energies

Article
Coordinated Control for Large-Scale Wind Farms
with LCC-HVDC Integration
Xiuqiang He 1 ID
, Hua Geng 1, *, Geng Yang 1 and Xin Zou 2
1 Department of Automation, Tsinghua University, Beijing 100084, China;
he-xq16@mails.tsinghua.edu.cn (X.H.); yanggeng@tsinghua.edu.cn (G.Y.)
2 State Power Economic Research Institute, State Grid Corporation of China, Beijing 102209, China;
zouxin@chinasperi.sgcc.com.cn
* Correspondence: genghua@tsinghua.edu.cn; Tel.: +86-10-6277-0559




Received: 30 July 2018; Accepted: 21 August 2018; Published: 23 August 2018


Abstract: Wind farms (WFs) controlled with conventional vector control (VC) algorithms cannot be
directly integrated to the power grid through line commutated rectifier (LCR)-based high voltage
direct current (HVDC) transmission due to the lack of voltage support at its sending-end bus. This
paper proposes a novel coordinated control scheme for WFs with LCC-HVDC integration. The
scheme comprises two key sub-control loops, referred to as the reactive power-based frequency (Q-f )
control loop and the active power-based voltage (P-V) control loop, respectively. The Q-f control,
applied to the voltage sources inverters in the WFs, maintains the system frequency and compensates
the reactive power for the LCR of HVDC, whereas the P-V control, applied to the LCR, maintains
the sending-end bus voltage and achieves the active power balance of the system. Phase-plane
analysis and small-signal analysis are performed to evaluate the stability of the system and facilitate
the controller parameter design. Simulations performed on PSCAD/EMTDC verify the proposed
control scheme.

Keywords: HVDC; line commutated converter; wind farm; frequency stability; frequency control;
voltage stability; voltage control; vector control; voltage-source converter

1. Introduction
The power system is facing unprecedented technical challenges due to abundant large-scale
renewable power plant integration [1–5]. In China, large-scale wind farms (WFs) are mainly built
in remote areas of the northwest. As the local alternating current (AC) network is quite weak and
the penetration level of wind power is extremely high, it is a critical technical issue to integrate and
deliver large-scale wind power into the southeastern power grid. Emerging ultra HVDC (UHVDC)
transmission technology is able to provide an available solution. To date, most of the UHVDC systems
that are being planned or that are already built are of the line commutated converter (LCC) type,
which are suitable for long-distance and large-capacity transmission, with advantages such as low
expenditure and power loss [4]. Recently, a ±800 kV UHVDC with 10 GW capacity is being planned
to deliver wind and solar power on the Tibetan Plateau into the eastern load center, and wind and
solar power accounts for about 85% of the transmission capacity. It is quite difficult for the system to
maintain stable operation when there is no traditional generating set that is available at the sending
end of the UHVDC, due to some special factors, e.g., circuit faults, but only islanded WFs and/or
photovoltaic power plants. Therefore, it is necessary to study the control scheme of large-scale WFs
with LCC-HVDC integration as a technical reserve [4–7].
Vector control (VC) algorithms are commonly used for the control of wind energy conversion
systems (WECSs). Conventional VC [8] is based on the orientation of the grid voltage vector, and

Energies 2018, 11, 2207; doi:10.3390/en11092207 www.mdpi.com/journal/energies

Energies 2018, 11, 2207 2 of 19

thus a stiff grid is required to ensure the stability of the systems [9]. Unlike the voltage source
converter in VSC-HVDC, the line commutated rectifier (LCR) in LCC-HVDC cannot actively generate
the referenced three-phase voltage, essentially because of the application of semi-controlled switching
devices, e.g., thyristors. Moreover, a steady and balanced commutation voltage is a prerequisite for the
operation of the LCR. Consequently, under the condition that there is no available voltage support at
the sending-end bus (SEB) of LCC-HVDC, the WFs with conventional VC algorithms cannot can be
integrated by the LCC-HVDC directly [10–12].
A simple approach is to introduce a static synchronous compensator (STATCOM) on the SEB
in order to provide the voltage support [10–12]. However, the extremely high reliability and large
capability is required for the STATCOM, which results in high operating costs and power loss [13].
Without voltage support, the critical issue in the system is to guarantee the stability of the SEB voltage
vector, including both the voltage stability and the frequency stability. Considering that the LCR is
controllable in terms of active power, both the frequency and voltage stability issues can be addressed
through the division of labor between the WF and the LCR [14–19], i.e., the voltage and the frequency
are controlled by the WF and the LCR respectively, or conversely.
For the doubly-fed induction generator (DFIG)-based offshore WF at steady states, the stator
voltage of the DFIG is the product of its stator flux and the SEB frequency [14]. Based on this
fact, the earliest approach, where the stator flux and the frequency are controlled by the WF and
the LCR respectively, is proposed in [14,15]. A similar approach can be found in [16], where the
stator voltage and the frequency are controlled by the WF and the LCR respectively. In both of the
approaches, the frequency is regulated by the active power of the WF, whereas the voltage is regulated
by the reactive power of the WF. Actually, there is a substantial amount of capacitive compensation
on the SEB of the LCC-HVDC or diode-based HVDC, leading to a strong coupling between the
bus voltage and the active power balance, and also between the frequency and the reactive power
balance [17,20–22]. Consequently, references [17,20–22] develop a novel control concept, where the
frequency is regulated by the reactive power, whereas the voltage is regulated by the active power.
There is another coordination approach developed in [18,19] where the voltage is controlled by both
the WF and the LCR. As a result, the strong coupling between the active and reactive power control
loops occurs, which would affect the dynamic performance of the system. Moreover, the frequency
stability was not addressed in the approach.
In the aforementioned approaches, the control algorithms of the WFs are based on the conventional
VC structure where phase-locked loops (PLLs) are employed to detect the phase-angle of the stator
voltage. It is reported that PLLs play an important role in the system dynamics, and the system stability
involving PLLs are quite complicated [9], especially when WFs are connected to the weak, or even
isolated grids [23,24]. There are some intensive studies regarding the PLL-less DFIG control algorithms
for standalone applications, referred to as the indirect self-orientated vector control (ISOVC) [25–28].
In the ISOVC, the phase-angle, adopted in the coordinate transformation between abc and dq reference
frames, is derived from a free running integral of the rated synchronous speed ω 0 instead of the PLL.
It is worth noting that the supplementary indirect orientation control is realized through modifying
the original active power loop, and thus the auxiliary torque and pitch angle control is required to
regulate the active power [25–28]. Since the probability of frequency instability due to the dynamic
characteristics of PLLs can be completely avoided in the ISOVC, it is applied to control standalone
DFIGs with LCC-HVDC integration in [29]. In order to be employed in multi-machine scenarios,
additional active power droop loop should be introduced into ω 0 to achieve synchronization and
power sharing among multiple machines [29,30]. Unfortunately, with such droop scheme, the WF
cannot always track its maximum power point with sacrificed economic benefits.
In this paper, a novel scheme with respect to the division of labor between the WF and the LCR is
proposed. On one hand, considering the coupling relationship between the frequency and the reactive
power balance [17–20], a novel indirect orientation control based on the reactive power loop instead
of the active power loop [25–28] is developed. In the scheme, reactive power droop is employed for
Energies 2018, 11, 2207 3 of 19

Energies 2018, 11, x FOR

synchronization andPEER REVIEW
reactive power sharing among multiple machines. Therefore, it would not affect 3 of 19
the active power tracking of the WF. On the other hand, the control objective of the LCR is to maintain
voltage
the SEB and the active
voltage power
stability. balance,
Actually, notthere
since onlyiscan the voltage
a strong stability
coupling betweenbe addressed,
the voltagebut also
and thethe WF
active
is able to capture the maximum active power with the proposed scheme.
power balance, not only can the voltage stability be addressed, but also the WF is able to captureThe proposed scheme
comprises
the maximum two active
key sub-control
power with loops. One is thescheme.
the proposed reactive The
power-based
proposedfrequency (Q-f) control
scheme comprises twoloop
key
for the voltage source inverters (VSIs) in the WF, where a novel ISOVC is developed
sub-control loops. One is the reactive power-based frequency (Q-f ) control loop for the voltage source to maintain the
SEB frequency and compensate reactive power for the LCR. The other is
inverters (VSIs) in the WF, where a novel ISOVC is developed to maintain the SEB frequency and the active power-based
voltage
compensate(P-V) control
reactive loopfor
power forthe
theLCR.
LCR,Thethrough
other iswhich the power-based
the active SEB voltage voltageis controlled to achieve
(P-V) control loop
the active power balance.
for the LCR, through which the SEB voltage is controlled to achieve the active power balance.
The rest of
The rest of the
the paper
paper isis organized
organized asas follows.
follows. Section
Section 22 depicts
depicts thethe mathematical
mathematical models,
models, andand
explains
explains the
the relationship
relationship between
between thethe voltage
voltage andand the active power
the active power balance,
balance, and and also between the
also between the
frequency
frequency and the reactive power balance. Section 3 proposes the Q-f and P-V control loops
and the reactive power balance. Section 3 proposes the Q-f and P-V control after
loops after
analyzing
analyzing the operational principles.
the operational principles. Section
Section 44 demonstrates
demonstrates the the stability
stability of of the
the proposed control
proposed control
scheme
scheme andand designs
designs the
the controller parameters. Section
controller parameters. Section 55 shows
shows the
the simulation
simulation results
results and
and verifies
verifies the
the
feasibility of the coordinated control scheme. Section 6 concludes
feasibility of the coordinated control scheme. Section 6 concludes this paper. this paper.

2.
2. System
System Modeling
Modeling
The topology of
The topology the studied
of the studied system
system is is shown
shown in in Figure
Figure 1.1. This
This study
study takes
takes permanent-magnet
permanent-magnet
synchronous
synchronous generator
generator(PMSG)-based
(PMSG)-basedWECSs WECSs asasananexample
example to study the coordination
to study the coordination between WFs
between
and LCC-HVDC. The proposed control scheme can be easily extended
WFs and LCC-HVDC. The proposed control scheme can be easily extended to doubly-fed induction to doubly-fed induction
generator
generator (DFIG)-based
(DFIG)-basedWECSs. WECSs.It It should
should bebenoticed
noticedthatthat
thethe
Q-f Q-f
control is applied
control in grid-side
is applied VSIs
in grid-side
of PMSG-based
VSIs of PMSG-based WECSs, whereas
WECSs, it is applied
whereas in rotor-side
it is applied converters
in rotor-side convertersof DFIG-based
of DFIG-based WECSs. For
WECSs.
simplicity, the PMSG-based WECS can be equivalent to a voltage source
For simplicity, the PMSG-based WECS can be equivalent to a voltage source inverter (VSI) in parallelinverter (VSI) in parallel
with aadirect
with directcurrent
current(DC) (DC) capacitor
capacitor andand a controlled
a controlled DC current
DC current sourcesource
[31]. In [31]. In to
order order to supply
supply energy
energy
for the system startup, batteries are installed at the DC bus of several units (no need for all for
for the system startup, batteries are installed at the DC bus of several units (no need all
units).
units). Especially, for the DFIG-based WECSs, rotor excitation of DFIGs should
Especially, for the DFIG-based WECSs, rotor excitation of DFIGs should be provided initially in the be provided initially
in the startup
startup process,process,
which canwhich can be accomplished
be accomplished throughthrough the rotor-side
the rotor-side converters converters
poweredpowered by the
by the batteries.
batteries. The WECSs are connected to the SEB of HVDC. The rectifier is of
The WECSs are connected to the SEB of HVDC. The rectifier is of the LCC type, and thus a substantialthe LCC type, and thus a
substantial
amount of AC amount
filtersofare
ACconfigured
filters are atconfigured
the SEB to at mitigate
the SEB theto mitigate the current and
current harmonics, harmonics,
meanwhileand
meanwhile
they provide they provide
reactive reactive
power power compensation,
compensation, which can which can be equivalent
be equivalent to a capacitor
to a capacitor bankbank
Cf atCthe
f at

the fundamental frequency.

fundamental frequency.

