive passive loads. Owing to the limitation of the mea-
surement devices, only the SM1 and SM6 capacitor volt-
ages of the lower arm are measured and displayed on the
oscilloscope.
In conventional individual voltage regulation control,
the adjusting signals are calculated in (18), whereas in the
discussed communication structures, i.e., Figs. 5 (b) and 4(c),

VOLUME 7, 2019 66813

 S. Song, J. Liu: Interpreting the Individual Capacitor Voltage Regulation Control of PSC-PWM MMC via Consensus Theory

FIGURE 8. Prototype in laboratory.

TABLE 1. Experimental platform parameters.

FIGURE 7. Simulation results under different communication structures:

(a) graph C , controller taking the average of all agents, (b) graph C1 ,
controller taking the average of a portion of agents, and (c) graph C2 ,
controller taking the average of neighboring agents.


KP iU
1u∗BU ,1,C2 = usmU ,6 −2usmU ,1 +usmU ,2 ,


the new adjusting signals are expressed by 


 6
 KP iU
1u ∗ usmU ,1 −2usmU ,2 +usmU ,3 ,


KP iU ,x 
,2,C =
1u∗BU ,x,C1 = − LC 1 usmU ,x ,

(30)


 BU 2 6
N  KP iU
1u∗ usmU ,2 −2usmU ,3 +usmU ,4 ,


KP iU ,x

BU ,3,C2 =
1u∗BU ,x,C2 =− LC 2 usmU ,x . (31) 6
KP iU (33)
N 1u ∗ = u −2u +u


,
,3 ,4 ,5

,4,C smU smU smU
 BU 2 6


By substituting (28) and (29) into (30) and (31), respec- 
 KP iU
1uBU ,5,C2 =
∗ usmU ,4 −2usmU ,5 +usmU ,6 ,



tively, the corresponding adjusting signals for each individual 


 6
SM can be obtained directly: 
1u∗ KP iU
usmU ,5 −2usmU ,6 +usmU ,1 .


BU ,6,C2 = 6


KP iU
1u∗BU ,1,C1 = usmU ,6−3usmU ,1 +usmU ,2 +usmU ,4 ,


where, for simplicity, the subscript x = a, b, and c denoting


6


phases A, B, and C is omitted in the final expression.

 KP iU
1u ∗ usmU ,1−3usmU ,2 +usmU ,3 +usmU ,5 ,




BU ,2,C =
1 6



 KP iU A. PERFORMANCE EXPERIMENTS
1u∗ usmU ,2−3usmU ,3 +usmU ,4 +usmU ,6 ,



BU ,3,C1 =
6 To show the rate of convergence of each communication
KP iU
1u ∗ ,



 BU ,4,C = u smU ,3−3usmU ,4 +u smU ,5 +usmU ,1 structure, the individual capacitor voltage regulation con-
1 6



 KP iU trol is disabled at the beginning. Voltage deviation occurs,
1u∗BU ,5,C1 = usmU ,4−3usmU ,5 +usmU ,6 +usmU ,2 ,





 6 as shown in Figs. 9(a)–9(c), which is then subjected to


1u∗ KP iU active regulation control. Fig. 9(a) shows the experimen-
usmU ,5−3usmU ,6 +usmU ,1 +usmU ,3 ,


BU ,6,C1 = 6

tal waveforms of the existing communication structure,
(32) i.e., graph C. The capacitor voltages of SM1 and SM6 quickly

