You are on page 1of 16

1176 IEEE TRANSACTIONS ON POWER ELECTRONICS, VOL. 29, NO.

3, MARCH 2014
A Unified Control Strategy for Three-Phase Inverter
in Distributed Generation
Zeng Liu, Student Member, IEEE, Jinjun Liu, Senior Member, IEEE, and Yalin Zhao
Abstract—This paper presents a unified control strategy that en-
ables bothislandedandgrid-tiedoperations of three-phase inverter
in distributed generation, with no need for switching between two
corresponding controllers or critical islanding detection. The pro-
posed control strategy composes of an inner inductor current loop,
and a novel voltage loop in the synchronous reference frame. The
inverter is regulated as a current source just by the inner induc-
tor current loop in grid-tied operation, and the voltage controller
is automatically activated to regulate the load voltage upon the
occurrence of islanding. Furthermore, the waveforms of the grid
current in the grid-tied mode and the load voltage in the islanding
mode are distorted under nonlinear local load with the conven-
tional strategy. And this issue is addressed by proposing a unified
load current feedforward in this paper. Additionally, this paper
presents the detailed analysis and the parameter design of the
control strategy. Finally, the effectiveness of the proposed control
strategy is validated by the simulation and experimental results.
Index Terms—Distributed generation (DG), islanding, load cur-
rent, seamless transfer, three-phase inverter, unified control.
I. INTRODUCTION
D
ISTRIBUTED generation (DG) is emerging as a viable
alternative when renewable or nonconventional energy
resources are available, such as wind turbines, photovoltaic ar-
rays, fuel cells, microturbines [1], [3]. Most of these resources
are connected to the utility through power electronic interfacing
converters, i.e., three-phase inverter. Moreover, DG is a suitable
formto offer high reliable electrical power supply, as it is able to
operate either in the grid-tied mode or in the islanded mode [2].
In the grid-tied operation, DG deliveries power to the utility
and the local critical load. Upon the occurrence of utility outage,
the islanding is formed. Under this circumstance, the DG must
be tripped and cease to energize the portion of utility as soon as
possible according to IEEE Standard 929-2000 [4]. However,
in order to improve the power reliability of some local critical
Manuscript received December 15, 2012; revised March 4, 2013; accepted
April 21, 2013. Date of current version September 18, 2013. This work was
supported in part by the National Basic Research Program (973 Program) of
China under Project 2009CB219705, and by the State Key Laboratory of Elec-
trical Insulation and Power Equipment under Project EIPE09109. This paper
was presented in part at the 26th IEEE Applied Power Electronics Conference
and Exposition, Fort Worth, TX, USA, March 6–11, 2011. Recommended for
publication by Associate Editor D. Xu.
The authors are with the State Key Lab of Electrical Insulation and
Power Equipment, School of Electrical Engineering, Xi’an Jiaotong Univer-
sity, Xi’an 710049, China (e-mail: zeng.liu@ieee.org; jjliu@mail.xjtu.edu.cn;
yobdc54@gmail.com).
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.2013.2262078
load, the DG should disconnect to the utility and continue to
feed the local critical load [5]. The load voltage is key issue of
these two operation modes, because it is fixed by the utility in
the grid-tied operation, and formed by the DG in the islanded
mode, respectively. Therefore, upon the happening of islanding,
DG must take over the load voltage as soon as possible, in order
to reduce the transient in the load voltage. And this issue brings
a challenge for the operation of DG.
Droop-based control is used widely for the power sharing of
parallel inverters [11], [12], which is called as voltage mode
control in this paper, and it can also be applied to DG to real-
ize the power sharing between DG and utility in the grid-tied
mode [13]–[16], [53]. In this situation, the inverter is always
regulated as a voltage source by the voltage loop, and the qual-
ity of the load voltage can be guaranteed during the transition
of operation modes. However, the limitation of this approach is
that the dynamic performance is poor, because the bandwidth
of the external power loop, realizing droop control, is much
lower than the voltage loop. Moreover, the grid current is not
controlled directly, and the issue of the inrush grid current dur-
ing the transition from the islanded mode to the grid-tied mode
always exists, even though phase-locked loop (PLL) and the
virtual inductance are adopted [15].
The hybrid voltage and current mode control is a popular
alternative for DG, in which two distinct sets of controllers
are employed [17]–[40]. The inverter is controlled as a current
source by one sets of a controller in the grid-tied mode, while
as a voltage source by the other sets of controller in the islanded
mode. As the voltage loop or current loop is just utilized in
this approach, a nice dynamic performance can be achieved.
Besides, the output current is directly controlled in the grid-tied
mode, and the inrush grid current is almost eliminated.
In the hybrid voltage and current mode control, there is a
need to switch the controller when the operation mode of DG
is changed. During the interval from the occurrence of utility
outage and switching the controller to voltage mode, the load
voltage is neither fixed by the utility, nor regulated by the DG,
and the length of the time interval is determined by the islanding
detection process. Therefore, the main issue in this approach is
that it makes the quality of the load voltage heavily reliant on the
speed and accuracy of the islanding detection method [6]–[10].
Another issue associated with the aforementioned approaches
is the waveform quality of the grid current and the load voltage
under nonlinear local load. In the grid-tied mode, the output
current of DG is generally desired to be pure sinusoidal [18].
When the nonlinear local load is fed, the harmonic component
of the load current will fully flow into the utility. A single-phase
DG, which injects harmonic current into the utility for mitigating
0885-8993 © 2013 IEEE
LIU et al.: UNIFIED CONTROL STRATEGY FOR THREE-PHASE INVERTER IN DISTRIBUTED GENERATION 1177
Fig. 1. Schematic diagram of the DG based on the proposed control strategy.
the harmonic component of the grid current, is presented in [41].
The voltage mode control is enhanced by controlling the DG to
emulate a resistance at the harmonic frequency, and then the
harmonic current flowing into utility can be mitigated [42].
In the islanded mode, the nonlinear load may distort the load
voltage [43], and many control schemes have been proposed to
improve the quality of the load voltage, including a multiloop
control method [43]–[46], resonant controllers [48], [49], sliding
mode control [47]. However, existing control strategies, dealing
with the nonlinear local load in DG, mainly focus on either the
quality of the grid current in the grid-tied mode or the one of the
load voltage in the islanded mode, and improving both of them
by a unified control strategy is seldom.
This paper proposes a unified control strategy that avoids
the aforementioned shortcomings. First, the traditional induc-
tor current loop is employed to control the three-phase inverter
in DG to act as a current source with a given reference in the
synchronous reference frame (SRF). Second, a novel voltage
controller is presented to supply reference for the inner induc-
tor current loop, where a proportional-plus-integral (PI) com-
pensator and a proportional (P) compensator are employed in
D-axis and Q-axis, respectively. In the grid-tied operation, the
load voltage is dominated by the utility, and the voltage com-
pensator in D-axis is saturated, while the output of the voltage
compensator in Q-axis is forced to be zero by the PLL. There-
fore, the reference of the inner current loop cannot regulated by
the voltage loop, and the DG is controlled as a current source
just by the inner current loop. Upon the occurrence of the grid
outage, the load voltage is no more determined by the utility,
and the voltage controller is automatically activated to regulate
the load voltage. These happen naturally, and, thus the proposed
control strategy does not need a forced switching between two
distinct sets of controllers. Further, there is no need to detect
the islanding quickly and accurately, and the islanding detec-
tion method is no more critical in this approach. Moreover,
the proposed control strategy, benefiting from just utilizing the
current and voltage feedback control, endows a better dynamic
performance, compared to the voltage mode control.
Third, the proposed control strategy is enhanced by introduc-
ing a unified load current feedforward, in order to deal with
the issue caused by the nonlinear local load, and this scheme
is implemented by adding the load current into the reference
of the inner current loop. In the grid-tied mode, the DG injects
harmonic current into the grid for compensating the harmonic
component of the grid current, and thus, the harmonic compo-
nent of the grid current will be mitigated. Moreover, the benefit
of the proposed load current feedforward can be extended into
the islanded operation mode, due to the improved quality of the
load voltage.
The rest of this paper is arranged as follows. Section II de-
scribes the proposed unified control strategy for three-phase
inverter in DG, including the power stage of DG, the basic idea,
and the control diagram. The detailed operation principle of DG
with the proposed control strategy is illustrated in Section III.
The parameter design and small signal analysis of the proposed
control system are given in Section IV. Section V investigates
the proposed control strategy by simulation and experimental
results. Finally, the concluding remarks are given in Section VI.
II. PROPOSED CONTROL STRATEGY
A. Power Stage
This paper presents a unified control strategy for a three-
phase inverter in DG to operate in both islanded and grid-tied
modes. The schematic diagramof the DGbased on the proposed
control strategy is shown by Fig. 1. The DG is equipped with
a three-phase interface inverter terminated with a LC filter. The
primary energy is converted to the electrical energy, which is
then converted to dc by the front-end power converter, and the
output dc voltage is regulated by it. Therefore, they can be
represented by the dc voltage source V
dc
in Fig. 1. In the ac side
of inverter, the local critical load is connected directly.
It should be noted that there are two switches, denoted by S
u
and S
i
, respectively, in Fig. 1, and their functions are different.
The inverter transfer switch S
i
is controlled by the DG, and the
utility protection switch S
u
is governed by the utility. When the
utility is normal, both switches S
i
and S
u
are ON, and the DG in
the grid-tied mode injects power to the utility. When the utility is
in fault, the switch S
u
is tripped by the utility instantly, and then
the islanding is formed. After the islanding has been confirmed
by the DG with the islanding detection scheme [6]–[10], the
switch S
i
is disconnected, and the DG is transferred from the
grid-tied mode to the islanded mode. When the utility is restored,
the DG should be resynchronized with the utility first, and then
the switch S
i
is turned ON to connect the DG with the grid.
1178 IEEE TRANSACTIONS ON POWER ELECTRONICS, VOL. 29, NO. 3, MARCH 2014
Fig. 2. Overall block diagram of the proposed unified control strategy.
B. Basic Idea
With the hybrid voltage and current mode control [17]–[40],
the inverter is controlled as a current source to generate the
reference power P
DG
+ jQ
DG
in the grid-tied mode. And its
output power P
DG
+ jQ
DG
should be the sum of the power
injected to the grid P
g
+ jQ
g
and the load demand P
load
+
jQ
load
, which can be expressed as follows by assuming that the
load is represented as a parallel RLC circuit:
P
load
=
3
2
·
V
2
m
R
(1)
Q
load
=
3
2
· V
2
m
·
_
1
ωL
− ωC
_
. (2)
In (1) and (2), V
m
and ω represent the amplitude and fre-
quency of the load voltage, respectively. When the nonlinear
local load is fed, it can still be equivalent to the parallel RLC
circuit by just taking account of the fundamental component.
During the time interval fromthe instant of islanding happen-
ing to the moment of switching the control system to voltage
mode control, the load voltage is neither fixed by the utility nor
regulated by the inverter, so the load voltage may drift from
the normal range [6]. And this phenomenon can be explained
as below by the power relationship. During this time interval,
the inverter is still controlled as a current source, and its output
power is kept almost unchanged. However, the power injected to
utility decreases to zero rapidly, and then the power consumed
by the load will be imposed to the output power of DG. If both
active power P
g
and reactive power Q
g
injected into the grid are
positive in the grid-tied mode, then P
load
and Q
load
will increase
after the islanding happens, and the amplitude and frequency of
the load voltage will rise and drop, respectively, according to
(1) and (2).
With the previous analysis, if the output power of inverter
P
DG
+ jQ
DG
could be regulated to match the load demand by
changing the current reference before the islanding is confirmed,
the load voltage excursion will be mitigated. And this basic idea
is utilized in this paper. In the proposed control strategy, the
output power of the inverter is always controlled by regulating
the three-phase inductor current i
Labc
, while the magnitude and
frequency of the load voltage v
Cabc
are monitored. When the
islanding happens, the magnitude and frequency of the load volt-
age may drift fromthe normal range, and then they are controlled
to recover to the normal range automatically by regulating the
output power of the inverter.
C. Control Scheme
Fig. 2 describes the overall block diagram for the proposed
unified control strategy, where the inductor current i
Labc
, the
utility voltage v
gabc
, the load voltage v
Cabc
, and the load current
i
LLabc
are sensed. And the three-phase inverter is controlled in
the SRF, in which, three phase variable will be represented
by dc quantity. The control diagram is mainly composed by
the inductor current loop, the PLL, and the current reference
generation module.
In the inductor current loop, the PI compensator is employed
in both D- and Q-axes, and a decoupling of the cross coupling
denoted by ω
0
L
f
/k
PWM
is implemented in order to mitigate the
couplings due to the inductor. The output of the inner current
loop d
dq