Local load
Tw
Battery WECS HVDC
…

SEB Bus
Trc
Main grid
Filter Filter

Figure 1. System
System topology.

For
For aa clear
clear description
description of the system
of the system model,
model, it
it can
can be
be divided
divided into
into three
three subsystems:
subsystems: the
the WECS
WECS
subsystem, the SEB subsystem and the HVDC subsystem. To capture the
subsystem, the SEB subsystem and the HVDC subsystem. To capture the fundamental power dynamic fundamental power
dynamic characteristics
characteristics of thethe
of the system, system, the switching
switching function
function models models
[10,11] [10,11]
of the of the converters
converters are
are employed.
employed. Moreover, the inverter of HVDC can be equivalent to a DC voltage source [14,15]
Moreover, the inverter of HVDC can be equivalent to a DC voltage source [14,15] since it is subjected since it
is subjected to a constant-voltage control and has little effect on the other subsystems
to a constant-voltage control and has little effect on the other subsystems under normal operations. under normal
operations. Figure
Figure 2 depicts 2 depicts
the the equivalent
equivalent circuit
circuit of the of system,
whole the whole system,
where the where the significations
significations of the
of the electrical
electrical variables and parameters are self-explanatory.
variables and parameters are self-explanatory.
Energies 2018, 11, 2207 4 of 19
Energies 2018, 11, x FOR PEER REVIEW 4 of 19

udc uwa Rw Lw iwa uba Rrc Lrc urca irca id Rd Ld

 

i1 uwb iwb ubb urcb ircb 
 
 udr udi
Cdc uwc iwc ubc urcc ircc


 

ica icb icc
VSI Cf LCR
WECS SEB HVDC

Equivalent circuit,
Figure 2. Equivalent
Figure circuit, where
where thethe wind
wind farm
farm (WF)
(WF) isis represented
represented byby aa single
single wind
wind energy
energy
conversion systems
conversion systems(WECS)
(WECS)just forfor
just a simplified system
a simplified model.
system Note Note
model. that the proposed
that control control
the proposed scheme
is applicable
scheme for multiple
is applicable WECSs WECSs
for multiple (see the (see
nextthe
section
next for morefor
section details).
more details).

2.1.
2.1. WECS
WECS Subsystem
Subsystem Model
Model
In
In the rated synchronous reference frame (RSEF) with the rated frequency ω ω00,, the
the WECS
WECS model
can
can be written as:

Lw  di (ωb
dtb)dt= −RRwwiiwd  ubdbd+  0 Lω
(
diLwd
 w /wd wd − u
mdmudc
d udc +
i Lw iwq
w 0wq
 (1)
(1)
LwdiLwq (ωbdt
/wq
di  )dt= −RRwwiiwq 
− uubqbq+mm u dc
qdcu −0 Lωw0iwd
Lw iwd
 w b wq q

Cdc dudc /(ωb dt) = i1 − md iwd + mq iwq
 
(2)
Cdc dudc b dt   i1  md iwd  mq iwq (2)
where the subscripts d and q represent the variables transformed from the three-phase abc reference
where
frame tothethe
subscripts
dq RSEF, dand
andωqb represent
is the basethe variablessimilarly
frequency, transformed from the three-phase abc reference
hereinafter.
frame to the dq RSEF, and ωb is the base frequency, similarly hereinafter.
2.2. SEB Subsystem Model
2.2. SEB Subsystem
Similarly, Model
the SEB model in the RSEF can be written as:
Similarly, the SEB model(in the RSEF can be written as:
C f dubd /(ωb dt) = iwd − ircd + ω0 C f ubq
C
fCdu dubd(ω
f bq / b ) 
b dt
dt =iiwq  i rcq−
wd − ircd
0ωC0fCubq
f ubd
(3)
 (3)
 f bq  b  wq rcq
 C du  dt  i  i   C
0 f bd
u
Let the amplitude and the phase-angle of the SEB voltage be:
Let the amplitude and the phase-angle of theqSEB voltage be:
Ubm = u2bd + u2bq
Ubm  u2  u2  (4)
φ = arctan bdubq /u bd , φ ∈ [0, 2π )
bq
(4)

  arctan ubq ubd ,  0, 2  
It can be obtained in the polar coordinate system that [19,20]:
It can be obtained in the polar coordinate system that [19,20]:
2
0.5dUbm /(ωb dt) = ( Pw − Prc )/C f (5)
0.5 dUbm2
bdt    Pw  Prc  C f (5)
 
2
ω0 + dφ /(ωb dt) = ω1 = (− Qw + Qrc )/ C f Ubm (6)
0  d b dt   1   Qw  Qrc  C f Ubm 
2

where Pw = ubd iwd + ubq iwq and Prc = ubd ircd + ubq ircq are the active powers from the WECS, and they
 (6)

are absorbed
where Pw = ubdby
iwd the
+ ubqHVDC
iwq andrespectively.
Prc = ubdircd + uAlso, Qw the
bqircq are = –u bd iwq +
active ubq iwd from
powers and Qthe –ubd ircqand
rc =WECS, + uthey
bq ircd are
the reactive
absorbed by powers
the HVDC from the WECS Also,
respectively. and theyQw =are–ubdabsorbed
iwq + ubqiwd by
andthe
QrcLCR= –ubdrespectively,
ircq + ubqircd areand 1 is the
theωreactive
SEB frequency.
powers from the WECS and they are absorbed by the LCR respectively, and ω1 is the SEB frequency.
Equations (5)
Equations (5)and
and(6)(6) laylay
thethe
foundations
foundations for this
for study. From the
this study. From perspective of the filter
the perspective capacitor
of the filter
parallel branch,
capacitor parallelthebranch,
WECS the canWECS
be seencanas abecontrolled
seen as a power source
controlled (Pw and
power sourceQw ),(Pwhile
w and theQw),LCRwhilecanthebe
seen as a controlled power
LCR can be seen as a controlled powerload (P rc and Q
load). Given that the WECS and the HVDC
rc (Prc and Qrc). Given that the WECS and the HVDC are are interconnected
by the filter capacitor,
interconnected by the two filtersignificant
capacitor, results can be drawn
two significant resultsfrom
canthe
be filer
drawn capacitor
from the point
filerofcapacitor
view. (1)
The voltage
point of view.amplitude Ubm is highly
(1) The voltage coupled
amplitude Ubmwith the active
is highly coupledpowerwith(Pthe
w − P );
active
rc (2) The
power phase-angle
(P w − Prc); (2)φ
(or the
The frequency ω
phase-angle  1 )(or
is highly coupledωwith
the frequency 1) is the reactive
highly coupledpowerwith w −reactive
(Qthe Qrc ) [20].power
A physical
(Qw − mechanism
Qrc) [20]. A
explanation is given as follows.
physical mechanism explanation is given as follows.
It is known that the total instantaneous power of the three-phase balanced capacitor branch
equals its active power (which is zero) at steady states, and the power exchanged within the three-
phase capacitors charges/discharges the capacitors. As a result, the three-phase capacitor circuit as a
Energies 2018, 11, 2207 5 of 19

It is known that the total instantaneous power of the three-phase balanced capacitor branch equals
its active power (which is zero) at steady states, and the power exchanged within the three-phase
capacitors charges/discharges the capacitors. As a result, the three-phase capacitor circuit as a whole
exhibits a certain reactive power. However, this conclusion becomes invalid during dynamic processes.
For Equation (5), if there is an active power deviation, e.g., Pw − Prc > 0, this indicates that the active
power generated by the WECS is larger than that absorbed by the LCR, then the extra active power will
charge the capacitors. Consequently, the instantaneous current amplitude will increase, which leads
the instantaneous voltage amplitude to increase too. Similarly, if there is a reactive power deviation for
Equation (6), e.g., (–Qw + Qrc ) > 0, which indicates that the reactive power generated by the WECS is
smaller than that absorbed by the LCR, then the voltage phase-angle (i.e., the instantaneous frequency)
will increase, assuming that the voltage amplitude keeps unchanged. As a consequence, the capacitive
reactance will decrease due to the frequency increase, and therefore the capacitor will generate more
reactive power to try to balance the reactive power. In fact, the capacitive parallel branch is in a dual
relationship with the inductive series branch in traditional power systems, and thus it is not difficult to
understand the foregoing coupling relationship. Moreover, it can be also found that a smaller Cf (the
absolute minimum filter guarantees Cf > 0) can result in a stronger coupling.

2.3. HVDC Subsystem Model

In the RSEF, the HVDC model can be written as [11,12]:
(
Lrc dircd /(ωb dt) = − Rrc ircd + ubd − urcd + ω0 Lrc ircq
(7)
Lrc dircq /(ωb dt) = − Rrc ircq + ubq − urcq − ω0 Lrc ircd
(

urcd = Urcm ircd sin α + ircq cos α

(8)
urcq = Urcm ircd cos α − ircq sin α

udr = Urcm cos α (9)

Ld did /(ωb dt) = udr − udi − Rd id (10)

where α is the firing angle, and Urcm is the voltage amplitude in the rectifier bridge side. Note that the
variables are in the per-unit system, and thus the rectifier coefficient is eliminated. Also, both the (12k
± 1) order harmonics in ac side and the 12k order harmonics in DC side are neglected in the typical
12-plus HVDC model, since the events of major concern are the fundamental power conversion rather
than high frequency dynamic.

3. Coordinated Control Scheme

Prior to describing the proposed coordinated control scheme, the control requirements should be
emphasized first.

(1) Voltage control: a stable voltage can offer voltage support for the WECSs, as well as the
commutation voltage for the LCR.
(2) Frequency control: the frequency stability should be maintained so that multiple WECSs are able
to operate synchronously.
(3) Active power balance: considering that the wind conditions are not controlled, the active power
generated from the WFs should be equal to that which is transmitted into the HVDC in real time.
(4) Reactive power balance: considering that the reactive power compensation capability of AC
filters is discontinuous, the WECSs should be able to compensate and share the insufficient or
excessive reactive power automatically.