66814 VOLUME 7, 2019

S. Song, J. Liu: Interpreting the Individual Capacitor Voltage Regulation Control of PSC-PWM MMC via Consensus Theory
FIGURE 9. Lower arm capacitor voltages: (a) graph C , controller taking
the average of all SM voltages, (b) graph C1 , controller taking the average
of a portion of SM voltages, and (c) graph C2 , controller taking the
average of neighboring SM voltages.
converge to the reference value. The response time
is ∼ 0.7 s. The voltage waveforms with communication
structure C1 are depicted in Fig. 9(b). The capacitor voltages
are also well regulated after activation. However, the response
time is ∼1.4 s, which is a factor of ∼2 longer than the previous
one. Fig. 9(c) shows the result of structure C2 , for which
the corresponding convergence time is ∼4.3 s. Note that
the control parameters used in these experiments are all the
same.
VOLUME 7, 2019
FIGURE 10. Lower arm capacitor voltages in load change test:
(a) graph C , controller taking the average of all SM voltages,
(b) graph C1 , controller taking the average of a portion of SM voltages,
and (c) graph C2 , controller taking the average of neighboring
SM voltages.
A load change test has also been conducted in the labo-
ratory. The resistance load is set to be 50 ohm at the begin-
ning. After some time, it is switched to 20 ohm. The results
for these three structures are shown in Fig. 10. Apparently,
the capacitor voltages are all well regulated in all these
cases.
In addition, Fig. 11 gives the experimental waveforms for
the MMC output voltage, output current, upper arm current,
and lower arm current of these three structures in steady state.
66815
 S. Song, J. Liu: Interpreting the Individual Capacitor Voltage Regulation Control of PSC-PWM MMC via Consensus Theory
FIGURE 11. MMC output voltages, output currents, upper arm currents,
and lower arm currents. (a) graph C , controller taking the average of all
SM voltages, (b) graph C1 , controller taking the average of a portion of
SM voltages, and (c) graph C2 , controller taking the average of
neighboring SM voltages.
The harmonic components and Total Harmonic Distortion
(THD) of output voltages and currents are also presented
in Figs. 12 and 13. We can see that the harmonic components
and THD are almost identical, indicating that different struc-
tures have no impact on the MMC operation.
The experimental results above verify that 1. consensus
(voltage regulation) is achieved in connected graphs and 2.
the existing communication connection has a higher rate
of convergence. It is observed that, in the experiment, the
66816
FIGURE 12. Harmonic components and THD of output voltages:
(a) graph C , controller taking the average of all SM voltages, (b) graph C1 ,
controller taking the average of a portion of SM voltages, and
(c) graph C2 , controller taking the average of neighboring SM voltages.
convergence times of different graphs are 0.7 : 1.4 : 4.3 ≈
1 1 1
6 : 3 : 1 , which coincides well with the theoretical analysis.
In addition, it has been proved that the algebraic connectivity
will reach its maximum value in a complete graph [40], and
the rate of convergence will also be the highest. This leads to
the conclusion that the existing individual capacitor voltage
regulation control is the fastest. It needs to be pointed out
that the only difference among these three communication
structures in MMC control is the adjusting signal calcula-
tion equation. Therefore, the complete-graph communication
structure in an MMC control system is always applied to give
the best control performance.
VOLUME 7, 2019
S. Song, J. Liu: Interpreting the Individual Capacitor Voltage Regulation Control of PSC-PWM MMC via Consensus Theory
FIGURE 13. Harmonic components and THD of output currents:
(a) graph C , controller taking the average of all SM voltages, (b) graph C1 ,
controller taking the average of a portion of SM voltages, and
(c) graph C2 , controller taking the average of neighboring SM voltages.