, together with the decoupling of the capacitor voltage
denoted by 1/k
PWM
, sets the reference for the standard space
vector modulation that controls the switches of the three-phase
inverter. It should be noted that k
PWM
denotes the voltage gain
of the inverter, which equals to half of the dc voltage in this
paper.
LIU et al.: UNIFIED CONTROL STRATEGY FOR THREE-PHASE INVERTER IN DISTRIBUTED GENERATION 1179
Fig. 3. Block diagram of the current reference generation module.
The PLL in the proposed control strategy is based on the
SRF PLL [50], [51], which is widely used in the three-phase
power converter to estimate the utility frequency and phase.
Furthermore, a limiter is inserted between the PI compensator
G
PLL
and the integrator, in order to hold the frequency of the
load voltage within the normal range in the islanded operation.
In Fig. 2, it can be found that the inductor current is regulated
to follow the current reference i
Lref dq
, and the phase of the
current is synchronized to the grid voltage v
gabc
. If the current
reference is constant, the inverter is just controlled to be a current
source, which is the same with the traditional grid-tied inverter.
The new part in this paper is the current reference generation
module shown in Fig. 2, which regulates the current reference to
guarantee the power match between the DG and the local load
and enables the DG to operate in the islanded mode. Moreover,
the unified load current feedforward, to deal with the nonlinear
local load, is also implemented in this module.
The block diagram of the proposed current reference gen-
eration module is shown in Fig. 3, which provides the current
reference for the inner current loop in both grid-tied and islanded
modes. In this module, it can be found that an unsymmetrical
structure is used in D- and Q-axes. The PI compensator is
adopted in D-axes, while the P compensator is employed in
Q-axis. Besides, an extra limiter is added in the D-axis. More-
over, the load current feedforward is implemented by adding the
load current i
LLdq
to the final inductor current reference i
Lref dq
.
The benefit brought by the unique structure in Fig. 3 can be rep-
resented by two parts: 1) seamless transfer capability without
critical islanding detection; and 2) power quality improvement
in both grid-tied and islanded operations. The current reference
i
Lredq
composes of four parts in D- and Q-axes respectively:
1) the output of voltage controller i
ref dq
; 2) the grid current
reference I
gref dq
; 3) the load current i
LLdq
; and 4) the current
flowing through the filter capacitor C
f
.
In the grid-tied mode, the load voltage v
Cdq
is clamped by the
utility. The current reference is irrelevant to the load voltage, due
to the saturation of the PI compensator in D-axis, and the output
of the P compensator being zero in Q-axis, and thus, the inverter
operates as a current source. Upon occurrence of islanding, the
voltage controller takes over automatically to control the load
voltage by regulating the current reference, and the inverter acts
as a voltage source to supply stable voltage to the local load;
this relieves the need for switching between different control
architectures.
Another distinguished function of the current reference gen-
eration module is the load current feedforward. The sensed
load current is added as a part of the inductor current refer-
ence i
Lref dq
to compensate the harmonic component in the grid
current under nonlinear local load. In the islanded mode, the load
current feedforward operates still, and the disturbance from the
load current, caused by the nonlinear load, can be suppressed
by the fast inner inductor current loop, and thus, the quality of
the load voltage is improved.
The inductor current control in Fig. 2 was proposed in pre-
vious publications for grid-tied operation of DG [18], and the
motivation of this paper is to propose a unified control strategy
for DG in both grid-tied and islanded modes, which is repre-
sented by the current reference generation module in Fig. 3.
The contribution of this module can be summarized in two as-
pects. First, by introducing PI compensator and P compensator
in D-axis and Q-axis respectively, the voltage controller is inac-
tivated in the grid-tied mode and can be automatically activated
upon occurrence of islanding. Therefore, there is no need for
switching different controllers or critical islanding detection,
and the quality of the load voltage during the transition from
the grid-tied mode to the islanded mode can be improved. The
second contribution of this module is to present the load current
feedforward to deal with the issue caused by the nonlinear local
load, with which, not only the waveform of the grid current in
grid-tied is improved, but also the quality of the load voltage in
the islanded mode is enhanced.
Besides, it should be noted that a three-phase unbalanced
local load cannot be fed by the DG with the proposed control
strategy, because there is no flow path for the zero sequence
current of the unbalanced load, and the regulation of the zero
sequence current is beyond the scope of the proposed control
strategy.
III. OPERATION PRINCIPLE OF DG
The operation principle of DG with the proposed unified
control strategy will be illustrated in detail in this section, and
there are in total four states for the DG, including the grid-tied
mode, transition from the grid-tied mode to the islanded mode,
the islanded mode, and transition from the islanded mode to the
grid-tied mode.
A. Grid-Tied Mode
When the utility is normal, the DG is controlled as a current
source to supply given active and reactive power by the inductor
current loop, and the active and reactive power can be given
by the current reference of D- and Q-axis independently. First,
the phase angle of the utility voltage is obtained by the PLL,
which consists of a Park transformation expressed by (3), a PI
1180 IEEE TRANSACTIONS ON POWER ELECTRONICS, VOL. 29, NO. 3, MARCH 2014
compensator, a limiter, and an integrator
_
x
d
x
q
_
=
2
3
_
_
_
_
_
cos θ cos
_
θ −
2
3
π
_
cos
_
θ +
2
3
π
_
−sin θ −sin
_
θ −
2
3
π
_
−sin
_
θ +
2
3
π
_
_
_
_
_
_
×
_
_
x
a
x
b
x
c
_
_
. (3)
Second, the filter inductor current, which has been trans-
formed into SRF by the Park transformation, is fed back and
compared with the inductor current reference i
Lref dq
, and the
inductor current is regulated to track the reference i
Lref dq
by
the PI compensator G
I
.
The reference of the inductor current loop i
Lref dq
seems
complex and it is explained as below. It is assumed that the utility
is stiff, and the three-phase utility voltage can be expressed as
_
¸
¸
¸
¸
¸
¸
_
¸
¸
¸
¸
¸
¸
_
v
ga
= V
g
cos θ

v
gb
= V
g
cos
_
θ



3
_
v
gc
= V
g
cos
_
θ

+

3
_
(4)
where V
g
is the magnitude of the grid voltage, and θ

is the actual
phase angle. By the Park transformation, the utility voltage is
transformed into the SRF, which is shown as
_
v
gd
= V
g
cos(θ