The proposed coordinated control scheme is able to achieve the requirements. The Q-f control
applied into the wind farm meets the requirements (2) and (4), whereas the P-V control applied into
Energies 2018, 11, 2207 6 of 19

Energies 2018, 11, x FOR PEER REVIEW 6 of 19

the HVDC rectifier meets the requirement (1) and (3). Thus, the wind farm and the HVDC cooperate
with Note that to
each other theachieve
resynchronization
system stability.capability is also of much significance for the system
uninterrupted
Note thatoperation in the case of a fault.
the resynchronization Underis
capability fault
alsoconditions,
of muchthe back-end converters
significance can be
for the system
controlled
uninterruptedto supply zero in
operation power temporarily,
the case of a fault.and then the
Under faultbatteries can be
conditions, theutilized
back-end again to help
converters
generate the SEB voltage
can be controlled after
to supply thepower
zero fault istemporarily,
cleared. More andtechnical
then thedetails willcan
batteries be given in future
be utilized work.
again to help
generate the SEB voltage after the fault is cleared. More technical details will be given in future work.
3.1. Q-f Control of WECSs
3.1. Q-f Control of WECSs
In contrast to the ISOVC in [25–28], where the active power loop is adopted to achieve indirect
In contrast
orientation to the ISOVC in [25–28],
and synchronization, a novelwhere
ISOVCthe active
will power loop
be developed is adopted
here. Since that to achieve indirect
the relationship
orientation
between theand synchronization,
frequency a novel
and reactive power,ISOVC will bein
as shown developed
Equationhere. Since
(6), the that the
reactive relationship
power loop is
between the frequency and reactive power, as shown in Equation (6), the reactive
adopted to achieve Q-f control. As depicted in Figure 3, two key modifications are made in the power loopQ-f
is
adoptedcompared
control Q-f the
to achievewith control. As depicted
conventional in Figure
VC with 3, two
the unity key modifications
power factor. One is thatare made in the Q-f
the self-defined
control compared
phase-angle is: with the conventional VC with the unity power factor. One is that the self-defined
phase-angle is:
∠U =  ω00 + ∠
U    0
Z
(11)
0 (11)
where 0 is an initial value, determined by the final value of the phase-angle at the end of system
where ∠0 is an initial value, determined by the final value of the phase-angle at the end of system
startup. The other is that the control object of the reactive power loop is regulating the q-axis voltage
startup. The other is that the control object of the reactive power loop is regulating the q-axis voltage
ubq instead of the reactive current iwq to zero. As a consequence, not only is the frequency ω1 clamped
ubq instead of the reactive current iwq to zero. As a consequence, not only is the frequency ω 1 clamped
when ubq = 0 since the actual phase-angle is consistent with the self-defined one, but the reactive
when ubq = 0 since the actual phase-angle is consistent with the self-defined one, but the reactive
current command iwq* can also be regulated automatically so that the WECS is able to compensate
current command iwq * can also be regulated automatically so that the WECS is able to compensate
reactive power for the LCR. Note that the active power control is still based on the maximum power
reactive power for the LCR. Note that the active power control is still based on the maximum power
point tracking (MPPT) control, as shown in Figure 3b. The operational principle of the Q-f control is
point tracking (MPPT) control, as shown in Figure 3b. The operational principle of the Q-f control is
described as follows.
described as follows.


iwd ud
udc PI 1 PI 2
SVPWM

+– +– ++
 dq
udc iwd ud

uq
iwq 0 +– PI 3 ++ abc
iwq uq
U b
ubdq PLL
dq
ubabc
iwdq
abc iwabc
...

(a)

iwd ud
udc PI 1 PI 2
SVPWM

+ +– ++
 – i ud
dq
udc  wd

iwq uq
ubq 0 +– PI 3 +– PI 4 ++ abc
ubq iwq uq
ubdq dq
ubabc

abc
 iw2 max  iwd
2 U  0  0
iwdq dq
abc iwabc
...

(b)

Figure 3.
Figure 3. (a)
(a) Conventional vector control
Conventional vector control (VC)
(VC) versus
versus (b)
(b) proposed
proposed reactive
reactive power-based frequency
power-based frequency
(Q-f ) control
(Q-f) control for
for voltage
voltagesource
sourceinverters
inverters(VSIs)
(VSIs)ofof
WECSs.
WECSs. InIn
actual practice,
actual thethe
practice, controlled variable
controlled ubq
variable
can be replaced by the local voltage information instead of sending-end bus (SEB)
ubq can be replaced by the local voltage information instead of sending-end bus (SEB) voltage voltage information.
Since the critical
information. information
Since is the
the critical voltage-phase
information angle
is the rather than the
voltage-phase voltage
angle amplitude,
rather than thethevoltage
spatial
distribution feature of the voltage amplitude has litter influences on the Q-f control performance.
amplitude, the spatial distribution feature of the voltage amplitude has litter influences on the Q-f
control performance.
Assuming that the system is in a steady state, if the frequency ω 1 suddenly starts to increase due
Assuming that
to a disturbance, the
e.g., system isincrease
a sudden in a steady state,
of the if thepower
reactive frequency
Qrc , ω 1 suddenly
the starts U
voltage vector tobincrease due
and current
to a disturbance,
vector e.g., a sudden
Iw under conventional VCincrease of the
is shown reactive
in Figure 4a.power
It can Qberc,observed
the voltage
thatvector Ub and current
the controller tracks
vector Iw underchange
the frequency conventional VC is
and cannot shown
output inreactive
the Figure 4a. It can
power to be observed
regulate the that the controller
frequency, tracks
which further
the frequency
leads change and
to the instability cannot output
of frequency. the reactive
In Figure 4b, thepower to regulate
self-oriented the frequency,
control which but
(11) is adopted, further
the
leads to the instability of frequency. In Figure 4b, the self-oriented control (11) is adopted, but the
reactive power loop still regulates iwq to zero. Under this condition, Ub rotates counterclockwise an
angle ϕ due to the increase of frequency. Thus, although the WECS outputs reactive power, Ub no
Energies 2018, 11, 2207 7 of 19

reactive power loop still regulates iwq to zero. Under this condition, U b rotates counterclockwise an
Energies 2018, 11, x FOR PEER REVIEW 7 of 19
angle φ due to the increase of frequency. Thus, although the WECS outputs reactive power, U b no
longer coincides with
longer coincides with the q-axisbecause
the q-axis becauseof ofaalack
lackofofsynchronization
synchronizationcontrol.control.Thereafter,
Thereafter, the
the d- d-
andandq-
q-axis controls are no longer decoupled, and moreover, multiple WECSs
axis controls are no longer decoupled, and moreover, multiple WECSs may become asynchronous may become asynchronous
due to the
due to the accumulation
accumulationof ofthe
theangle
angleerrors.
errors.InIn Figure
Figure 4c,4c,
thethe issue
issue is completely
is completely addressed
addressed wherewhere
the
the control object is u = 0 instead of i = 0. In other words, the phase-angle is indirectly
control object is ubq = 0 instead of iwq = 0. In other words, the phase-angle is indirectly controlled to
bq wq controlled
to follow
follow thethe angle
angle generated
generated by by
thethe rated
rated frequency
frequency 0 , and
ω0,ωand the the controller
controller output
output signal
signal is exactly
is exactly the
the reactive
reactive current
current reference.
reference. Consequently,
Consequently, thethethird
thirdproportional-integral
proportional-integral(PI (PI3)3 )regulator
regulatorisis able
able toto
produce an exact reactive current reference i * under the condition that u = 0.
produce an exact reactive current reference iwq under the condition that ubq = 0.
wq * bq

ωpll dpll d d d

Ub Ub Ub
Iw Iw Iw
 

q qpll ω0 q ubq ω0 q iwq ω0

(a) (b) (c)

Figure 4.
Figure 4. Vector
Vector diagrams
diagrams in
in the
the rated
rated synchronous
synchronous reference
reference frame
frame (RSRF)
(RSRF) under
under different
different control
control
algorithms: (a)
algorithms: (a) Conventional
Conventional VC
VC with
with iiwq == 0;
0; (b) Self-oriented control
(b) Self-oriented control with
with iiwq == 0;
0; (c) Proposed Q-f
(c) Proposed Q-f
wq wq
control with ubq = 0.
control with ubq = 0.

It should be noted that the PI3 regulator cannot work well in multi-WECS scenarios due to a
It should be noted that the PI3 regulator cannot work well in multi-WECS scenarios due to
reactive current circulation among the WECSs without a sharing scheme. To this end, a simple
a reactive current circulation among the WECSs without a sharing scheme. To this end, a simple
approach is to improve the PI-type regulator into a P-type one, resulting in a droop characteristic.
approach is to improve the PI-type regulator into a P-type one, resulting in a droop characteristic.
Thus, when a multi-machine system is subjected to the Q-f droop control, the reactive power can be
Thus, when a multi-machine system is subjected to the Q-f droop control, the reactive power can
compensated and shared automatically, and both the frequency stability and the synchronization
be compensated and shared automatically, and both the frequency stability and the synchronization
stability can be realized. With the P-type control, the voltage vector will no longer coincide with the
stability can be realized. With the P-type control, the voltage vector will no longer coincide with the
d-axis. By defining an appropriate range of the included angle between them, and thereby setting an
d-axis. By defining an appropriate range of the included angle between them, and thereby setting an
appropriate proportional coefficient, it is doable to ensure that the included angle is small enough
appropriate proportional coefficient, it is doable to ensure that the included angle is small enough and
and close to zero at steady states.
close to zero at steady states.
According to the P-type control, for one WECSj, it can be obtained that:
According to the P-type control, for one WECSj , it can be obtained that:
iwqj  kpj ubqj (12)
iwqj = −k pj ubqj (12)
where ubqj can be considered to be approximately the same for different units. For one thing, in actual
where
practice,ubqjubqj
can= be0 atconsidered
the time to be approximately
when WECSj switches the from
same the
for different units. For one
pre-synchronization stagething,
to thein
actual practice,
connection to thebqj u = 0 at the time when WECS switches from the pre-synchronization
sending-end grid by means of jphase-locked loops. For another, after the connection stage to the
connection to the sending-end
to the sending-end grid, the phase-locked grid by meansloops of phase-locked
are withdrawn.loops.Then,
For another, after the connection
the phase-angle difference
to the sending-end grid, the phase-locked loops are withdrawn.
between two units during normal operating conditions are also eliminated by their Then, the phase-angle difference
initial
between
synchronoustwo units during
reference normal operating conditions are also eliminated by their initial synchronous
frames.
reference
On theframes.
basis of Equation (12), the reactive power can be written as:
On the basis of Equation (12), the reactive power can be written as:
 
  kpj ubqj
Qwj  ubdj iwqj  ubqj iwdj  kpj ubdj  iwdj ubqj (13)
Qwj = −ubdj iwqj + ubqj iwdj = k pj ubdj + iwdj ubqj = k0 pj ubqj (13)
Figure 5 shows the droop curves of the reactive currents and the reactive powers with respect to
the d-axis
Figurevoltage.
5 shows It the
candroop
be observed
curvesin ofFigure 5a thatcurrents
the reactive a large and
proportional
the reactive coefficient
powers is ablerespect
with to result
to
in a large shared reactive current. In Figure 5b, the sharing coefficient is given
the d-axis voltage. It can be observed in Figure 5a that a large proportional coefficient is able to resultby Equation (13). Since
thea voltage
in vectorreactive
large shared and thecurrent.
d-axis are not strictly
In Figure 5b, thecoincident, the reactive
sharing coefficient is power
given by is inevitably
Equation (13).related to
Since
active current
the voltage vector andiwdj. Even so, ifare
the d-axis a quite large kpjcoincident,
not strictly is taken, then the voltage
the reactive powervector and the d-axis
is inevitably related will
to
substantially
the active current coincide. In this
iwdj . Even so, ifpoint,
a quiteitlargeis known fromthen
kpj is taken, Equation (13) vector
the voltage that theandreactive
the d-axispower
will
contributed by
substantially the active
coincide. current
In this point,iwdjit is
is relatively
known from small. In short,
Equation (13)the
thatproportional coefficient
the reactive power of each
contributed
unitthecan
by activebe current
calculatediwdj isand designed
relatively based
small. on Equation
In short, (13). Incoefficient
the proportional practicalofapplications,
each unit canthe be
optimization
calculated andcontrol
designed andbased
the quantitative
on Equation allocation of reactive
(13). In practical power can the
applications, be achieved
optimizationconsidering
control
the unit capacity limit. For example, a large proportional coefficient can be set for the unit with a
small active power output so that it can share more reactive power.
Energies 2018, 11, 2207 8 of 19

and the quantitative allocation of reactive power can be achieved considering the unit capacity limit.
For example, a large proportional coefficient can be set for the unit with a small active power output
Energies
so that 2018, 11,share
it can x FORmore
PEER REVIEW
reactive power. 8 of 19

iwq Qw
Qw1
iwq1

iwq2 Qw2

ubqj ubq ubqj ubq

kp2 kp1 k'p2 k'p1
(a) (b)

Figure 5.
Figure 5. Reactive
Reactivecurrent
currentand
andreactive
reactivepower
power sharing relationship
sharing with
relationship thethe
with P-type droop
P-type control.
droop (a)
control.
Reactive
(a) current
Reactive sharing
current relationship;
sharing (b)(b)
relationship; reactive power
reactive sharing
power relationship.
sharing relationship.