B. CURRENT-TRACKING DYNAMICS EXPERIMENTS
To test the impact of current-tracking dynamics on the
consensus-based interpretation, second-order circulating
current suppression controller is added to the MMC con-
trol system to eliminate the second-order component in the
arm current. If this controller works well, arm currents will
only contain DC and fundamental frequency components.
Figs. 14(a)–14(c) give the arm currents in different situations:
Fig. 14(a) for the circulating current controller disabled,
with a large second-order component appearing in the arm
currents; Fig. 14(b) for the current tracked with large errors,
where a second-order component can still be observed;
VOLUME 7, 2019
FIGURE 14. Arm currents with second-order circulating current controller:
(a) controller disabled, (b) current tracked with large errors, and
(c) current well tracked.
Fig. 14(c), for a well-tracked current, in which only DC and
fundamental frequency components appear in the waveform.
For the tracked with large errors and current well-tracked
situations, the corresponding lower arm capacitor voltages are
shown in Figs. 15 and 16, respectively. The capacitor volt-
ages with circulating current controller disabled are already
shown in Fig. 9. It is obvious that, though the current-tracking
dynamics are different, the experimental results are simi-
lar and the convergence times of different communication
structures are ∼ 61 : 31 : 11 . In other words, the current tracking
66817
 S. Song, J. Liu: Interpreting the Individual Capacitor Voltage Regulation Control of PSC-PWM MMC via Consensus Theory
FIGURE 15. Lower arm capacitor voltages when current tracked with
large errors: (a) graph C , controller taking the average of all SM voltages,
(b) graph C1 , controller taking the average of a portion of SM voltages,
and (c) graph C2 , controller taking the average of neighboring SM
voltages.
dynamics will not deteriorate the consensus-based interpreta-
tion, confirming the analysis in Section III.
C. P CONTROLLER PARAMETER EXPERIMENTS
The experiments above tested the impact of current-tracking
dynamics. Because the P controller parameter KP is also
part of the diffusion coefficient, experiments were conducted
to verify its influence. Figs. 17(a) –17(c) show the lower
66818
FIGURE 16. Lower arm capacitor voltages when current well tracked:
(a) graph C , controller taking the average of all SM voltages, (b) graph C1 ,
controller taking the average of a portion of SM voltages, and
(c) graph C2 , controller taking the average of neighboring SM voltages.
arm capacitor voltages in different communication structures,
while the P controller parameter KP is doubled compared to
that used in Figs. 9(a) –9(c). The settling time in Fig. 17(a) is
∼0.3 s, ∼0.6 s in Fig. 17(b), and ∼2.0 s in Fig. 17(c).
As discussed in [22], the rate of convergence is propor-
tional to the diffusion coefficient. From the experimental
results, this is also verified: KP doubled and convergence time
was almost halved. This relationship is helpful in designing
system control parameters. Based on the existing rate of
VOLUME 7, 2019
S. Song, J. Liu: Interpreting the Individual Capacitor Voltage Regulation Control of PSC-PWM MMC via Consensus Theory
FIGURE 17. Lower arm capacitor voltages, KP doubled: (a) graph C ,
controller taking the average of all SM voltages, (b) graph C1 , controller
taking the average of a portion of SM voltages, and (c) graph C2 ,
controller taking the average of neighboring SM voltages.
convergence, we can directly adjust the P controller parame-
ter KP according to the proportional relationship to get the
desired control performance. This linear relationship may
also be derived through a system transfer function, while with
help of consensus theory, it can be directly obtained.
V. CONCLUSION
In this paper, based on mathematical derivation, we find that
individual capacitor voltage regulation control in PSC-PWM
VOLUME 7, 2019
MMC can be interpreted by consensus theory. This provides