− θ)
v
gq
= V
g
sin(θ

− θ).
(5)
v
gq
is regulated to zero by the PLL, so v
gd
equals the mag-
nitude of the utility voltage V
g
. As the filter capacitor voltage
equals the utility voltage in the gird-tied mode, v
Cd
equals the
magnitude of the utility voltage V
g
, and v
Cq
equals zero, too.
In the D-axis, the inductor current reference i
Lref d
can be
expressed by (6) according to Fig. 3
i
Lref d
= I
gref d
+ i
LLd
− ω
0
C
f
· v
Cq
. (6)
The first part is the output of the limiter. It is assumed that
the given voltage reference V
max
is larger than the magnitude
of the utility voltage v
Cd
in steady state, so the PI compensator,
denoted by G
V D
in the following part, will saturate, and the
limiter outputs its upper value I
gref d
. The second part is the
load current of D-axis i
LLd
, which is determined by the charac-
teristic of the local load. The third part is the proportional part
−ω
0
C
f
· v
Cq
, where ω
0
is the rated angle frequency, and C
f
is the capacitance of the filter capacitor. It is fixed as v
Cq
de-
pends on the utility voltage. Consequently, the current reference
i
Lref d
is imposed by the given reference I
gref d
and the load
current i
LLd
, and is independent of the load voltage.
In the Q-axis, the inductor current reference i
Lref q
consists
of four parts as
i
Lref q
= v
Cq
· k
Gvq
+ I
gref q
+ i
LLq
+ ω
0
C
f
· v
Cd
(7)
where k
Gvq
is the parameter of the P compensator, denoted by
G
V Q
in the following part. The first part is the output of G
V Q
,
Fig. 4. Simplified block diagram of the unified control strategy when DG
operates in the grid-tied mode.
which is zero as the v
Cq
has been regulated to zero by the PLL.
The second part is the given current reference I
gref q
, and the
third part represents the load current in Q-axis. The final part
is the proportional part −ω
0
C
f
· v
Cd
, which is fixed since v
Cd
depends on the utility voltage. Therefore, the current reference
i
Lref q
cannot be influenced by the external voltage loop and is
determined by the given reference I
gref q
and the load current
i
LLq
.
With the previous analysis, the control diagramof the inverter
can be simplified as Fig. 4 in the grid-tied mode, and the inverter
is controlled as a current source by the inductor current loop
with the inductor current reference being determined by the
current reference I
gref dq
and the load current i
LLdq
. In other
words, the inductor current tracks the current reference and the
load current. If the steady state error is zero, I
gref dq
represents
the grid current actually, and this will be analyzed in the next
section.
B. Transition From the Grid-Tied Mode to the Islanded Mode
When the utility switch S
u
opens, the islanding happens, and
the amplitude and frequency of the load voltage will drift due
to the active and reactive power mismatch between the DG and
the load demand. The transition, shown in Fig. 5, can be divided
into two time interval. The first time intervals is from the instant
of turning off S
u
to the instant of turning off S
i
when islanding
is confirmed. The second time interval begins from the instant
of turning off inverter switch S
i
.
During the first time interval, the utility voltage v
gabc
is still
the same with the load voltage v
Cabc
as the switch S
i
is in ON
state. As the dynamic of the inductor current loop and the voltage
loop is much faster than the PLL [52], while the load voltage
and current are varying dramatically, the angle frequency of the
LIU et al.: UNIFIED CONTROL STRATEGY FOR THREE-PHASE INVERTER IN DISTRIBUTED GENERATION 1181
Fig. 5. Operation sequence during the transition from the grid-tied mode to
the islanded mode.
Fig. 6. Transient process of the voltage and current when the islanding
happens.
load voltage can be considered to be not varied. The dynamic
process in this time interval can be described by Fig. 6, and it is
illustrated later.
In the grid-tied mode, it is assumed that the DG injects ac-
tive and reactive power into the utility, which can be expressed
by (8) and (9), and that the local critical load, shown in (10),
represented by a series connected RLC circuit with the lagging
power factor
P
g
=
3
2
· (v
Cd
i
gd
+ v
Cq
i
gq
) =
3
2
v
Cd
i
gd
(8)
Q
g
=
3
2
· (v
Cq
i
gd
− v
Cd
i
gq
) = −
3
2
v
Cd
i
gq
(9)
Z
sload
= R
s
+ jωL
s
+
1
jωC
s
= R
s
+ j
_
ωL
s