In order to maintain the active power benefit of WFs, the active power priority principle can be
In order to maintain the active power benefit of WFs, the active power priority principle can be
adopted. As shown in Figure 3, according to the real-time active current iwd, the reactive current iwq is
adopted. As shown in Figure 3, according to the real-time active current iwd , the reactive current iwq is
limited as follows:
limited as follows: q
2
i =± ii22wmax−i 2iwd
iwqlim
wq lim w max wd
(14)
(14)
Equation (14) defines the reactive power margin. There is no steady-state equilibrium point
Equation (14) defines the reactive power margin. There is no steady-state equilibrium point of
of reactive power, assuming that the reactive power demand exceeds the reactive power margin.
reactive power, assuming that the reactive power demand exceeds the reactive power margin.
Considering that the reactive power demand can be adjusted by the centralized reactive power
Considering that the reactive power demand can be adjusted by the centralized reactive power
compensation device, and that a WF has the minimum reactive power compensation capability,
compensation device, and that a WF has the minimum reactive power compensation capability, this
this study recommends the following reactive power compensation scheme: (1) For the entire WF,
study recommends the following reactive power compensation scheme: (1) For the entire WF,
according to its rated capacity and maximum active power output to determine the minimum reactive
according to its rated capacity and maximum active power output to determine the minimum
power compensation capability. When the real-time reactive power output becomes larger than the
reactive power compensation capability. When the real-time reactive power output becomes larger
minimum compensation capacity, the centralized reactive power compensation should be adjusted
than the minimum compensation capacity, the centralized reactive power compensation should be
in time, such as the conventional filter or the high-performance static var compensator (SVC) or
adjusted in time, such as the conventional filter or the high-performance static var compensator (SVC)
STATCOM. Consequently, the reactive power demand becomes smaller, and within the minimum
or STATCOM. Consequently, the reactive power demand becomes smaller, and within the minimum
compensation capability. (2) For each unit, according to its real-time active current, adjust the limitation
compensation capability. (2) For each unit, according to its real-time active current, adjust the
of the reactive current according to Equation (14).
limitation of the reactive current according to Equation (14).
3.2. P-V Control of LCR
3.2. P-V Control of LCR
When the active power Pw from the WECS increases, it has been known that Ubm increases
When the active power Pw from the WECS increases, it has been known that Ubm increases
according to Equation (5) assuming that Prc remains constant. Actually, if Ubm increases, Prc will
according to Equation (5) assuming that Prc remains constant. Actually, if Ubm increases, Prc will
increase too. However, there is no doubt that the steady-state Ubm will become larger from the
increase too. However, there is no doubt that the steady-state Ubm will become larger from the
perspective of the whole circuit if the firing angle remains unchanged. Only if the LCR controller
perspective of the whole circuit if the firing angle remains unchanged. Only if the LCR controller
reduces the firing angle α, leading to more absorbed active power Prc by the LCR, can Ubm return back
reduces the firing angle α, leading to more absorbed active power Prc by the LCR, can Ubm return back
its reference value. A detailed analysis is performed as follows.
its reference value. A detailed analysis is performed as follows.
Based on the conclusions in [5], the time constant of the rectifier currents is quite small, about
Based on the conclusions in [5], the time constant of the rectifier currents is quite small, about
tens of milliseconds, under the constant-voltage control of the inverter of HVDC. Thus, the current
tens of milliseconds, under the constant-voltage control of the inverter of HVDC. Thus, the current
transients in Equations (7) and (10) can be ignored while analyzing the active power balance, which
transients in Equations (7) and (10) can be ignored while analyzing the active power balance, which
gives rise to:
gives rise to:
Peq = Ubm Urcm sin δ/(ω0 Lrc )
(15)
sin δ/(
ω00LLrcrc) − Urcm
2 /(ω L )
QeqP=eq

U U
bm UUrcm cos
bm rcm 0 rc
(15)
where Peq and Qeq are the active Qeq and
Ubmreactive 0 Lrcof rectifier
Urcm cos power 2
Urcm 0bridge,
Lrc  and δ is the phase-angle
difference between U b and U rc , and it can be seen as a power angle. Note that the internal resistor Rrc
where
is ignoredPeq in
and Qeq are(15).
Equation the Given
active that
andthereactive
powerpower of rectifier
factor angle of the bridge,
rectifier and δ is
bridge the phase-angle
(excluding Rrc and
difference between
Lrc ) is α, i.e.,: U b and Urc, and it can be seen as a power angle. Note that the internal resistor Rrc

is ignored in Equation (15). Given that theQpower factor angle of the rectifier bridge (excluding(16)
eq = Peq tan α
Rrc
and Lrc) is α, i.e.,:
Qeq  Peq tan  (16)

Combining Equations (9), (10), (15), and (16), it can be obtained that:
Energies 2018, 11, 2207
x FOR PEER REVIEW 9 of 19

Ubm(15),
Combining Equations (9), (10),   (16),
cos and 
 uitdr can 
 udibe obtained
id Rd  udi that: (17)

Furthermore, consideringUthe effect

bm cos ofαthe
(δ + ) =leakage
udr = uinductor
di + id Rd L
≈rc uon
di the power factor angle of (17)
the
LCR (including Trc), it exists as:
Furthermore, considering the effect of the leakage inductor Lrc on the power factor angle of the
udr  Ubm cos   id Rc (18)
LCR (including Trc ), it exists as:
where Rc is the equivalent commutation u dr = Uwith
resistor bm cos − id Rvalue
anαactual c 6/πω0Lrc. Substituting Equation (18)
(18)
into Equations
where (9)equivalent
Rc is the and (10) yields that:
commutation resistor with an actual value 6/πω 0 Lrc . Substituting
Equation (18) into Equations (9) and (10) yields that:
Ubm cos   udi  id Rc  Rd   (19)
U cos α = udi + id ( Rc + Rd ) (19)
It can be assumed that both Ubm bm and udi remain constant under the controls of both the rectifier
and inverter, and thus cos(δ + α) remains constant, according to Equation (17). Actually, when Pw
It can be assumed that both Ubm and udi remain constant under the controls of both the rectifier
increases, the power angle δ increases, whereas the firing angle α decreases along with the increase
and inverter, and thus cos(δ + α) remains constant, according to Equation (17). Actually, when Pw
of id in Equation (19). Therefore, δ + α remains approximately unchanged. Since that δ + α is the power
increases, the power angle δ increases, whereas the firing angle α decreases along with the increase of
factor angle between Ub and Irc:
id in Equation (19). Therefore, δ + α remains approximately unchanged. Since that δ + α is the power
factor angle between U b and Irc : Prc  Ubm I rcm cos    
(20)
Qrc  Ubm I rcm sin    
Prc = Ubm Ircm cos(δ + α)
(20)
Qrc = Ubm Ircm sin(δ + α)
Let
Let U
Ubb be
beoriented
orientedto tothe
thed-axis
d-axisof
ofthe
therated
ratedsynchronous
synchronous reference
reference frame
frame (RSRF)
(RSRF) (consistent
(consistent with
with
the orientation of the WECS), and the vector diagrams of the SEB is depicted
the orientation of the WECS), and the vector diagrams of the SEB is depicted in Figure 6. Inin Figure 6. Figure
In Figure
6a,
6a, the active current iwd is smaller, and it can be assumed that the reactive current Ic from the filters
the active current iwd is smaller, and it can be assumed that the reactive current Ic from the filters
can
can be supplied to
be supplied to the rectifier exactly.
the rectifier exactly. In
In Figure
Figure 6b,
6b, however,
however, iiwd becomes larger, resulting in more
wd becomes larger, resulting in more
required reactive power for the LCR. Assuming that Ic remains constant due to a time-delay to adjust
required reactive power for the LCR. Assuming that Ic remains constant due to a time-delay to adjust
the
the filters,
filters, the
the voltage
voltage amplitude
amplitude U Ubm, ,and
andthe
bm phase-angle φ of
thephase-angle ofUUb are
are controlled
controlled by
by the
b
the LCR
LCR and
and the
the
WECS respectively,
respectively, and then the WECS can output the the required
required reactive
reactive current
current automatically.
automatically. Thus,
Thus,
Icc,, together with i , is able to provide the reactive current of the LCR.
together with iwq , is able to provide the reactive current of the LCR.
wq Moreover, the power angle
increases, whereas the firing angle decreases,
decreases, and
and UUbb remains constant.

d d
Ub j0 Lrc I rc Ub
j0 Lrc I rc
Urc
Urc Irc
 iwd
Iw 
 Irc 
Iw
q Ic q Ic iwq
(a) (b)

that U
Figure 6. Vector diagrams under the condition that Ubb remains constant. (a) A smaller active current
iwd (orPPww););(b)
wd(or (b)aalarger
largeractive
activecurrent iwd
currentiwd..

After
After analyzing
analyzing thethe power
power characteristics
characteristics of of the
the system,
system, the
the P-V
P-V control
control loop
loop can
can be
be designed
designed as as
follows.
follows. According
According to to Equation
Equation (19),
(19), there is aa linear
there is linear relationship
relationship between
between U Ubmcosα
cosα and
and id.. In
i In fact,
fact, α
α
bm d
changes within a narrow range of around 20°◦ [14], and therefore, it can be approximately
changes within a narrow range of around 20 [14], and therefore, it can be approximately considered considered
that
that cosα remains unchanged,
unchanged, leading
leading toto an
an approximate
approximate linear
linear relationship
relationship between
between UUbm and id.
cosα remains bm and id .
Consequently,
Consequently, aalinear
linearregulator,
regulator,such
suchasas
PI,PI,
cancanbe be employed
employed to control
to control UbmU bm, as shown in Figure 7,
, as shown in Figure 7, and
and its output is the DC current reference id*. In the inner current loop, the typical regulator is applied.
its output is the DC current reference id *. In the inner current loop, the typical regulator is applied.
Note
Note that
that the
the optional
optional compensation
compensation can can bebe performed
performed in in the
the outer
outer loop,
loop, so
so as
as to obtain the
to obtain same
the same
control gain at different operating points.
control gain at different operating points.
Energies 2018, 11, 2207 10 of 19
Energies 2018, 11, x FOR PEER REVIEW 10 of 19

 compensation
U bm id 
+– PI 5 +– PI 6 –


+
U bm cos 
id 

Figure 7. Proposed power based voltage (P-V) control for LCR.