a theoretical proof of its validity. In addition, the impact of
different communication structures on control performance
is discussed. The analysis proves that the highest rate of con-
vergence is achieved in the existing communication structure.
The current-tracking dynamics issue is also discussed, which
is found not to deteriorate the consensus-based interpretation.
Moreover, the proportional relationship between the rate of
convergence and the P controller parameter is justified, which
is helpful in the design of system control parameters for
researchers and power electronics engineers. Experiments are
conducted on a seven-level down-scaled MMC prototype,
verifying the analysis.
REFERENCES
[1] H. Akagi, ‘‘Classification, terminology, and application of the modular
multilevel cascade converter (MMCC),’’ IEEE Trans. Power Electron.,
vol. 26, no. 11, pp. 3119–3130, Nov. 2011.
[2] N. Flourentzou, V. G. Agelidis, and G. D. Demetriades, ‘‘VSC-based
HVDC power transmission systems: An overview,’’ IEEE Trans. Power
Electron., vol. 24, no. 3, pp. 592–602, Mar. 2009.
[3] M. Saeedifard and R. Iravani, ‘‘Dynamic performance of a modular multi-
level back-to-back HVDC system,’’ IEEE Trans. Power Del., vol. 25, no. 4,
pp. 2903–2912, Oct. 2010.
[4] S. Debnath, J. Qin, B. Bahrani, M. Saeedifard, and P. Barbosa, ‘‘Operation,
control, and applications of the modular multilevel converter: A review,’’
IEEE Trans. Power Electron., vol. 30, no. 1, pp. 37–53, Jan. 2015.
[5] A. J. Korn, M. Winkelnkemper, and P. Steimer, ‘‘Low output frequency
operation of the modular multi-level converter,’’ in Proc. IEEE Energy
Convers. Congr. Expo. (ECCE), Sep. 2010, pp. 3993–3997.
[6] M. Hagiwara and H. Akagi, ‘‘Control and experiment of pulsewidth-
modulated modular multilevel converters,’’ IEEE Trans. Power Electron.,
vol. 24, no. 7, pp. 1737–1746, Jul. 2009.
[7] A. Nami, L. Jiaqi, F. Dijkhuizen, and G. D. Demetriades, ‘‘Modu-
lar multilevel converters for HVDC applications: Review on converter
cells and functionalities,’’ IEEE Trans. Power Electron., vol. 30, no. 1,
pp. 18–36, Jan. 2015.
[8] M. A. Perez, S. Bernet, J. Rodriguez, S. Kouro, and R. Lizana, ‘‘Circuit
topologies, modeling, control schemes, and applications of modular mul-
tilevel converters,’’ IEEE Trans. Power Electron., vol. 30, no. 1, pp. 4–17,
Jan. 2015.
[9] N. Thitichaiworakorn, M. Hagiwara, and H. Akagi, ‘‘Experimental ver-
ification of a modular multilevel cascade inverter based on double-star
bridge-cells (MMCI-DSBC),’’ IEEE Trans. Ind. Appl., vol. 50, no. 1,
pp. 509–519, Jan./Feb. 2014.
[10] M. Guan, Z. Xu, and H. Chen, ‘‘Control and modulation strategies for
modular multilevel converter based HVDC system,’’ in Proc. 37th Annu.
Conf. IEEE Ind. Electron. Soc., Nov. 2011, pp. 849–854.
[11] P. M. Meshram and V. B. Borghate, ‘‘A simplified nearest level con-
trol (NLC) voltage balancing method for modular multilevel converter
(MMC),’’ IEEE Trans. Power Electron., vol. 30, no. 1, pp. 450–462,
Jan. 2015.
[12] F. Zhao, G. Xiao, M. Liu, and D. Yang, ‘‘A fast sorting strategy based
on a two-way merge sort for balancing the capacitor voltages in modular
multilevel converters,’’ J. Power Electron., vol. 17, no. 2, pp. 346–357,
Mar. 2017.
[13] Q. Tu, Z. Xu, and L. Xu, ‘‘Reduced switching-frequency modulation and
circulating current suppression for modular multilevel converters,’’ IEEE
Trans. Power Del., vol. 26, no. 3, pp. 2009–2017, Jul. 2011.
[14] J. Qin and M. Saeedifard, ‘‘Reduced switching-frequency voltage-
balancing strategies for modular multilevel HVDC converters,’’ IEEE
Trans. Power Del., vol. 28, no. 4, pp. 2403–2410, Oct. 2013.
[15] P. Hu and D. Jiang, ‘‘A level-increased nearest level modulation method
for modular multilevel converters,’’ IEEE Trans. Power Electron., vol. 30,
no. 4, pp. 1836–1842, Apr. 2015.
[16] E. Solas, G. Abad, J. A. Barrena, S. Aurtenetxea, A. Cárcar, and L. Zając,
‘‘Modular multilevel converter with different submodule concepts—Part I:
Capacitor voltage balancing method,’’ IEEE Trans. Ind. Electron., vol. 60,
no. 10, pp. 4525–4535, Oct. 2013.
66819
 S. Song, J. Liu: Interpreting the Individual Capacitor Voltage Regulation Control of PSC-PWM MMC via Consensus Theory
[17] E. Solas, G. Abad, J. A. Barrena, S. Aurtenetxea, A. Cárcar, and L. Zając,
‘‘Modular multilevel converter with different submodule concepts—Part
II: Experimental validation and comparison for HVDC application,’’ IEEE
Trans. Ind. Electron., vol. 60, no. 10, pp. 4536–4545, Oct. 2013.
[18] B. Li, R. Yang, D. Xu, G. Wang, W. Wang, and D. Xu, ‘‘Analy-
sis of the phase-shifted carrier modulation for modular multilevel con-
verters,’’ IEEE Trans. Power Electron., vol. 30, no. 1, pp. 297–310,
Jan. 2015.
[19] J.-Y. Choi and B.-M. Han, ‘‘An improved phase-shifted carrier PWM for
modular multilevel converters with redundancy sub-modules,’’ J. Power
Electron., vol. 16, no. 2, pp. 473–479, Mar. 2016.
[20] K. Li, L. Yuan, Z. Zhao, S. Lu, and Y. Zhang, ‘‘Fault-tolerant con-
trol of MMC with hot reserved submodules based on carrier phase
shift modulation,’’ IEEE Trans. Power Electron., vol. 32, no. 9,
pp. 6778–6791, Sep. 2017.
[21] N. A. Lynch, Distributed Algorithms. San Francisco, CA, USA:
Morgan Kaufmann, 1997.
[22] R. Olfati-Saber and R. M. Murray, ‘‘Consensus problems in networks of
agents with switching topology and time-delays,’’ IEEE Trans. Autom.
Control, vol. 49, no. 9, pp. 1520–1533, Sep. 2004.
[23] R. Olfati-Saber, J. A. Fax, and R. M. Murray, ‘‘Consensus and coop-
eration in networked multi-agent systems,’’ Proc. IEEE, vol. 95, no. 1,
pp. 215–233, Jan. 2007.
[24] J. Qin, Q. Ma, Y. Shi, and L. Wang, ‘‘Recent advances in consensus of
multi-agent systems: A brief survey,’’ IEEE Trans. Ind. Electron., vol. 64,
no. 6, pp. 4972–4983, Jun. 2017.
[25] J. Qin, Q. Ma, H. Gao, Y. Shi, and Y. Kang, ‘‘On group synchronization
for interacting clusters of heterogeneous systems,’’ IEEE Trans. Cybern.,
vol. 47, no. 12, pp. 4122–4133, Dec. 2017.
[26] F. Xiao, Y. Shi, and W. Ren, ‘‘Robustness analysis of asynchronous
sampled-data multiagent networks with time-varying delays,’’ IEEE Trans.
Autom. Control, vol. 63, no. 7, pp. 2145–2152, Jul. 2018.
[27] H. Li, Y. Shi, and W. Yan, ‘‘On neighbor information utilization in dis-
tributed receding horizon control for consensus-seeking,’’ IEEE Trans.
Cybern., vol. 46, no. 9, pp. 2019–2027, Sep. 2016.
[28] Z. Peng, D. Wang, H. Zhang, and G. Sun, ‘‘Distributed neural net-
work control for adaptive synchronization of uncertain dynamical mul-
tiagent systems,’’ IEEE Trans. Neural Netw. Learn. Syst., vol. 25, no. 8,
pp. 1508–1519, Aug. 2014.
[29] J. Yu and Y. Shi, ‘‘Scaled group consensus in multiagent systems with
first/second-order continuous dynamics,’’ IEEE Trans. Cybern., vol. 48,
no. 8, pp. 2259–2271, Aug. 2018.
[30] L. Y. Lu and C. C. Chu, ‘‘Consensus-based droop control synthesis for
multiple DICs in isolated micro-grids,’’ IEEE Trans. Power Syst., vol. 30,
no. 5, pp. 2243–2256, Sep. 2015.
[31] F. Guo, C. Wen, J. Mao, and Y.-D. Song, ‘‘Distributed secondary volt-
age and frequency restoration control of droop-controlled inverter-based
microgrids,’’ IEEE Trans. Ind. Electron., vol. 62, no. 7, pp. 4355–4364,
Jul. 2015.
[32] L.-Y. Lu and C.-C. Chu, ‘‘Consensus-based secondary frequency and
voltage droop control of virtual synchronous generators for isolated AC
micro-grids,’’ IEEE J. Emerg. Sel. Topics Circuits Syst., vol. 5, no. 3,
pp. 443–455, Sep. 2015.
[33] M. Fiedler, ‘‘Algebraic connectivity of graphs,’’ Czechoslovak Math. J.,
vol. 23, no. 2, pp. 298–305, 1973.
[34] R. O. Saber and R. M. Murray, ‘‘Consensus protocols for net-
works of dynamic agents,’’ in Proc. Amer. Control Conf., Jun. 2003,
pp. 951–956.
[35] M. Hagiwara, R. Maeda, and H. Akagi, ‘‘Control and analysis of the
modular multilevel cascade converter based on double-star chopper-
cells (MMCC-DSCC),’’ IEEE Trans. Power Electron., vol. 26, no. 6,
pp. 1649–1658, Jun. 2011.
[36] Z. Li, P. Wang, Z. Chu, H. Zhu, Y. Luo, and Y. Li, ‘‘An inner current sup-