1
ωC
s
_
= R
s
+ jX
s
. (10)
When islanding happens, i
gd
will decrease from positive to
zero, and i
gq
will increase from negative to zero. At the same
time, the load current will vary in the opposite direction. The
load voltage in D- and Q-axes is shown by (11) and (12), and
each of them consists of two terms. It can be found that the
load voltage in D-axis v
Cd
will increase as both terms increase.
However, the trend of the load voltage in Q-axis v
Cq
is uncertain
because the first term decreases and the second term increases,
and it is not concerned for a while
v
Cd
= i
LLd
· R
s
− i
LLq
· X
s
(11)
v
Cq
= i
LLq
· R
s
+ i
LLd
· X
s
. (12)
With the increase of the load voltage in D-axis v
Cd
, when
it reaches and exceeds V
max
, the input of the PI compensator
G
V D
will become negative, so its output will decrease. Then,
the output of limiter will not imposed to I
gref d
any longer, and
the current reference i
Lref d
will drop. With the regulation of
the inductor current loop, the load current in D-axis i
LLd
will
decrease. As a result, the load voltage in D-axis v
Cd
will drop
and recover to V
max
. After i
LLd
has almost fallen to the normal
value, the load voltage in Q-axis v
Cq
will drop according to
(12). As v
Cq
is decreased from zero to negative, then the input
of the PI compensator G
PLL
will be negative, and its output will
drop. In other words, the angle frequency ω will be reduced.
If it falls to the lower value of the limiter ω
min
, then the angle
frequency will be fixed at ω
min
.
Consequently, at the end of the first time interval, the load
voltage in D-axis v
Cd
will be increased to and fixed at V
max
,
and the angle frequency of the load voltage ω will drop. If it is
higher than the lower value of the limiter ω
min
, the PLL can still
operate normally, and the load voltage in Q-axis v
Cq
will be
zero. Otherwise, if it is fixed at ω
min
, the load voltage in Q-axis
v
Cq
will be negative. As the absolute values of v
Cd
and v
Cq
,
at least the one of v
Cd
, are raised, the magnitude of the load
voltage will increase finally.
The variation of the amplitude and frequency in the load volt-
age can also be explained by the power relationship mentioned
before. When the islanding happens, the local load must ab-
sorb the extra power injected to the grid, as the output power
of inverter is not changed instantaneously. According to (1), the
magnitude of the load voltage V
m
will rise with the increase of
P
load
. At the same time, the angle frequency ω should drop, in
order to consume more reactive power with (2). Therefore, the
result through the power relationship coincides with the previ-
ous analysis.
The second time interval of the transition begins from the
instant when the switch S
i
is open after the islanding has been
confirmed by the islanding detection method. If the switch S
i
opens, the load voltage v
Cabc
is independent with the grid volt-
age v
gabc
. At the same time, v
gabc
will reduce to zero theo-
retically as the switch S
u
has opened. Then, the input of the
compensator G
PLL
becomes zero and the angle frequency is
invariable and fixed to the value at the end of the first interval.
Under this circumstance, v
Cdq
is regulated by the voltage loop,
and the inverter is controlled to be a voltage source.
With the previous analysis, it can be concluded that the drift
of the amplitude and frequency in the load voltage is restricted
in the given range when islanding happens. And the inverter
is transferred from the current source operation mode to the
voltage source operation mode autonomously. In the hybrid
voltage and current mode control [17]–[40], the time delay of
islanding detection is critical to the drift of the frequency and
magnitude in the load voltage, because the drift is worse with
the increase of the delay time. However, this phenomenon is
avoided in the proposed control strategy.
C. Islanded Mode
In the islanded mode, switching S
i
and S
u
are both in OFF
state. The PLL cannot track the utility voltage normally, and the
angle frequency is fixed. In this situation, the DGis controlled as
a voltage source, because voltage compensator G
V D
and G
V Q
can regulate the load voltage v
Cdq
. The voltage references in D-
and Q-axis are V
max
and zero, respectively. And the magnitude
of the load voltage equals to V
max
approximately, which will
1182 IEEE TRANSACTIONS ON POWER ELECTRONICS, VOL. 29, NO. 3, MARCH 2014
Fig. 7. Simplified block diagram of the unified control strategy when DG
operates in the islanded mode.
be analyzed in Section IV. Consequently, the control diagram of
the three-phase inverter in the islanded mode can be simplified
as shown in Fig. 7.
In Fig. 7, the load current i
LLdq
is partial reference of the
inductor current loop. So, if there is disturbance in the load
current, it will be suppressed quickly by the inductor current
loop, and a stiff load voltage can be achieved.
D. Transition From the Islanded Mode to the Grid-Tied Mode
If the utility is restored and the utility switch S
u
is ON, the
DG should be connected with utility by turning on switch S
i
.
However, several preparation steps should be performed before
turning on switch S
i
.
First, as soon as utility voltage is restored, the PLL will track
the phase of the utility voltage. As a result, the phase angle of
the load voltage v
Cabc
will follow the grid voltage v
gabc
. If the
load voltage v
Cabc
is in phase with the utility voltage, v
gd
will
equal the magnitude of the utility voltage according to (5).
Second, as the magnitude of the load voltage V
max
is larger
than the utility voltage magnitude V
g
, the voltage reference V
ref
will be changed to V
g
by toggling the selector S fromterminals 1
to 2. As a result, the load voltage will equal to the utility voltage
in both phase and magnitude.
Third, the switch S
i
is turned on, and the selector S is reset
to terminal 1. In this situation, the load voltage will be held by
the utility. As the voltage reference V
ref
equals V
max
, which
is larger than the magnitude of the utility voltage V
g
, so the
PI compensator G
V D
will saturate, and the limiter outputs its
upper value I
gref d
. At the same time, v
Cq
is regulated to zero
by the PLL according to (5), so the output of G
V Q
will be
zero. Consequently, the voltage regulators G
V D
and G
V Q
are
inactivated, and the DG is controlled as a current source just by
the inductor current loop.
IV. ANALYSIS AND DESIGN
In this section, the three-phase inverter with the proposed
control strategy is analyzed and designed in both steady state
and transient state. In the steady state, the operation points of
DG in both grid-tied and islanded modes are analyzed, and
the limiters and references are selected. In the transient state,
compensators in both inductor current loop and the external
voltage loop are designed based on the small-signal model, and
the impact of the load current feedforward is analyzed as well.
A. Steady State
1) Analysis of Operation Points: As analyzed previously, in
the grid-tied mode, the inverter is controlled as a current source,
and the current reference for the inductor current loop i
Lref dq
is
expressed by (6) and (7). The steady-state error will be zero with
the PI compensator in the inductor current loop, so the inductor
current in steady state can be expressed as follows:
_
i
Ld
= I
gref d
− ω
0
C
f
· v
Cq
+ i
LLd
i
Lq
= v
Cq
· k
Gvq
+ ω
0
C
f
· v
Cd
+ I
gref q
+ i
LLq
.
(13)
In the SRF, the relationship between the voltage and the cur-
rent of the filter capacitor in steady state can be expressed by
_
i
Cd
= −v
Cq
· ωC
f
i
Cq
= v
Cd
· ωC
f
(14)
where ω represents the angle frequency of the DG and C
f
denotes capacitance of the filter capacitor. As a result, the output
current of the inverter i
odq
can be gained
_
¸
_
¸
_
i
od
= i
Ld
− i
Cd
= I
gref d
− (ω
0
− ω) · C
f
· v
Cq
+ i
LLd
i
oq
= i
Lq
− i
Cq
= v
Cq
· k
Gvq
+ I
gref q
+(ω
0
− ω) · C
f
· v
Cq
+ i
LLq
.
(15)
As angle frequency ω is very close to the rated angle fre-
quency ω
0
, the output current i
odq
can be simplified as
_
i
od
= I
gref d
+ i
LLd
i
oq
= v
Cq
· k
Gvq
+ I
gref q
+ i
LLq
.
(16)
It can be found that the output current follows I
gref dq
and the
load current i
LLdq
, as v
Cq
equals zero in the grid-tied mode. The
active and reactive power injected into utility can be obtained
as follows. Consequently, the active power and reactive power
flowing from the inverter to utility can be given by I
gref d
and
I
gref q
, respectively
_
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
_
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
_
P
g
=
3
2
· [v
Cd
(i
od
− i
LLd
) + v
Cq
(i
oq
− i
LLq
)]
=
3
2
· v
Cd
I
gref d
Q
g
=
3
2
· [v
Cq
(i
od
− i
LLd
) − v
cCd
(i
oq
− i
LLq
)]
= −
3
2
· v
Cd
I
gref q
.
(17)
In the islanded mode, the inverter is controlled as a volt-
age source by the external voltage loop. In the D-axis, v
Cd
is
LIU et al.: UNIFIED CONTROL STRATEGY FOR THREE-PHASE INVERTER IN DISTRIBUTED GENERATION 1183
regulated by the PI compensator G
V D
, so the steady state error
will be zero and v
Cd
can be expressed as follows:
v
Cd
= V
ref
(18)
where V
ref
is the voltage reference in D-axis.
In the Q-axis, the regulator G
V Q
is P compensator, so the
steady state error may not be zero. As the load current is added
to the inductor reference, the condition shown as below can be
achieved
v
Cq
· k
Gvq
+ I
gref q
= 0. (19)
And then, the load voltage in Q-axis can be expressed by
(20). It should be noted that the absolute value of v
Cq
rises with
the increase of the current reference I
gref q
which is related to
the reactive power injected into the utility
v
Cq
= −
I
gref q
k
Gvq
. (20)
The magnitude of the load voltage V
m
can be represented as
follows. It equals to V
ref
approximately, because v
Cq
should be
much lower than V
ref
with proper current reference I
gref q
V
m
=
¸
V
2
ref
+
_

I
gref q
k
Gvq
_
2
≈ V
ref
. (21)
When the islanding happens, the angle frequency is restricted
in the given range by the limiter. As analyzed previously, the an-
gle frequency in the islanded mode is determined in the first time
interval of the transition from the grid-tied made to the islanded
mode. According to (20), if the current reference I
gref q
is set
to zero, v
Cq
is zero. Then, it means that the angle frequency
ω does not vary in the first time interval of the transition, and
it should equal ω
g 0
, which denotes the angle frequency of the
utility before islanding happens. Consequently, the angle fre-
quency of the load voltage ω in the islanded mode is determined
by the current reference I
gref q
, and it can be expressed by (22),
where ω
min
and ω
max
represent the upper and lower values of
the limiter shown in Fig. 2, respectively
ω =
_
¸
_
¸
_
ω
min
, I
gref q
> 0
ω
g0
, I
gref q
= 0
ω
max
, I
gref q
< 0.
(22)
2) Selection of References and Limiters: In the grid-tied
mode, the active power injected into the grid P
g
is given by
the current reference I
gref d
, and it is the upper value of the
limiter in D-axis. Therefore, the selection of I
gref d
depends on
the power rating of the inverter. For the current reference I
gref q
,
firstly it determines the reactive power injected into utility Q
g
in
the grid-tied mode according to (17), and then it also affects the
magnitude of the load voltage in the islanded mode according
to (21). As a result, the reactive power Q
g
cannot be very large,
in order to make the magnitude of the load voltage within the
normal range in the islanded mode.
In the grid-tied mode, V
max
should be larger than the magni-
tude of the utility voltage V
g
. At the same time, the load voltage
is determined by V
max
in the islanded mode by (21), so V
max
should not be much larger than V
g
. Therefore, it is selected as
the maximum magnitude of the utility voltage in this paper.
According to IEEE standard 1547-2003 [5], the range of the
normal grid voltage is 0.88–1.1 p.u., so V
max
can be selected as
V
max
= 1.1 ·