4. Stability Analysis and Parameter Design

4.1. Stability Analysis of Q-f Control

While analyzing
analyzingthe thestability
stabilityofofthe Q-f
the Q-fcontrol
control of of
thethe
WECS
WECS subsystem,
subsystem, the the
voltage amplitude
voltage amplitude Ubm
can
U be assumed
bm can be assumed to betoconstant, andand
be constant, bothboth
the the
active current
active iwd and
current thethe
iwd and reactive power
reactive powerQrcQare seen
rc are as
seen
external disturbances. Moreover, the current dynamics can be neglected since
as external disturbances. Moreover, the current dynamics can be neglected since they are generally they are generally much
faster faster
much than those
than of power.
those Considering
of power. only the
Considering only outer loop of
the outer theof
loop Q-fthecontrol, it canitbe
Q-f control, obtained
can that:
be obtained
that:

iwq = k p3 + k i3 /s (0 − Ubm sin φ) (21)
 
iwq  kp 3  ki 3 s  0  Ubm sin   (21)
Let iwq = i0 wq + kp3 (0 − Ubm sin φ), and rewrite Equations (6) and (21) as:
Let iwq = i'wq + kp3(0 − Ubmsin  ), and rewrite Equations (6) and (21) as:
dφ /dt = ωb Ubm i0 wq − k p3 Ubm 


sin φ cos φ
d dt  − Ubbm iwd 
U i   k U sin  cos
wq φ +p 3Qrc
bm sin ]/ C f Ubm  − ωb ω0
bm 2  (22)
(22)
Ubm iwd sin   Qrc  C f Ubm
0
di wq /dt = −k i3 Ubm sin φ
2
  b0  (23)

From Equations (22) and (23), it candibe  observed

wq
dt  ki 3U thatsin
bm
 is a coupling relationship between(23)
there the
0
state variables φ and i wq , which determines the reactive power-frequency dynamic characteristics.
The phase-plane
From Equations (22) analysis [32] it
and (23), can can bebe performed
observed to that demonstrate the stability
there is a coupling of the simplified
relationship between
the state variables  and i′wq, which determines the reactive power-frequency dynamic
second-order nonlinear system consisting of Equations (22) and (23). Clearly, the equilibrium point of
the system is:
characteristics. (
The phase-plane analysis [32] φcan = be0 performed to demonstrate the stability of the simplified
0 2
 (24)
second-order nonlinear system consisting 0 C f Ubm − (22)
i wq = ofωEquations Qrc and /Ubm (23). Clearly, the equilibrium point
of the system is:
Let a = ω b (kp3 Ubm + iwd )/(Cf Ubm ), and b = ki3 ω b /Cf . The linearized system at Equation (24) is ∆ẋ
= A∆x with ∆x = [∆φ, ∆i0 wq ] and:   0

 "

 wq 
i   0C f Ubm
−a
2
 Qrc Ubm # 
b/(k i3 Ubm )
(24)

A= (25)
Let a = ωb(kp3Ubm + iwd)/(CfUbm), and b = − ki3kωi3bU
/Cbm 0
f. The linearized system at Equation (24) is Δẋ = AΔx
with Δx = [Δ  , Δi'wq] and:
The characteristic equation is λ2 + aλ + b = 0. Clearly, the system is small-signal stable, since
a > 0 and b > 0. If a2 − 4b < 0, Equation (24)
A
 will  ki 3Ufocus,
a be a bstable bm 
 otherwise it is a stable node. In the
 (25)
former case, the motion near the equilibrium   ki 3point
Ubm will 0converge  in the form of oscillations. While
in the latter case, there is no oscillation during the convergence and an asymptote exists around the
The characteristic
equilibrium point: equation is λ2 + aλ + b = 0. Clearly, the system is small-signal stable, since a > 0
and b > 0. If a − 4b < 0, Equation (24) will
2
∆i0bewq a=stable−k i3 Ufocus, ∆φ
bm /λ1otherwise it is a stable node. In the former
(26)
case, the motion near the equilibrium point will converge in the form of oscillations. While in the
where λ1 is the eigenvalue with a smaller modulus.
latter case, there is no oscillation during the convergence and an asymptote exists around the
Taking a concrete case as an example: Ubm = 1.0 pu, iwd = 0.8 pu, Qrc = Cf = 0.21 pu in an initial
equilibrium point:
state, and they remain unchanged in the following convergence process. Also, set kp3 = 0.6, ki3 =
50. Since that the reactive power from Cfiwq is exactly 1  to Qrc , since Qrc = Cf , Qw = 0, and(26)
 ki 3Ubmsupplied the
equilibrium point is the origin. As shown in Figure 8, the equilibrium points of Equation (22) and
where
those ofλ1Equation
is the eigenvalue
(23) formwith a smaller
the two modulus.
equilibrium curves respectively in the φ − i0 wq phase plane, and
Taking a concrete case as an example: Ubm = 1.0 pu, iwd = 0.8 pu, Qrc = Cf = 0.21 pu in an initial state,
and they remain unchanged in the following convergence process. Also, set kp3 = 0.6, ki3 = 50. Since
that the reactive power from Cf is exactly supplied to Qrc, since Qrc = Cf, Qw = 0, and the equilibrium
point is the origin. As shown in Figure 8, the equilibrium points of Equation (22) and those of
Energies 2018, 11, x FOR PEER REVIEW 11 of 19
Energies 2018, 11, 2207 11 of 19
Equation (23) form the two equilibrium curves respectively in the ϕ − i′wq phase plane, and the
intersection, i.e., the origin O, is the equilibrium point of the system. In Figure 8, the range of the state
the intersection, i.e., the origin O, is the equilibrium point of the system. In Figure 8, the range of the
variable i′wq is [i′0 wqmin, i′0 wqmax]. 0
state variable i wq is [i wqmin , i wqmax ].
 max  iwq max  kp 3Ubm sin   ip
iwq 2
w max
 iwd  kp 3Ubm sin 
i0 2
wqmax = iwqmax + k p3 Ubm sin φ = piwmax − iwd + k p3 Ubm sin φ (27)
(27)
 min
i0 wqmin
iwq = iwqmin
iwq min+kkp3p 3U
Ubm sin    i 2 i2wmax
bm sin φ = − w max
 iwd
− kp+
iwd Ubm sin φ
U p3sin
3 kbm

where
where iwmax
wmaxisisthe
the maximum current of
maximum current ofthe
theWECS.
WECS.Moreover,
Moreover,thethe phase-angle
phase-angle  should
φ should be in be
theinrange
the
( − π/2, π/2) so as to ensure the negative feedback property of the controller.
range (−π/2, π/2) so as to ensure the negative feedback property of the controller.

2.0
1.5 di'wq/dt = 0
P2
1.0 i'wqmax asymptote di'wq/dt > 0
0.5 P3 di'wq/dt < 0
i'wq (pu)

P1
0.0
O dϕ/dt = 0
ϕmin = –π/2

-0.5

ϕmax = π/2
-1.0 i'wqmin dϕ/dt > 0
-1.5 dϕ/dt < 0
-2.0
-2.0 -1.5 -1.0 -0.5 0.0 0.5 1.0 1.5 2.0
ϕ (rad)

Figure 8.
8. Phase
Phase plane
plane φ
ϕ−− i′iwq
0 with the Q-f control.
Figure wq with the Q-f control.

In Figure 8, the domain (−π/2, π/2) × [i′wqmin0, i′wqmax] 0is divided into four sections. In each section,
In Figure 8, the domain (−π/2, π/2) × [i , i wqmax ] is divided into four sections. In each
the horizontal and vertical arrows indicate thewqmin directions of motions of Equations (22) and (23)
section, the horizontal and vertical arrows indicate the directions of motions of Equations (22) and (23)
respectively, and the actual direction of the state trajectory is the synthesis direction. Taking the initial
respectively, and the actual direction of the state trajectory is the synthesis direction. Taking the initial
point P1 as an example, as shown by the black arrow, the trajectory traverses the blue curve vertically
point P1 as an example, as shown by the black arrow, the trajectory traverses the blue curve vertically
and then enters the section where P2 is located. Then, it moves to the lower right, and traverses the
and then enters the section where P2 is located. Then, it moves to the lower right, and traverses the
red curve in a vertical direction. Thereafter, the state trajectory will move along the asymptote to the
red curve in a vertical direction. Thereafter, the state trajectory will move along the asymptote to the
steady state point. Similar cases occur in other sections. Therefore, it can be concluded that the system
steady state point. Similar cases occur in other sections. Therefore, it can be concluded that the system
is locally asymptotically stable in the domain.
is locally asymptotically stable in the domain.
It should also be assumed that the initial point is I with the position (−0.1, 0.1), and the
It should also be assumed that the initial point is I with the position (−0.1, 0.1), and the
convergence process of the second-order simplified system is illustrated in Figure 9. From Figure 9a,
convergence process of the second-order simplified system is illustrated in Figure 9. From Figure 9a,
it can be observed that the system converges along the asymptote, since the equilibrium point is a
it can be observed that the system converges along the asymptote, since the equilibrium point is
stable node. In Figure 9b, the equilibrium point become a stable focus due to a large ki3. Thus, although
a stable node. In Figure 9b, the equilibrium point become a stable focus due to a large ki3 . Thus,
the convergence speed become large, there is an overshoot in the motion near the stable focus.
although the convergence speed become large, there is an overshoot in the motion near the stable
Therefore, it can be concluded that a large ki3 can facilitate the system converge speed, causing both
focus. Therefore, it can be concluded that a large ki3 can facilitate the system converge speed, causing
overshoot and oscillation. A trade-off should be considered to design an appropriate value for the
both overshoot and oscillation. A trade-off should be considered to design an appropriate value for
control parameter ki3. Similar work can be made to further study the effects of other parameters on
the control parameter ki3 . Similar work can be made to further study the effects of other parameters on
the system dynamic behaviors.
the system dynamic behaviors.
It is noteworthy that, when i0 wq reaches the limitation i0 wqmin or i0 wqmax , the system state moves
along the boundaries and it can still converge once the state reaches another section across the blue
curve as long as the equilibrium point is in the allowable domain. In Equation (27), the range of i0 wq
is related to iwd , i.e., the active power Pw . While in Equation (24), the equilibrium point is related
to the required reactive power Qw . The different equilibrium curves (or points) and domains under
different Pw and Qw are depicted in Figure 10. In particular, the equilibrium point P1 (P1 0 ) are beyond
the domain, when both Pw and Qw are quite large, such as 1.0 and 0.9 (−0.9). The problem must be
avoided in practice. To this end, a proper capacity of filters could be designed to result in a smaller Qw .
 0.1005

asympt
i'wq (pu)
6 di'wq/dt = 0 6
0.1000
4 4 0.0995
0 5 10 10 -4
2 2
dϕ/dt = 0
O
0 0
-10 -5 0 5 ×0.01 0 0.05 0.10 0.15
0
Energies 2018, 11, 2207 ϕ (rad) t (s) 12 of 19
Energies 2018, 11, x FOR PEER REVIEW
0.05 12 of 19

0.10
×0.01 ×0.01 0.1015
10
0.15 tI(s) 10
(a) 0.1010

asymptote
8 8
×0.01 ×0.01 0.1005

i'wq (pu)i'wq (pu)

6 di'wq/dt = 0
12 612
10 10 0.1000
4 I 48 0.0995
8
0 5 10 10-4
6
2 26
4 di' wq/dt
dϕ/dt ==0 0
O 4
0
2 dϕ/dt = 0 02
0 -10 O
-5 0 5 ×0.01 0 0.05 0.10 0.15
0
-2 ϕ (rad)
-2 t (s)
0.05 -10 -6 -2 2 6 ×0.01 0 2 4 6 8 10
0 t (ms)
0.10
ϕ (rad) ×0.001
2 5
4
4
0.15 t (s) (a)3
6 2
×0.01 1 ×0.01
8 0
12 12-1
10 t (ms)
10 10 -4 -2 0 2 ×0.001
I
8 (b)8
i'wq (pu)

6 6
4 di'wq/dt = 0 4
Figure 9. Convergence process of the
2 dϕ/dt = 0second-order simplified
2 system composed of Equations (22) and
0 O 0
(23), from the initial point I. (a) kp3 = 0.6, ki3 = 50; (b) kp3 = 0.3, ki3 = 2000.
-2 -2