pressing method for modular multilevel converters,’’ IEEE Trans. Power
Electron., vol. 28, no. 11, pp. 4873–4879, Nov. 2013.
[37] M. Zhang, L. Huang, W. Yao, and Z. Lu, ‘‘Circulating harmonic current
elimination of a CPS-PWM-based modular multilevel converter with a
plug-in repetitive controller,’’ IEEE Trans. Power Electron., vol. 29, no. 4,
pp. 2083–2097, Apr. 2014.
[38] L. He, K. Zhang, J. Xiong, and S. Fan, ‘‘A repetitive control scheme
for harmonic suppression of circulating current in modular multilevel
converters,’’ IEEE Trans. Power Electron., vol. 30, no. 1, pp. 471–481,
Jan. 2015.
66820
[39] S. Geng, Y. Gan, Y. Li, L. Hang, and G. Li, ‘‘Novel circulating cur-
rent suppression strategy for MMC based on quasi-PR controller,’’ in
Proc. IEEE Appl. Power Electron. Conf. Expo. (APEC), Mar. 2016,
pp. 3560–3565.
[40] D. Mosk-Aoyama, ‘‘Maximum algebraic connectivity augmentation is NP-
hard,’’ Oper. Res. Lett., vol. 36, no. 6, pp. 677–679, Nov. 2008.
SHUGUANG SONG received the B.S. degree
from Xi’an Jiaotong University, Xi’an, China,
in 2014, where he is currently pursuing the Ph.D.
degree, both in electrical engineering. His cur-
rent research interests include multilevel convert-
ers in power quality, HVDC, and motor drive
applications.
JINJUN LIU (M’97–SM’10–F’19) received the
B.S. and Ph.D. degrees in electrical engineering
from Xi’an Jiaotong University (XJTU), Xi’an,
China, in 1992 and 1997, respectively.
He then joined the XJTU Electrical Engineer-
ing School as a Faculty Member. From 1999 to
2002, he was with the Center for Power Electronics
Systems, Virginia Polytechnic Institute and State
University, Blacksburg, VA, USA, as a Visiting
Scholar. In 2002, he was promoted to the position
of Full Professor and then the Head of the Power Electronics and Renewable
Energy Center, XJTU, which now comprises 17 faculty members and over
100 graduate students and carries one of the leading power electronics
programs in China. From 2005 to 2010, he was the Associate Dean of
the Electrical Engineering School, XJTU, and the Dean for Undergraduate
Education of XJTU (2009–2015). He is currently a XJTU Distinguished
Professor of power electronics, sponsored by the Chang Jiang Scholars
Program of Chinese Ministry of Education. He has coauthored three books
(including one textbook), has published over 400 technical papers in peer-
reviewed journals and conference proceedings, holds nearly 50 invention
patents (China/US), and has delivered for many times plenary keynote
speeches and tutorials at the IEEE conferences or China national conferences
in power electronics area. His research interests include power quality control
and utility applications of power electronics, micro-grids for sustainable
energy and distributed generation, and more/all electronic power systems.
Dr. Liu has received for eight times governmental awards at the national
level or provincial/ministerial level for scientific research/teaching achieve-
ments. He also received the 2006 Delta Scholar Award, the 2014 Chang Jiang
Scholar Award, the 2014 Outstanding Sci-Tech Worker of the Nation Award,
and the IEEE TRANSACTIONS ON POWER ELECTRONICS 2016 Prize Paper Award.
He has served as the IEEE Power Electronics Society Region 10 Liaison
and then China Liaison for 10 years, an Associate Editor for the IEEE
TRANSACTIONS ON POWER ELECTRONICS, for 12 years, and has been the Vice
President for membership of the IEEE PELS, since 2015. He is on the Board
of China Electrotechnical Society and was elected the Vice President of the
CES Power Electronics Society, in 2013. Since 2013, he has been the Vice
President for International Affairs, China Power Supply Society (CPSS), and
has been the inaugural Editor-in-Chief of the CPSS Transactions on Power
Electronics and Applications, since 2016. Since 2013, he has been serving
as the Vice Chair of the Chinese National Steering Committee for College
Electric Power Engineering Programs.
VOLUME 7, 2019