2 · V
n
(23)
where V
n
represents the RMS value of the rated phase voltage.
In order to guarantee that the PLL operates normally in the
grid-tied mode, the utility angle frequency ω should not touch
the upper value ω
max
or lower value ω
min
of the limier in the
PLL. Besides, the angle frequency ω is restricted between ω
max
and ω
min
in the islanded mode, and it should not drift from
the normal value too far. So, ω
max
and ω
min
are selected as
the maximum and minimum angle frequencies allowed by the
utility standard.
B. Transient State
1) Small-Signal Model of the Power Stage: Before the com-
pensators in the voltage and current loops are designed and the
transient performance is analyzed, the three-phase inverter in the
DG needs to be modeled. According to the power stage shown
in Fig. 1, the dc-link voltage V
dc
is regulated by the front-end
converter in DG. Then, it is assumed that the dc voltage V
dc
is
very stiff, and its dynamic is not concerned in this paper. There-
fore, the average model of the power stage can be described
by
V
dc
2
·
_
_
_
d
a
d
b
d
c
_
_
_ = L
f
·
d
dt
_
_
_
i
La
i
Lb
i
Lc
_
_
_+ R
l
·
_
_
_
i
La
i
Lb
i
Lc
_
_
_+
_
_
_
v
Ca
v
Cb
v
Cc
_
_
_
(24)
_
_
_
i
La
i
Lb
i
Lc
_
_
_ = C
f
·
d
dt
_
_
_
v
Ca
v
Cb
v
Cc
_
_
_+
_
_
_
i
LLa
i
LLb
i
LLc
_
_
_+
_
_
_
i
ga
i
gb
i
gc
_
_
_. (25)
In (24), d
a
, d
b
, and d
c
are the average duty cycle of each leg
varying from −1 to 1, and R
l
represents the equivalent series
resistance of the filter inductor. Then, the average model in the
SRF can obtained with the Park transformation shown in (3),
which is represented by
V
dc
2
·
_
d
d
d
q
_
= L
f
·
d
dt
_
i
Ld
i
Lq
_
+
_
0 −ωL
f
ωL
f
0
_
·
_
i
Ld
i
Lq
_
+ R
l
·
_
i
Ld
i
Lq
_
+
_
v
Cd
v
Cq
_
(26)
_
i
Ld
i
Lq
_
= C
f
·
d
dt
_
v
Cd
v
Cq
_
+
_
0 −ωC
f
ωC
f
0
_
·
_
v
Cd
v
Cq
_
+
_
i
LLd
i
LLq
_
+
_
i
gd
i
gq
_
. (27)
With the stiff dc voltage V
dc
, the small-signal model will be
the same as the average model. Then, it can be found that there
are couplings introduced by the inductor L
f
and capacitors C
f
between D and Q-axes in the SRF, and these couplings can
be mitigated by the decoupling components ω
0
L
f
/k
PWM
and
1184 IEEE TRANSACTIONS ON POWER ELECTRONICS, VOL. 29, NO. 3, MARCH 2014
TABLE I
PARAMETERS OF THE POWER STAGE
Fig. 8. Bode plot of the control-to-current transfer function in grid-tied and
islanded modes.
ω
0
C
f
in Fig. 3 [12]. Therefore, the small-signal model can be
simplified into two identical SISOsystems, which is represented
by (28) and the subscript d and q are ignored
_
¸
¸
_
¸
¸
_
V
dc
2
·
ˆ
d = L
f
·
d
dt
ˆ
i
L
+ R
l
·
ˆ
i
L
+ ˆ v
C
ˆ
i
L
= C
f
·
d
dt
ˆ v
C
+
ˆ
i
LL
+
ˆ
i
g
.
(28)
2) Design and Analysis of the Current Loop: The inductor
current loop should operate normally to regulate the inductor
current loop in both islanded and grid-tied modes. In the islanded
mode, the small-signal model of the control-to-current can be
obtained according to (28), which is shown as
G
id1
(s) =
ˆ
i
L
(s)
ˆ
d(s)
=
V
dc
2
·
sC
f
s
2
L
f
C
f
+ sR
l
C
f
+ 1
. (29)
However, in the grid-tied mode, the dynamic of the capacitor
C
f
is ignored due to the stiff utility [34], and the small-signal
model of the control-to-current is described by
G
id2
(s) =
ˆ
i
L
(s)
ˆ
d(s)
=
V
dc
2
·
1
sL
f
+ R
l
. (30)
The parameters of the power stage in this paper are shown
in Table I, and the Bode plot of the control-to-current transfer
function in both of operation modes is shown in Fig. 8. It can
be found that huge difference appears in the low and medium
Fig. 9. Bode plot of the loop gain of the inner current loop.
Fig. 10. Block diagram of the simplified voltage loop.
frequency range, and it is difficult to design the compensator G
I
to achieve good performance in both of operation modes.
The reason for this difference between the islanded mode
and the grid-tied mode is that the inductor current is coupled
with the capacitor voltage in the islanded mode. To mitigate
this difference, the capacitor voltage is fed forward with the
coefficient 1/k
PWM
in Fig. 2, and then the inductor current
can be decoupled with the capacitor voltage. Consequently, the
transfer function of control to current in the islanded mode
is changed to be close to the one in the grid-tied mode, and
the current compensator G
I
can be designed based on unified
transfer function shown by (30).
The PI compensator G
I
in (31) is designed, and the digital
delay caused by the pulse width modulation (PWM) and sample
is considered as well. The loop gain of the current loop is shown
in Fig. 9, with the crossover frequency of 1100 Hz, and the phase
margin of 65

G
i
(s) = k
Gi
·
1 +
s
ω
G i
s
. (31)
3) Design and Analysis of the Voltage Loop: The voltage
loop just operates in the islanded mode to regulate the load
voltage, and the simplified block diagram is shown in Fig. 10,
where G
ic
(s) and G
vi
(s) denote the closed-loop transfer function
of an inductor loop and the impedance of the filter capacitor C
f
,
respectively.
In the D-axis, G
V D
is a PI compensator shown in (32), while
a P compensator G
V Q
expressed by (33) is used in Q-axis.
These two compensators are designed, and the loop gain of the
current loop is shown in Fig. 11. It can be found that there is a
little difference in the low frequency range. The phase margin
is set to 55