-10 -6 -2 2 6 ×0.01 0 2 4 6 8 10
0 t (ms)
It is noteworthy that, when i′wq reaches the limitation ϕ (rad) ×0.001i′wqmin or i′wqmax , the system state moves along
2 5
the boundaries and it can still4 converge once the state 4 reaches another section across the blue curve
3
as long as the equilibrium point 6 is in the allowable domain. 2
1
In Equation (27), the range of i′wq is related
8 0
to iwd, i.e., the active power Pw. While t (ms)
in Equation (24),-1 the equilibrium point is related to the required
10
-4 -2 0 2 ×0.001
reactive power Qw. The different equilibrium curves (b) (or points) and domains under different Pw and
Qw are depicted in Figure 10. In particular, the equilibrium point P1 (P1′) are beyond the domain, when
Figure 9. Convergence process
process of of the
thesecond-order
second-order simplified
simplified systemsystem composed
composed of of Equations
Equations (22)
(22) and
Pw and9.QConvergence
both Figure w are quite large, such as 1.0 and 0.9 (−0.9). The problem must be avoided in practice. To
and
(23),
(23), from
from the
the initial
initial point
point I.
I. (a)
(a) kkp3 == 0.6,
0.6,ki3 = =
k 50; (b)
50; (b)k p3
k = 0.3,
= k
0.3, =
i3 k 2000.
= 2000.
this end, a proper capacity of filters p3 could bei3designed top3result ini3a smaller Qw.
It is noteworthy that,
1.5
when i′wq reaches the limitation i′wqmin or i′wqmax, the system state moves along
the boundaries and it can still converge i'wqmax
once the state reaches another section across the blue curve
as long as the equilibrium 1.0 point is in the allowable domain. P1 In Equation (27), the range of i′wq is related
to iwd, i.e., the active power
0.5 Pw . While in Equation (24), the equilibrium point is related to the required
reactive power Qw. The different equilibrium curves (or points) and domains under different Pw and
i'wq (pu)

0.0
Qw are depicted in Figure 10. In particular, the equilibrium point P1 (P1′) are beyond the domain, when
ϕmax = π/2

both Pw and Qw are quite -0.5 large, such as 1.0 and 0.9 (−0.9). The problem must be avoided in practice. To
ϕmin = –π/2

this end, a proper capacity of filters could be designed P1'to result in a smaller Qw.
-1.0 i'wqmin
1.5
-1.5
-2 -1.5
i'wqmax -1 -0.5 0 0.5 1 1.5 2
1.0 ϕ (rad)
P1
Figure 10. Different equilibrium curves and domains under different Pw and Qw.
Figure 10.0.5Different equilibrium curves and domains under different Pw and Qw .
i'wq (pu)

4.2. Stability Analysis of0.0

P-V Control
ϕmax = π/2

4.2. Stability Analysis of P-V Control

-0.5
A simplified second-order equation of the HVDC subsystem can be obtained when the current
ϕmin = –π/2

A simplified second-order equation of the HVDC subsystem can be obtained when the current
transients are ignored.-1.0 P1'
i'wqmin
transients are ignored.
  

∗
-1.5 id i≈  k p5
kp 5 + kkii55 /s
s (U Ubm−UU bm ) (28)
(28)
d bm bm
-2 -1.5 -1 -0.5
 ϕ (rad) 0 0.5 1 1.5 2
2
dUbm /dt ≈ 2ωb Pw − udi id − Rd i2d /C f (29)
Figure 10. Different equilibrium curves and domains under different Pw and Qw.
According to Equations (28) and (29), when Pw increases, the voltage amplitude Ubm will increase,
andStability
4.2. thus theAnalysis
control of
loop
P-VEquation
Control (28) will adjust the DC current reference. Once the firing angle
regulated by the inner loop decreases, the actual DC current id will increase. According to Equation (29),
A simplified second-order equation of the HVDC subsystem can be obtained when the current
Ubm will stop increasing and start to decrease, and finally the system will reach a new steady state.
transients are ignored.

 
id  kp 5  ki 5 s Ubm  Ubm   (28)
 According to
According to Equations
Equations (28)
(28) and
and (29),
(29), when
when PPww increases,
increases, the
the voltage
voltage amplitude
amplitude U Ubm will increase,
bm will increase,
and thus the control loop Equation (28) will adjust the DC current reference. Once
and thus the control loop Equation (28) will adjust the DC current reference. Once the firing angle the firing angle
regulated by
regulated by the
the inner
inner loop
loop decreases,
decreases, the
the actual
actual DC
DC current
current iidd will
will increase.
increase. According
According to to Equation
Equation
(29), U
(29), Ubm will stop
bm will stop increasing
increasing and
and start
start to
to decrease,
decrease, and
and finally
finally the
the system
system will
will reach
reach aa new
new steady
steady
Energies 2018, 11, 2207 13 of 19
state.
state.

4.3. Small-Signal
4.3. Small-Signal Analysis and and Parameters DesignDesign
4.3. Small-Signal Analysis Analysis and Parameters
Parameters Design
Taking one
Taking one of of typical
typical operating
operating points,
points, i.e.,
i.e., PPw = 0.8 pupu and Q Qrc = 0.21 pu; pu; as an an example, the the
Taking one of typical operating points, i.e., Pww == 0.8
0.8 pu and
and Qrcrc == 0.21
0.21 pu; as as an example,
example, the
small-signal model
small-signal model of of the
the overall
overall system
system cancan be
be established
established in in the
the RSEF,
RSEF, as as shown
shown in in Figure
Figure 11.11. For
For
small-signal model of the overall system can be established in the RSEF, as shown in Figure 11. For the
the inner
the inner loop regulator,
regulator, PI PI22 and
and PIPI4,, of
of the
the WECS,
WECS, the the parameters
parameters can can be set as as kkp2p2 (k
(kp4)) is
is 1.0,
1.0, and kki2i2
inner looploop regulator, PI2 and PI4 , of 4the WECS, the parameters can be setbeassetkp2 (kp4 ) isp41.0, and kand
i2 (ki4 )
(ki4i4)) is
(k is 10,
10, leading
leading to to aa closed-loop
closed-loop bandwidth
bandwidth about
about 167 167 Hz.
Hz. Moreover,
Moreover, typical
typical parameters such such
is 10, leading to a closed-loop bandwidth about 167 Hz. Moreover, typical parametersparameters such as kp1 = 4.0,
as
as k = 4.0, k = 50 can be set for the outer loop active power regulator PI , leading to a closed-loop
ki1 k=p150 = 4.0, ki1 =set50for
canthebeouter
set for theactive
outerpower
loop active power
PI1 ,regulator
leading toPIa1,closed-loop
leading to abandwidthclosed-loop
p1 i1 1
can be loop regulator of
bandwidth of
bandwidth of about
about 20 20 Hz.
Hz.
about 20 Hz.

LinearizingU
U bm
uudc
dc + Linearizing bm
PI55 ––+ PI
PI66
PI11
PI ++–– PI22
PI ++–– u Linearizing urcd
rcd Linearizing urcd
Eq.(4) PI
Eq.(1) iiwd uubd iircd

Eq.(10)
iidd
Eq.(4) +
u wd 

Eq.(10)
LLww wd Eq.(1) wd
Eq.(3)
bd
Eq.(7) Eq.(8) and
and u
LLww Eq.(2) iiwq
uuwq Eq.(2) Eq.(3) uubq Eq.(7) iircq Eq.(8) udrdr
uurcq
wq bq rcq
PI33 PI44 + wq (9)
(9)
uubq PI ++–– PI +++ rcq
 +
bq –1
–1
WECS
WECS
VSI controller
VSI controller SEB model
SEB model LCRcontroller
LCR controller
model
model
Figure
Figure 11.
Figure 11. Small-signal
11. Small-signal model
Small-signal model of
model of the
of the system.
the system.
system.

In particular,
In particular, attention
attention should
should be be paid
paid to
to the
the parameters
parameters of of the
the developed
developed outer
outer loop
loop reactive
reactive
In particular, attention should be paid to the parameters of the developed outer loop reactive
power control. As shown in Figure 12a, when k becomes larger, the
power control. As shown in Figure 12a, when ki3 becomes larger, the damping of PI3 increases.
i3 damping of PI 3 increases.
power control. As shown in Figure 12a, when ki3 becomes larger, the damping of PI3 increases.
However, the
However, the loop
loop will
will couple
couple withwith the
the DC
DC voltage
voltage when
when kki3i3 is is too
too large,
large, resulting
resulting in
in aa pair
pair of
of
However, the loop will couple with the DC voltage when ki3 is too large, resulting in a pair of conjugate
conjugate modes.
conjugate modes. ItIt can
can be
be seen
seen from
from Figure
Figure 12b
12b that
that when
when kkp3p3 increases,
increases, the
the damping
damping of of PI
PI33 decreases,
decreases,
modes. It can be seen from Figure 12b that when kp3 increases, the damping of PI3 decreases, which
which thereafter leads to the coupling. However, the coupling disappears as k continues
which thereafter leads to the coupling. However, the coupling disappears as kp3 continues to increase.
p3 to increase.
thereafter leads to the coupling. However, the coupling disappears as kp3 continues to increase. Finally,
p3  [1.5, 2.0], and ki3 
Finally, kp3  [1.5, 2.0], and ki3  [20, 140] can be taken. Within the tuned parameter ranges, the closed-
Finally, k [20, 140] can be taken. Within the tuned parameter ranges, the closed-
kp3 ∈ [1.5, 2.0], and ki3 ∈ [20, 140] can be taken. Within the tuned parameter ranges, the closed-loop
loop bandwidth
loop bandwidth is is about
about 4040 Hz.
Hz.
bandwidth is about 40 Hz.

10
10
PI3+Δu +Δudcdc PI5,6
axis

PI PI
Imag.axis

5,6
55 Δudcdc3
Δu PI33
PI PI2,4
PI 2,4
00
(a) Imag.

PI11
PI
-5
-5 PI5,6
PI 5,6
-10
-10
(a)

-80
-80 -60
-60 -40
-40 -20
-20 00
Real axis
Real axis
10
10 PI33+Δu
+Δudcdc
axis

PI
Imag.axis

PI5,6
PI 5,6
55 PI33
PI Δudcdc
Δu PI33
PI PI
PI2,4
(b) Imag.

2,4
00
PI11
PI
-5
-5 PI5,6
PI 5,6
(b)

-10
-10
-120
-120 -100
-100 -80
-80 -60
-60 -40
-40 -20
-20 00
Real axis
Real axis

Figure 12.
Figure 12. Root loci
loci of the
the small-signal
small-signal model.
model. (a)
(a) kkp3p3 == 1.3
1.3 and kki3i3 changes
changes from 50
50 to
to 140;
140; (b)
(b) kki3i3 == 130
130
Figure 12. Root
Root loci of
of the small-signal model. (a) kp3 = 1.3 and and ki3 changes from
from 50 to 140; (b) ki3 = 130
and
and k changes from 0.3 to 2.0.
and kkp3 changes
changesfrom
from0.30.3to
to2.0.
p3
p3 2.0.