, and the crossover frequency is around 600 Hz in
LIU et al.: UNIFIED CONTROL STRATEGY FOR THREE-PHASE INVERTER IN DISTRIBUTED GENERATION 1185
Fig. 11. Bode plot of the loop gain of the voltage loop in D- and Q-axes.
both D- and Q-axes
G
vd
(s) = k
Gvd
·
1 +
s
ω
G v d
s
(32)
G
vq
(s) = k
Gvq
. (33)
4) Impact of Load Current Feedforward: In Fig. 10, the load
current
ˆ
i
LL
is a part of the inductor current reference, and the
disturbance from the load current can be suppressed by the
inductor current loop directly. To evaluate the effect of the load
current feedforward in the islanded mode, the transfer function
of the output impedance is derived. The output impedances with
and without load current feedforward are expressed by
Z
o1
(s) =
ˆ v
C
(s)
ˆ
i
LL
(s)
= −
G
vi
(s) · [1 − G
ic
(s)]
1 + G
v
(s) · G
ic
(s) · G
vi
(s)
(34)
Z
o2
(s) =
ˆ v
C
(s)
ˆ
i
LL
(s)
= −
G
vi
(s)
1 + G
v
(s) · G
ic
(s) · G
vi
(s)
. (35)
Comparing (34) and (35), it can be found that an extra factor
[1 − G
ic
(s)] appears in the output impedance with load current
feedforward, and the magnitude of the output impedance will
be reduced in the low frequency range because the gain of
the closed-loop transfer function G
ic
(s) closes to unity in the
bandwidth of the current loop. The Bode plot of the output
impedance of these two conditions is shown in Fig. 12, and
it can be seen that the magnitude of the output impedance is
reduced from dc to 600 Hz with the load current feedforward.
Consequently, the quality of the load voltage v
Cabc
will be
improved with the load current feedforward.
In the grid-tied mode, the inductor current is regulated by the
inductor current loop directly, and the inductor current reference
is mainly composed by the current reference I
gref dq
, and the
load current i
LLdq
. If the load current is not fed forward, the
output current i
odq
of the inverter will be fixed by I
gref dq
. As a
result, the disturbance of the load current will be fully injected
into the utility, and this can be represented by
ˆ
i
g
(s)
ˆ
i
LL
(s)
= −1. (36)
Fig. 12. Bode plot of the output impedance with and without the load current
feedforward, when DG operates in the islanded mode.
Fig. 13. Bode plot of the transfer function from load current to grid current
with and without the load current feedforward, when DGoperates in the grid-tied
mode.
With the load current feedforward, the disturbance of the
load current can be compensated by the inverter, and the trans-
fer function from the load current to the grid current can be
described by (37). The Bode plots of transfer function [see (36)
and (37)] are shown in Fig. 13, and the gain is mitigated up to
1050 Hz with the load current feedforward and therefore, the
quality of the grid current can be improved
ˆ
i
g
(s)
ˆ
i
LL
(s)
= G
ic
(s) − 1. (37)
V. SIMULATION AND EXPERIMENTAL RESULTS
A. Simulation Results
To investigate the feasible of the proposed control strategy,
the simulation has been done in PSIM. The power rating of a
three-phase inverter is 3 kWin the simulation. The parameters in
the simulation are shown in Tables I and II. The RMS of the rated
phase voltage is 115 V, and the voltage reference V
max
is set
as 10% higher than the rated value. The rated utility frequency
1186 IEEE TRANSACTIONS ON POWER ELECTRONICS, VOL. 29, NO. 3, MARCH 2014
TABLE II
PARAMETERS IN THE CONTROL SYSTEM
Fig. 14. Simulation waveforms of load voltage v
C a
, grid current i
g a
, and
inductor current i
La
when DG is in the grid-tied mode under condition of the
step down of the grid current reference from 9 A to 5 A with: (a) conventional
voltage mode control, and (b) proposed unified control strategy.
is 50 Hz, and the upper and the lower values of the limiter in
the PLL are given as 0.2 Hz higher and lower than the rated
frequency, respectively.
In the grid-tied mode, the dynamic performance of the con-
ventional voltage mode control and the proposed unified control
strategy is compared by stepping down the grid current refer-
ence from 9 A to 5 A. The simulation result of the voltage
mode control is shown in Fig. 14(a), and the current reference is
changed at the moment of 14 s. It is found that dynamic process
lasts until around 15.2 s. In the proposed unified control strategy,
the simulation result is represented in Fig. 14(b) and the time
interval of the dynamic process is less than 5 ms. Comparing
Fig. 15. Simulation waveforms of load voltage v
C a
, grid current i
g a
, and
inductor current i
La
when DG is transferred from the grid-tied mode to the
islanded mode with: (a) conventional hybrid voltage and current mode control,
and (b) proposed unified control strategy.
the simulation results above, it can be seen that the dynamic
performance of the proposed unified control strategy is better
than the conventional voltage mode control.
During the transition from the grid-tied mode to the islanded
mode, the proposed unified control strategy is compared with
the hybrid voltage and current mode control, and the simulation
scenario is shown as follows: 1) Initially, the utility is normal,
and the DG is connected with the utility; 2) at 0.5 s, islanding
happens; and 3) at 0.52 s, the islanding is confirmed.
Simulate results with the hybrid voltage and current mode
control is shown in Fig. 15(a). It can be seen that the grid cur-
rent drop to zero at 0.5 s, and that the load voltage is seriously
distorted from 0.5 to 0.52 s. Then, the load voltage is recovered
to the normal value after 0.52 s. Fig. 15(b) presents the simu-
late results with the proposed unified control strategy. Initially,
the magnitude of grid current is 9 A and follows the current
reference I
gref dq
. The magnitude and frequency of the load
voltage are held by the utility. After the islanding happens, the
amplitude of the load voltage increases a little to follow the
voltage reference V
max
, and the output current of DG decreases
autonomously to match the load power demand.
LIU et al.: UNIFIED CONTROL STRATEGY FOR THREE-PHASE INVERTER IN DISTRIBUTED GENERATION 1187
Fig. 16. Diagram of the experimental prototype of DG.
Fig. 17. Experimental waveforms when DG is in the islanded mode: CH1,
load current i
LLa
, 5 A/div; CH3, load voltage v
C a
, 100 V/div; CH4, inductor
current i
La
, 5 A/div.
Fig. 18. Experimental waveforms when DG is in the grid-tied mode: CH1,
load current i
LLa
, 5 A/div; CH2, grid voltage v
g a
, 100 V/div; CH4, grid current
i
g a
, 10 A/div.
Comparing the simulation results above, it can be found that
the voltage quality is improved deeply by the proposed con-
trol strategy in the transition from the grid-tied mode to the
islanded mode, and the speed of the islanding detection is no
more critical.
Fig. 19. Experimental waveforms when DG is transferred from the grid-tied
mode to the islanded mode with (a) conventional hybrid voltage and current
mode control, and (b) proposed unified control strategy: CH2, load voltage
v
C a
, 100 V/div; CH3, grid current i
g a
, 10 A/div; CH4, inductor current i
La
,
10 A/div.
B. Experimental Results
To verify the proposed control strategy, an experimental pro-
totype of DG has been established, which is shown in Fig. 16.
The utility for the DG is emulated by a three-phase transformer
and a voltage regulator connected with the actual utility. The
rated line voltage of the actual utility is 380 V. The emulated
utility is called as utility in the below. Moreover, the inverter
in the DG is fed by a three-phase diode rectifier, and the
dc-bus voltage is set to 400 V approximately by the voltage
1188 IEEE TRANSACTIONS ON POWER ELECTRONICS, VOL. 29, NO. 3, MARCH 2014
Fig. 20. Experimental waveforms when DG is transferred from the islanded mode to the grid-tied mode: CH1, grid voltage v
g a
, 100 V/div; CH2, load voltage
v
C a
, 100 V/div; CH3, grid current i
g a
, 10 A/div; CH4, inductor current i
La
, 10 A/div.
regulator. Besides, the control system is implemented fully dig-
itally by digital signal processor TMS320F28335 from Texas
Instruments. The parameters in the experimental DG, shown in
Tables I and II, are identical to the ones used in the simulation.
Fig. 17 shows the experimental waveforms when DG is in
the islanded mode, and it can be seen that the magnitude of the
load voltage equals 180 Vapproximately, and the total harmonic
distortion (THD) of the load voltage is 0.9%.
Fig. 18 shows the experimental waveforms when DG is in the
grid-tied mode. The magnitude of the grid current is closed to
9 A, and the THD of the grid current is 3.6% approximately.
When the DG is transferred from the grid-tied mode to the
islanded mode, the experimental results with the traditional hy-
brid voltage and current mode control and the proposed unified
control strategy are given in Fig. 19. Before the islanding hap-
pens, the magnitude of the load voltage is around 163 V. The
current injected into the utility i
ga
is in phase with the utility
voltage, and the magnitude of i
ga
is approximately 9 A. When
the switch S
u
opens, the islanding happens, and the grid current
i
ga
drops to zero. In the traditional hybrid voltage and current
mode control, it can be found that the load voltage is seriously
distorted upon the occurrence of islanding. And this condition
lasts until the islanding is confirmed by DG and the control
structure is changed to regulate the load voltage. However, with
the proposed unified control strategy, the distortion of the load
voltage is obviously improved, and the magnitude of the load
voltage increases slightly and is close to 180 V.
Fig. 20 shows the process when DGis transferred fromthe is-
landed mode to the grid-tied mode. From0 ms to around 300 ms,
the phase of the load voltage is regulated to resynchronize with
the utility voltage, and the phase difference is reduced gradually.
Then, the magnitude of the load voltage is regulated to equal
the utility voltage. At the moment of 350 ms, the switch S
i
is
turned on, and the current injected into the grid i
ga
increases
smoothly without huge inrush current, and the load voltage is
stable during the transition.
Fig. 21 shows the experimental waveforms when DG feeds
nonlinear load in the islanded mode. It can be seen that the
distortion of the load voltage is improved by the load current
feedforward, and the THD of the load voltage is reduced from
4.7% to 3.2%.
Fig. 21. Experimental waveform when DG feeds nonlinear load in islanded
mode (a) with load current feedforward and (b) without load current feedfor-
ward: CH1, load current i
LLa
, 5 A/div; CH3, load voltage v
C a
, 100 V/div;
CH4, inductor current i
La
, 5 A/div.
When the power of the nonlinear load is varied, the THD of
the load voltage is also changed, and the experimental results are
shown in Fig. 22. It can be seen that with the load current feed-
forward, the THD of the load voltage can always be mitigated
under different power of the nonlinear load.
Fig. 23 shows the experimental waveforms when DG feeds
nonlinear load in the grid-tied mode. It can be seen that with the
load current feedforward, there is harmonic component in the
LIU et al.: UNIFIED CONTROL STRATEGY FOR THREE-PHASE INVERTER IN DISTRIBUTED GENERATION 1189
Fig. 22. Variation of the THD of the load voltage with the power of the
nonlinear load when DG is in the islanded mode.
Fig. 23. Experimental waveforms when DG feeds nonlinear load in the grid-
tied mode (a) with load current feedforward and (b) without load current feed-
forward: CH1, inductor current i
La
, 10 A/div; CH2, load current i
LLa
, 5 A/div;
CH3, grid voltage v
g a
, 100 V/div; CH4, grid current i
g a
, 10 A/div.
inductor current i
La
, and the harmonics component in the grid
current is reduced.
The THD of the grid current under different power of the
nonlinear load and different amplitude of the grid fundamental
Fig. 24. Variation of the THD of the grid current when DG is in the grid-tied
mode with (a) the power of the nonlinear load, and (b) the amplitude of the grid
current.
current is investigated, and the experimental results are shown
in Fig. 24. In Fig. 24(a), the amplitude of the grid fundamental
current is set at 7 A, and the power of nonlinear load is varied. It
can be found that with the increase of the load power, the THDof
the grid current rises, and the THD of the grid current is reduced
with the load current feedforward. In Fig. 24(b), the power of
the nonlinear load is set at around 600 W, and the amplitude of
the grid fundamental current is changed. It can be seen that the
THD of the grid current can be mitigated at different magnitude
of the grid fundamental current.
VI. CONCLUSION
A unified control strategy was proposed for three-phase in-
verter in DG to operate in both islanded and grid-tied modes,
with no need for switching between two different control ar-
chitectures or critical islanding detection. A novel voltage con-
troller was presented. It is inactivated in the grid-tied mode,
and the DG operates as a current source with fast dynamic
1190 IEEE TRANSACTIONS ON POWER ELECTRONICS, VOL. 29, NO. 3, MARCH 2014
performance. Upon the utility outage, the voltage controller can
automatically be activated to regulate the load voltage. More-
over, a novel load current feedforward was proposed, and it can
improve the waveform quality of both the grid current in the
grid-tied mode and the load voltage in the islanded mode. The
proposed unified control strategy was verified by the simulation
and experimental results.
REFERENCES
[1] R. C. Dugan and T. E. McDermott, “Distributed generation,” IEEE Ind.
Appl. Mag., vol. 8, no. 2, pp. 19–25, Mar./Apr. 2002.
[2] R. H. Lasseter, “Microgrids and distributed generation,” J. Energy Eng.,
vol. 133, no. 3, pp. 144–149, Sep. 2007.
[3] C. Mozina, “Impact of green power distributed generation,” IEEE Ind.
Appl. Mag., vol. 16, no. 4, pp. 55–62, Jul./Aug. 2010.
[4] IEEE Recommended Practice for Utility Interface of Photovoltaic(PV)
Systems, IEEE Standard 929-2000, 2000.
[5] IEEE Standard for Interconnecting Distributed Resources with Electric
Power Systems, IEEE Standard 1547-2003, 2003.
[6] J. Stevens, R. Bonn, J. Ginn, and S. Gonzalez, Development and Testing
of an Approach to Anti-Islanding in Utility-Interconnected Photovoltaic
Systems. Livermore, CA, USA: Sandia National Laboratories, 2000.
[7] A. M. Massoud, K. H. Ahmed, S. J. Finney, and B. W. Williams, “Har-
monic distortion-based island detection technique for inverter-based dis-
tributed generation,” IET Renewable Power Gener., vol. 3, no. 4, pp. 493–
507, Dec. 2009.
[8] T. Thacker, R. Burgos, F. Wang, and D. Boroyevich, “Single-phase is-
landing detection based on phase-locked loop stability,” in Proc. 1st IEEE
Energy Convers. Congr. Expo., San Jose, CA, USA, 2009, pp. 3371–3377.
[9] S.-K. Kim, J.-H. Jeon, J.-B. Ahn, B. Lee, and S.-H. Kwon, “Frequency-
shift acceleration control for anti-islanding of a distributed-generation
inverter,” IEEE Trans. Ind. Electron., vol. 57, no. 2, pp. 494–504, Feb.
2010.
[10] A. Yafaoui, B. Wu, and S. Kouro, “Improved active frequency drift anti-
islanding detection method for grid connected photovoltaic systems,”
IEEE Trans. Power Electron., vol. 27, no. 5, pp. 2367–2375, May 2012.
[11] J. M. Guerrero, L. Hang, and J. Uceda, “Control of distributed uninter-
ruptible power supply systems,” IEEE Trans. Ind. Electron., vol. 55, no. 8,
pp. 2845–2859, Aug. 2008.
[12] M. C. Chandorkar, D. M. Divan, and R. Adapa, “Control of parallel con-
nected inverters in standalone ACsupply systems,” IEEETrans. Ind. Appl.,
vol. 29, no. 1, pp. 136–143, Jan./Feb. 1993.
[13] Y. Li, D. M. Vilathgamuwa, and P. C. Loh, “Design, analysis, and real-
time testing of a controller for multibus microgrid system,” IEEE Trans.
Power Electron., vol. 19, no. 5, pp. 1195–1204, Sep. 2004.
[14] F. Gao and M. R. Iravani, “A control strategy for a distributed generation
unit in grid-connected and autonomous modes of operation,” IEEE Trans.
Power Del., vol. 23, no. 2, pp. 850–859, Apr. 2008.
[15] S.-H. Hu, C.-Y. Kuo, T.-L. Lee, and J. M. Guerrero, “Droop-controlled
inverters with seamless transition between islanding and grid-connected
operations,” in Proc. 3rd IEEE Energy Convers. Congr. Expo., Phoenix,
AZ, USA, 2011, pp. 2196–2201.
[16] L. Arnedo, S. Dwari, V. Blasko, and S. Park, “80 kW hybrid solar in-
verter for standalone and grid connected applications,” in Proc. 27th IEEE
Appl. Power Electron. Conf. Expo., Orlando, FL, USA, 2012, pp. 270–
276.
[17] R. Tirumala, N. Mohan, and C. Henze, “Seamless transfer of grid-
connected PWM inverters between utility-interactive and stand-alone
modes,” in Proc. 17th IEEE Appl. Power Electron. Conf. Expo., Dallas,
TX, USA, 2002, pp. 1081–1086.
[18] R. Teodorescu and F. Blaabjerg, “Flexible control of small wind turbines
with grid failure detection operating in stand-alone and grid-connected
mode,” IEEE Trans. Power Electron., vol. 19, no. 5, pp. 1323–1332, Sep.
2004.
[19] H. Zeineldin, M. I. Marei, E. F. El-Saadany, and M. M. A. Salama, “Safe
controlled islanding of inverter based distributed generation,” in Proc. 35th
IEEE Power Electron. Spec. Conf., Aachen, Germany, 2004, pp. 2515–
2520.
[20] H. Zeineldin, E. F. El-Saadany, and M. M. A. Salama, “Intentional is-
landing of distributed generation,” in Proc. IEEE Power Eng. Soc. Gen.
Meeting, San Francisco, CA, USA, 2005, pp. 1496–1502.
[21] S. Jung, Y. Bae, S. Choi, and H. Kim, “Alowcost utility interactive inverter
for residential fuel cell generation,” IEEE Trans. Power Electron., vol. 22,
no. 6, pp. 2293–2298, Nov. 2007.
[22] R. Majumder, A. Ghosh, G. Ledwich, and F. Zare, “Control of parallel
converters for load sharing with seamless transfer between grid connected
and islanded modes,” in Proc. IEEE Power Eng. Soc. Gen. Meeting, Pitts-
burgh, PA, USA, 2008, pp. 1–7.
[23] H. Tao, J. L. Duarte, and M. A. M. Hendrix, “Line-interactive UPS using
a fuel cell as the primary source,” IEEE Trans. Ind. Electron., vol. 55,
no. 8, pp. 3012–3021, Aug. 2008.
[24] D. N. Gaonkar, G. N. Pillai, and R. N. Patel, “Seamless transfer of micro-
turbine generation systemoperation between grid-connected and islanding
modes,” Electr. Power Compon. Syst., vol. 37, no. 2, pp. 174–188, Jan.
2009.
[25] N. Reza, G. Gevorg, A. Mehrdad, and M. Mishel, “Grid-tied and stand-
alone operation of distributed generation modules aggregated by cascaded