A similar
A similar analysis
analysis cancan be
be performed
performed to to tune
tune the
the parameters
parameters of of the
the rectifier
rectifier P-V P-V controller.
controller. In In
A similar
Figure 13a,
13a, when analysis can
when kki5i5 becomes be
becomes larger, performed
larger, the to
the oscillationtune the
oscillation mode
mode inparameters
in PI
PI5,6 of the
disappears, butrectifier
but the P-V
the damping controller.
damping of of PI
PI66
Figure 5,6 disappears,
In Figure
tends to 13a, when
decrease andki5 the
becomes
damping larger, the oscillation
characteristics of mode
both in PI
the disappears,
voltage
5,6 and current but the of damping
the SEB of
are
tends to decrease and the damping characteristics of both the voltage and current of the SEB are
PI6 tends to decrease and thethe
damping
larger kkcharacteristics of the
bothdamping
the voltage andofcurrent is.of the SEBkp5are 
deteriorated.
deteriorated. In Figure
In Figure 13b,
13b, the larger is, the
p5 is,
p5 the smaller
smaller the damping factor
factor of PI PI5,6
5,6 is. Finally,
Finally, kp5 
deteriorated.
[0.2, 0.5], In Figure
 [40,
0.5], kki5i5  [40, 100] 13b,
100] can
can bethe larger
be selected. k
selected. Noteis, the
Note that
p5 smaller
that typical the damping
typical parameters
parameters such factor
such as of PI
as kkp6p6 == 5,6 is.
2.0 and Finally, k
and kki6i6 == 20
20 are
p5 ∈
are
[0.2, 2.0
[0.2, 0.5], ki5 ∈ [40, 100] can be selected. Note that typical parameters such as kp6 = 2.0 and ki6 = 20 are
employed in the inner loop. Within the suggested parameter ranges, the outer and inner closed-loop
bandwidths are approximately 5 Hz and 110 Hz respectively.
Energies
Energies2018,
2018,11,
11,xxFOR
FORPEER
PEERREVIEW
REVIEW 14
14of
of19
19

employed
employed
Energies 2018,in11,the
in the inner
2207innerloop.
loop.Within
Withinthe
thesuggested
suggestedparameter
parameterranges,
ranges,the
theouter
outerand
andinner
innerclosed-loop
closed-loop
14 of 19
bandwidths are approximately 5 Hz and 110 Hz respectively.
bandwidths are approximately 5 Hz and 110 Hz respectively.
10
10 PI
PI5,65,6

axis
5

Imag.axis
5 Δi
Δid+Δudc PI3 Δu
PI3 Δudcdc PI PI6 6PI
PI1 1
d+Δudc
0

(a) Imag.
0
Δi +Δubd Δi PI
Δiwqwq+Δu
2000
-5 2000
1000
Δircd
rcd+Δubd +Δubqbq PI2,42,4
-5 1000
-10
0
0
Δi
Δi
wd Δi

(a)
-10 -1000 wd Δid
d
-2000 Δi wd+PI5
-1000
-2000 Δiwd+PI5
-1200 -1000 -800 -600 -400 -200
-1200 -1000 -800 -600 -400 -200
-150 -100 -50 0
-150 -100 -50 0
Real
Realaxis
axis
10
10
(b) Imag. axis
Imag.axis 5
5
PI
PI5,65,6
0
0
PIPI2,42,4
PI
PI1 1 PI
-5 PI5,6
-5 5,6
(b)

-10
-10
-20 -15 -10 -5 0
-20 -15 -10 -5 0
Real
Realaxis
axis

Figure
Figure13.
Figure 13.Root
13. Rootloci
Root lociof
loci ofthe
of thesmall-signal
the small-signal model.
small-signalmodel. (a)
(a)kkp5
model.(a) kp5p5===0.4
0.4and
0.4 andkkki5i5i5changes
and changesfrom
changes from50
from 50to
50 to1000;
to 1000; (b)
(b)kkki5
1000;(b) i5 =
=50
i5 = 50
50
and
and k
and kkp5
p5 changes from
changes from
p5 changes
0.1
from 0.1 to
0.1 to 2.0.
to 2.0.
2.0.

5.
5. Simulation
Simulation Results
Results
Simulations
Simulations are arecarried
are carriedout
carried outon
out onPSCAD/EMTDC
on PSCAD/EMTDC
PSCAD/EMTDC to
to verify
verify the
the proposed
to verify proposed
the proposed control
control scheme.
scheme.
control The
The
scheme.
simulated
simulated system
system is
isshown
shown in
inFigure
Figure 14,
14, and
and the
the detailed
detailed parameters
parameters of
of the
the system
system
The simulated system is shown in Figure 14, and the detailed parameters of the system are listed are
are listed
listed in
in Table
Table 1.1.
The
The
in employed
employed
Table monopole LCC-HVDC
monopole monopole
1. The employed LCC-HVDC model is from
model is from
LCC-HVDC modelthe CIGRE
theis CIGRE
from thebenchmark
benchmark model
CIGRE benchmarkmodel [33],[33], both
model the
both[33],
the
rectifier
both theand
rectifier and inverter
andof
inverter
rectifier of which
whichare
inverter LCC-based.
arewhich
of LCC-based. The
Therated
are LCC-based.ratedcapability
rated of
capability
The ofthe
thesystem
system
capability isis1000
of the 1000
systemMVA,
MVA, and
and
is 1000
the
thereactive
MVA,reactive power
power
and the capability
reactive powerof
capability each
eachset
setof
ofcapability ofAC
of AC
each11/13
set order
11/13 ACharmonic
order
of harmonic
11/13 orderfilters
filters isis50
harmonic 50MVar.
MVar.
filtersThe
The increase
is 50increase
MVar.
and
The decrease
and increase of
decreaseand the
of the DC-side
DC-side
decrease current
of current of the WECS
of thecurrent
the DC-side WECS of can
can simulate
thesimulate
WECS canthe changes
the changes
simulateof of the
thethe input power
input power
changes of theof ofthe
the
input
back end.
back end.
power of the back end.

Wind
WindFarm
Farm 0.5968H0.5968H
0.5968H0.5968H
35kV
35kV 35:213.5
35:213.5 2.5Ω
2.5Ω 2.5Ω
2.5Ω 209.2:230
209.2:230
YY ∆∆ ∆∆ YY
26.0uF

500kV
26.0uF

500kV

YY YY YY YY
11/13th
11/13th 11/13th
11/13th

Figure
Figure14.
14.Simulated
Simulatedsystem.
system.
Figure 14. Simulated system.

Table
Table1.
1.Parameters
Parametersof
ofthe
thesimulated
simulatedsystem.
system.
Table 1. Parameters of the simulated system.
CCdcdc 90,000
90,000uFuFfor
for1.5
1.5MVA
MVAcapacity
capacity
Wind Energy Conversion System Cdc Rw 90,000 uF for 0.001 1.5 MVApu capacity
Wind Energy Conversion System
Wind Energy Conversion System
R w 0.001 pu
Rw 0.001 pu
Lw 0.3 pu
Lw Lw 0.30.3
pu pu
Sending-End
Sending-End Bus
Bus C
C f 0.05 pu for each set of
offilter
Sending-End Bus Cf f
0.05 pu for each set set
0.05 pu for each filter
of filter
RRrcrc 0.001 pu
0.001 pu
Rrc Lrc 0.001
0.18 pupu
High Voltage
High Direct
Voltage Current
Direct System
Current L rc
L rc 0.18
0.18 pu pu
High Voltage Direct Current System Ld 1.1936
System Ld Ld 1.1936
1.1936 HH H
Rd RdR d 5 Ω5 Ω
5 Ω

5.1.
5.1.System
5.1. System Startup
Startup
The
The WFWF is is equivalent
equivalent toto aa single
single WECS,
WECS,and andthe
thesystem
systemstartup
startup process
processisis shown
shownin
shown in Figure
Figure 15.
15.
Before
Before the
Beforethe system
thesystem
system startup,
startup, the
thethe
startup, capacitor
capacitor of
of the
capacitor the
ofWECS WECS is
is charged
the WECS charged by the configured
by the configured
is charged battery,
battery, and
by the configured and
thereafter
battery, and
thereafter
the the
thethree-phase
three-phase
thereafter AC voltageAC
three-phase voltage
of the
AC of
ofthe
SEB can
voltage be SEB
the SEBcancanbe
generated generated
by
be the WECSby
generated atthe
by WECS
0–0.3
the s, as at
WECS at0–0.3
0–0.3ins,s,Figure
shown as
asshown
15a.
shown
At 0.3 s, the DC-side current of the WECS starts to increase, and then the HVDC is unblocked, with the
Energies 2018, 11, x FOR PEER REVIEW 15 of 19
Energies 2018, 11, 2207 15 of 19

in Figure 15a. At 0.3 s, the DC-side current of the WECS starts to increase, and then the HVDC is
unblocked,
P-V controlwith
in thethe P-V control
rectifier in the
whereas therectifier whereas the
constant-voltage constant-voltage
control control
in the inverter. in the inverter.
Henceforth, the DC
Henceforth,
voltage of HVDC is generated gradually, as shown in Figure 15g. At 0.4 s, the batteryAt
the DC voltage of HVDC is generated gradually, as shown in Figure 15g. 0.4 s, the
configured
battery configured at the DC bus of the WECS is withdrew, and the Q-f control in the WECS is
at the DC bus of the WECS is withdrew, and the Q-f control in the WECS is switched on. Then, the
switched on. Then, the active power continues increasing until to the rated point, as shown in Figure
active power continues increasing until to the rated point, as shown in Figure 15b,f. In Figure 15a,
15b,f. In Figure 15a, it can be observed that both the system frequency ω1 and the voltage of the SEB
it can be observed that both the system frequency ω 1 and the voltage of the SEB Ubm remain stable
Ubm remain stable and are maintained at 1.0 pu in the final steady state. The active power generated
and are maintained at 1.0 pu in the final steady state. The active power generated from the WECS
from the WECS can be delivered into the receiving end grid through the LCC-HVDC transmission.
can be delivered into the receiving end grid through the LCC-HVDC transmission. The AC filters are
The AC filters are connected to the SEB gradually when the reactive power from the WECS Qw is
connected to the SEB gradually when the reactive power from the WECS Qw is large than 0.1 pu (see
large than 0.1 pu (see Figure 15c), which can be regarded as the reactive power limit of the WECS.
Figure 15c), which can be regarded as the reactive power limit of the WECS. The reactive current iwq
The reactive current iwq of the WECS is regulated automatically, so as to compensate the reactive
of the WECS is regulated automatically, so as to compensate the reactive power for the LCR under
power for the LCR under the condition that ubq = 0, as shown in Figure 15d,f. Moreover, Figure 15e
the condition that ubq = 0, as shown in Figure 15d,f. Moreover, Figure 15e indicates that the DC-side
indicates that the DC-side voltage of the WECS remains stable under the active power control of the
voltage of the WECS remains stable under the active power control of the WECS after the battery
WECS after the battery is withdrawn.
is withdrawn.