boost converters,” J. Power Electron., vol. 10, no. 1, pp. 97–105, Jan. 2010.
[26] G. M. S. Azevedo, F. Bradaschia, M. C. Cavalcanti, F. A. S. Neves,
J. Rocabert, and P. Rodriguez, “Safe transient operation of microgrids
based on master-slave configuration,” in Proc. 3rd IEEE Energy Convers.
Congr. Expo., Phoenix, AZ, USA, 2011, pp. 2191–2195.
[27] I. J. Balaguer, Q. Lei, S. Yang, U. Supatti, and F. Z. Peng, “Control for
grid-connected and intentional islanding operations of distributed power
generation,” IEEE Trans. Ind. Electron., vol. 58, no. 1, pp. 147–157, Jan.
2011.
[28] S.-M. Lee, K.-N. Sung, T.-K. Lee, and W.-C. Lee, “UPS/APF power
conversion equipment with a seamless mode transfer,” in Proc. 8th IEEE
Int. Conf. Power Electron., Jeju, Korea, 2011, pp. 2769–2776.
[29] M. Ranjbar, H. Ebrahimirad, S. Mohaghegh, and A. Ghaleh, “Seamless
transfer of three-phase grid-interactive microturbine inverter between grid-
connected and stand-alone modes,” in Proc. 19th Iranian Conf. Electr.
Eng., Tehran, Iran, 2011, pp. 1–6.
[30] T.-S. Hwang and S.-Y. Park, “A seamless control strategy of distributed
generation inverter for critical load safety under strict grid disturbance,” in
Proc. 27th IEEE Appl. Power Electron. Conf. Expo., Orlando, FL, USA,
2012, pp. 254–261.
[31] F.-S. Pai, “An improved utility interface for microturbine generation sys-
tem with stand-alone operation capabilities,” IEEE Trans. Ind. Electron.,
vol. 53, no. 5, pp. 1529–1537, Oct. 2006.
[32] F.-S. Pai and S.-J. Huang, “Design and operation of power converter
for microturbine powered distributed generator with capacity expansion
capability,” IEEE Trans. Energy Convers., vol. 23, no. 1, pp. 110–118,
Mar. 2008.
[33] D. Dong, T. Thacker, R. Burgos, D. Boroyevich, F. Wang, and B. Giewont,
“Control design and experimental verification of a multi-function single-
phase bidirectional PWM converter for renewable energy systems,” in
Proc. 13th Eur. Conf. Power Electron. Appl., Barcelona, Spain, 2009,
pp. 1–10.
[34] D. Dong, T. Thacker, I. Cvetkovic, R. Burgos, D. Boroyevich, F. F. Wang,
and G. Skutt, “Modes of operation and system-level control of single-
phase bidirectional PWM converter for microgrid systems,” IEEE Trans.
Smart Grid, vol. 3, no. 1, pp. 93–104, Mar. 2012.
[35] G. Shen, D. Xu, and D. Xi, “Novel seamless transfer strategies for fuel
cell inverters from grid-tied mode to off-grid mode,” in Proc. 20th IEEE
Appl. Power Electron. Conf. Expo., Austin, TX, USA, 2005, pp. 109–113.
[36] G. Shen, D. Xu, and X. Yuan, “A novel seamless transfer control strategy
based on voltage amplitude regulation for utility-interconnected fuel cell
inverters with an LCL-filter,” in Proc. 37th IEEE Power Electron. Spec.
Conf., Jeju, Korea, 2006, pp. 1–6.
[37] G. Shen, D. Xu, and X. Yuan, “Instantaneous voltage regulated seamless
transfer control strategy for utility-interconnected fuel cell inverters with
an LCL-filter,” in Proc. 5th IEEE Int. Power Electron. Motion Control
Conf., Shanghai, China, 2006, pp. 1–5.
[38] S. Huang, L. Kong, and H. Xu, “Control algorithm research on seamless
transfer for distributed resource with a LCL filter,” in Proc. 3rd Int. Conf.
Electr. Utility DRPT, Nanjing, China, 2008, pp. 2810–2814.
[39] Q. Lei, S. Yang, and F. Peng, “Multi-loop control algorithms for seamless
transition of grid-connected inverter,” in Proc. 25th IEEE Appl. Power
Electron. Conf. Expo., Palm Springs, CA, USA, 2010, pp. 844–848.
[40] J.-H. Jung, E.-C. Nho, I.-D. Kim, H.-G. Kim, T.-W. Chun, and N.-S. Choi,
“Seamless control mode transfer of a PCS for islanding mode operation,”
in Proc. 8th IEEEInt. Conf. Power Electron., Jeju, Korea, 2011, pp. 2658–
2662.
[41] M. Cirrincione, M. Pucci, and G. Vitale, “A single-phase DG generation
unit with shunt active power filter capability by adaptive neural filtering,”
IEEE Trans. Ind. Electron., vol. 55, no. 5, pp. 2093–2110, May 2008.
LIU et al.: UNIFIED CONTROL STRATEGY FOR THREE-PHASE INVERTER IN DISTRIBUTED GENERATION 1191
[42] J. He, Y. W. Li, and M. S. Munir, “A flexible harmonic control approach
through voltage-controlled DG-grid interfacing converters,” IEEE Trans.
Ind. Electron., vol. 59, no. 1, pp. 444–455, Jan. 2012.
[43] Q. Lei, F. Z. Peng, and S. Yang, “Multiloop control method for high-
performance microgrid inverter through load voltage and current decou-
pling with only output voltage feedback,” IEEE Trans. Power Electron.,
vol. 26, no. 3, pp. 953–960, Mar. 2011.
[44] P. C. Loh and D. G. Holmes, “Analysis of multiloop control strategies for
LC/CL/LCL-filtered voltage-source and current-source inverters,” IEEE
Trans. Ind. Appl., vol. 41, no. 2, pp. 644–654, Mar./Apr. 2005.
[45] P. C. Loh, M. J. Newman, D. N. Zmood, and D. G. Holmes, “A compar-
ative analysis of multiloop voltage regulation strategies for single and
three-phase UPS systems,” IEEE Trans. Power Electron., vol. 18, no. 5,
pp. 1176–1185, Sep. 2003.
[46] M. J. Ryan and R. D. Lorenz, “A high performance sine wave inverter
controller with capacitor current feedback and “back-EMF” decoupling,”
in Proc. 26th IEEEPower Electron. Spec. Conf., Atlanta, GA, USA, 1995,
pp. 507–513.
[47] M. N. Marwali and A. Keyhani, “Control of distributed generation
systems—Part I: Voltages and currents control,” IEEE Trans. Power Elec-
tron., vol. 19, no. 6, pp. 1541–1550, Nov. 2004.
[48] A. Hasanzadeh, O. C. Onar, H. Mokhtari, and A. Khaligh, “A
proportional-resonant controller-based wireless control strategy with a re-
duced number of sensors for parallel-operated UPSs,” IEEE Trans. Power
Del., vol. 25, no. 1, pp. 468–478, Jan. 2010.
[49] M. Savaghebi, A. Jalilian, J. C. Vasquez, and J. M. Guerrero, “Secondary
control for voltage quality enhancement in microgrids,” IEEETrans. Smart
Grid, vol. 3, no. 4, pp. 1893–1902, Dec. 2012.
[50] S.-K. Chung, “A phase tracking system for three phase utility interface
inverters,” IEEE Trans. Power Electron., vol. 15, no. 3, pp. 431–438, May
2000.
[51] F. Blaabjerg, R. Teodorescu, M. Liserre, and A. V. Timbus, “Overview
of control and grid synchronization for distributed power generation sys-
tems,” IEEE Trans. Ind. Electron., vol. 53, no. 5, pp. 1398–1409, Oct.
2006.
[52] D. Dong, J. Li, D. Boroyevich, P. Mattavelli, I. Cvetkovic, and Y. Xue,
“Frequency behavior and its stability of grid-interface converter in dis-
tributed generation systems,” in Proc. 27th IEEE Appl. Power Electron.
Conf. Expo., Orlando, FL, USA, 2012, pp. 1887–1893.
[53] Y. A. R. I. Mohamed, H. H. Zeineldin, M. M. A. Salama, and
R. R. Seethapathy, “Seamless formation and robust control of distributed
generation microgrids via direct voltage control and optimized dynamic
power sharing,” IEEE Trans. Power Electron., vol. 27, no. 3, pp. 1283–
1294, Mar. 2012.
Zeng Liu (S’09) received the B.S. degree in electri-
cal engineering from Hunan University, Changsha,
China, in 2006, and the M.S. degree in electrical
engineering from Xi’an Jiaotong University, Xi’an,
China, in 2009, where he is currently working toward
the Ph.D. degree.
His research interests include control of single-
phase and multiphase power converters for uninter-
rupted power supply and utility application, model-
ing, and analysis and control of distributed power
system based on three-phase ac bus.
Jinjun Liu (M’97–SM’10) 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 Engineering
School as a teaching faculty. In 1998, he led the
founding of XJTU/Rockwell Automation Laboratory
and served as the Lab Director. From 1999 until early
2002, he was with the Center for Power Electron-
ics Systems, Virginia Polytechnic Institute and State
University, USA, as a Visiting Scholar. He then came
back to XJTU and in late 2002 was promoted to a Full Professor and the Head
of the Power Electronics and Renewable Energy Center, XJTU. During 2005
to early 2010, he served as the Associate Dean with the School of Electrical
Engineering, XJTU. He currently also serves as the Dean for Undergraduate Ed-
ucation, XJTU. He coauthored three books, published more than 100 technical
papers, holds 13 patents. His research interests include power quality control,
renewable energy generation and utility applications of power electronics, and
modeling and control of power electronic systems.
Dr. Liu is an AdCom member of the IEEE Power Electronics Society and
serves as Region 10 Liaison. He received several national, provincial, or minis-
terial awards for scientific or career achievements, and the 2006 Delta Scholar
Award. He also serves as an Associate Editor for the IEEE TRANSACTIONS ON
POWER ELECTRONICS. He is an AdCom member and the Chair of Student Ac-
tivities Committee for IEEE Xi’an Section. He is on the Executive Board and
serving as a Deputy Secretary-General for the China Power Electronics Society,
and also on the Executive Board and serving as a Deputy Secretary-General for
the China Power Supply Society.
Yalin Zhao received the B.S. degree in electrical
engineering from Xi’an Jiaotong University, Xi’an,
China, in 2011, where he is currently working toward
the M.S. degree.
His research interests include dead-time compen-
sation, stability analysis, and control of inverters.