1.2
(f) WECS (e) WECS dc-side (d) SEB voltage (pu) (c) Reactive power (pu) (b) Active power (pu) (a) Voltage and

Ub m
freq. (pu)

1.1 ω1
1
0.9
0.8
0 0.5 1 1.5 2 2.5 3
1
0.8
0.6
Pw
0.4 Prc
0.2 Piv
0
0 0.5 1 1.5 2 2.5 3
0.6

0.4 Qw
Qfilter
0.2 Add a set of filter once Qw > 0.1 pu Qrc

0 0.5 1 1.5 2 2.5 3

1.5

1
ubd
0.5
ubq
0

-0.5
0 0.5 1 1.5 2 2.5 3
1.2
1.1
voltage (pu)

1
0.9
Battery is withdrew
0.8
0 0.5 1 1.5 2 2.5 3
0.9
current (pu)

0.5 iwd
iwq
0.1

-0.3
0 0.5 1 1.5 2 2.5 3
(g) HVDC voltage

1
and current (pu)

0.8
0.6 udr
0.4 id
0.2 udi
0
0 0.5 1 1.5 2 2.5 3
Time (s)

Figure
Figure 15.
15. Simulation
Simulationresult
resultofofthe
thesystem
systemstartup. (a) (a)
startup. Sending-end busbus
Sending-end voltage andand
voltage frequency; (b)
frequency;
active power; (c) reactive power; (d) dq-axis sending-end bus voltage; (e) WECS DC-link
(b) active power; (c) reactive power; (d) dq-axis sending-end bus voltage; (e) WECS DC-link voltage; voltage; (f)
WECS
(f) WECSoutput current;
output (g) (g)
current; HVDC
HVDC DC-link voltage
DC-link andand
voltage current.
current.
 Energies 2018, 11, 2207 16 of 19
Energies 2018, 11, x FOR PEER REVIEW 16 of 19

5.2.Operation
5.2. Operationunder
underDisturbances
Disturbances
Underdisturbances,
Under disturbances,such suchasasfluctuations
fluctuationsofofactive
activepower
powerand andreactive
reactivepower,
power,the thesimulation
simulation
results are shown in Figure 16, where the WF is equivalent to a single
results are shown in Figure 16, where the WF is equivalent to a single WECS with the PI-type Q-f WECS with the PI-type Q-f
control.InInFigure
control. Figure 16,16,
at at
0.50.5 s, the
s, the DC-side
DC-side current
current of the
of the WECSWECS starts
starts to decrease,
to decrease, whichwhich simulates
simulates the
the decrease of the input active power from the front end of the WECS.
decrease of the input active power from the front end of the WECS. Figure 16a shows that a small Figure 16a shows that a small
dropofofthe
drop theSEB voltageUUbmbmoccurs.
SEBvoltage occurs.Then,Then,the
thevoltage
voltagereturns
returnstotothe therated
ratedvalue
valueunderunderthe P-Vcontrol.
theP-V control.
AtAt2.0
2.0s,s,the
theDC-side
DC-sidecurrent
currentrises
risestotothe
theoriginal
originalvalue,
value,andandaacontrary
contraryphenomenon
phenomenoncan canbebeobserved
observed
in Figure 16b. It should be noticed that the AC filters cannot be removed or
in Figure 16b. It should be noticed that the AC filters cannot be removed or added due to a time delay. added due to a time delay.
Therefore, the reactive power difference between
Therefore, the reactive power difference between Qrcrc and Qfilter Q and Q can be compensated
filtercan be compensated by Qww by Q . From 44ss
. From
toto99s,s,several
severalsets
setsofoffilters
filters are
are added
added and and removed
removed intentionally,
intentionally,in inorder
ordertotoverify
verifythetheperformance
performanceof
the Q-f control. From Figure 16d, it can be observed that the
of the Q-f control. From Figure 16d, it can be observed that the reactive current reactive current i wq automatically changes
iwq automatically
with thewith
changes demand for reactive
the demand for power
reactiveunder
power theunder
condition that ubq =that
the condition 0, guaranteeing the frequency
ubq = 0, guaranteeing the
stability. Accordingly, the reactive power differences between Q
frequency stability. Accordingly, the reactive power differences between Qrc and Qfilter can berc and Q filter can be automatically
compensatedcompensated
automatically by the WECS,by asthe
shown
WECS, in Figure
as shown 16c.in Figure 16c.

1.1
(c) Reactive power (pu) (b) Active power (pu) (a) Voltage, Freq (pu)

Ubm
1.05
ω1
1
0.95
0.9
0 2 4 6 8 10
1.1
Input active power decreases
1
0.9
0.8 Pw
0.7 Prc
0.6 Piv
Input active power increases
0.5
0 1 2 3 4 5 6 7 8 9 10
0.6
0.5
0.4 Remove two sets of
0.3 filters Qw
Add two sets of
Qfiter
0.2 filters Qrc
0.1
0
-0.1
0 1 2 3 4 5 6 7 8 9 10
0.2
(d) ubq and iwq (pu)

ubq
0.1 iwq
0
-0.1

-0.2
0 1 2 3 4 5 6 7 8 9 10
Time (s)

Figure 16. Simulation results under external disturbances, where the WF is represented by a single
Figure 16. Simulation results under external disturbances, where the WF is represented by a single
WECS with the PI-type Q-f control. (a) Sending-end bus voltage and frequency; (b) active power; (c)
WECS with the PI-type Q-f control. (a) Sending-end bus voltage and frequency; (b) active power;
reactive power; (d) dq-axis sending-end bus voltage.
(c) reactive power; (d) dq-axis sending-end bus voltage.

In order to evaluate the performance of the Q-f droop control in a multi-machine WF, the WF is
In order
represented bytotwo
evaluate
WECSs thewith
performance
a P-type Q-fof the Q-f control.
droop droop control in a multi-machine
The capabilities of the twoWF, the WF
WECSs areis
represented by two WECSs with a P-type Q-f
the same (500 MVA) but the droop coefficients are different (2.0 versus 1.0). The simulation result the
droop control. The capabilities of the two WECSs are is
same (500
shown MVA) 17.
in Figure butThe
the droop
DC-sidecoefficients
currents of arethe
different
WECSs (2.0 versus 1.0).
decrease at 0.5The simulation
s and result is
then increase at shown
2.5 s.
in Figure
From 4.5 s17. Thes,DC-side
to 7.5 several currents
sets of ACof the WECSs
filters decrease atto0.5
are connected thes and
SEBthen increase
gradually. Theat 2.5 s. From
similar 4.5 s
results
to 7.5 s, several
compared sets of16
with Figure AC filters
can are connected
be obtained. to thethere
However, SEB is
gradually. The similar
a small static error inresults compared
ubq1,2 because of
with Figure 16 can be obtained. However, there is a small static error in u
the adopted P-type droop control, as shown in Figure 17d. Moreover, from Figure 17d, it can adopted
bq1,2 because of the be also
P-type droop
observed that control, as shown
the shared in Figure
reactive current 17d.
by Moreover,
WECS1 iwq1 from Figurethat
is twice 17d,ofit can
WECS be 2also
iwq2,observed
due to the that
relationship between their droop coefficients.
Energies 2018, 11, 2207 17 of 19

the shared reactive current by WECS1 iwq1 is twice that of WECS2 iwq2 , due to the relationship between
Energies 2018, 11, x FOR PEER REVIEW 17 of 19
their droop coefficients.

1.05

(c) Reactive power (pu) (b) Active power (pu) (a) Voltage, Freq (pu)
Ubm
1.025 ω1
1
1.002
0.975 1.000
0.998
4 5 6 7 8
0.95
0 1 2 3 4 5 6 7 8 9

1
Pw1
0.8 Pw2
0.6 Input active power increases Pr c
0.4
Input active power decreases
0.2
0 1 2 3 4 5 6 7 8 9
0.7
0.6
0.5
0.4 Qw1
0.3 Qw2
0.2 Add five sets of filters Qfilter
0.1 Qr c
0
-0.1
0 1 2 3 4 5 6 7 8 9
0.15
(d) ubq and iwq (pu)

0.1 ubq1 ubq2 iwq1 iwq2

0.05
0
-0.05
-0.1
-0.15
0 1 2 3 4 5 6 7 8 9
Time (s)

17. Simulation
Figure 17.
Figure Simulationresults
resultsunder
underexternal disturbances,
external where
disturbances, the WF
where theisWFrepresented by two by
is represented WECSs
two
with thewith
WECSs theQ-f
P-type droop
P-type control.
Q-f droop (a) Sending-end
control. bus voltage
(a) Sending-end busand frequency;
voltage (b) active power;
and frequency; (c)
(b) active
reactive power; (d) dq-axis sending-end bus voltage. Note that P , P denote the
power; (c) reactive power; (d) dq-axis sending-end bus voltage. Note that Pw1, Pw2 denote the active
w1 w2 active powers
outputted
powers by both by
outputted theboth
WECSs, respectively,
the WECSs, similar for
respectively, Qw1 , for
similar Qw2Q. w1
Q,filter
Qw2denotes the reactive
. Qfilter denotes power
the reactive
generated by the filters.
power generated by the filters.

It should
should be noted
noted that the dynamic
dynamic behaviors
behaviors of of the
the system
system isis affected
affected by both the system
system
parameters
parameters and and control
controlparameters.
parameters.The Thesmall-signal
small-signalstability analysis
stability analysisperformed
performedin Section 4.3 just
in Section 4.3
illustrates the results
just illustrates of oneofofone
the results the of
typical operating
the typical points. points.
operating Similar repetitive work canwork
Similar repetitive be made
can beto
further
made tostudy thestudy
further effectthe
of effect
control
ofparameters on the system
control parameters on the stability at different
system stability operating
at different points.
operating
Furthermore, as shown
points. Furthermore, in Figure
as shown 9, the
in Figure phase-plane
9, the phase-planeanalysis, together
analysis, togetherwith
withtime-domain
time-domain statestate
trajectory based
based on
on Equations
Equations(22)
(22)and
and(23),
(23),ororEquations
Equations (28)
(28) andand (29)
(29) cancan
be be employed
employed to evaluate
to evaluate the
the effects
effects of the
of the system
system parameters
parameters andand control
control parameters
parameters on on
thethe dynamic
dynamic behaviors
behaviors of of
thethe system
system in
in practical
practical applications.
applications.

6. Conclusions
6. Conclusions
This paper
This paper proposed
proposed aa novel
novel coordinated
coordinated control
control scheme
scheme for
for WFs
WFs with
with LCC-HVDC
LCC-HVDCintegration.
integration.
The scheme comprises the Q-f control loop in the WECSs, and the P-V control
The scheme comprises the Q-f control loop in the WECSs, and the P-V control loop in the LCR. loop in the LCR.
The
Q-f Q-f control
Thecontrol maintains
maintains the system
the system frequency
frequency and compensates
and compensates forreactive
for the the reactive
powerpower
for theforLCR
the
LCR automatically,
automatically, whereas
whereas the P-V P-V control
thecontrol maintains
maintains the ACthebusAC bus voltage
voltage and realizes
and realizes thepower
the active active
power balance
balance of the of the sending-end
sending-end bus of bus of HVDC.
the the HVDC. Thus,
Thus, thethe scheme
scheme addressesboth
addresses boththe
thevoltage
voltage and
and
frequency stability, based on thethe coordination
coordination between
between thethe WF
WF and
and the
the LCR.
LCR.
The distinguishing features of the scheme can be concluded as
The distinguishing features of the scheme can be concluded as follows: follows: (1) there (1)
are no commonly
there are no
used PLLs in the controllers of WECSs, and consequently, the frequency
commonly used PLLs in the controllers of WECSs, and consequently, the frequency and synchronization stability
and
issues introducedstability
synchronization by PLLsissues
can be avoided; (2)
introduced bythe reactive
PLLs can bepower droop
avoided; (2)instead of thepower
the reactive active power
droop
instead of the active power droop is adopted while being applied to achieve synchronization control
and reactive power sharing in multi-machine systems, and therefore, the maximum power point
tracking of WFs remains unaffected; (3) the scheme can be utilized in more universal scenarios, as
long as the core topology is the VSI with LCR connection, such as WFs and photovoltaic power plants
Energies 2018, 11, 2207 18 of 19

droop is adopted while being applied to achieve synchronization control and reactive power sharing in
multi-machine systems, and therefore, the maximum power point tracking of WFs remains unaffected;
(3) the scheme can be utilized in more universal scenarios, as long as the core topology is the VSI
with LCR connection, such as WFs and photovoltaic power plants with LCC-based rectifier HVDC
integration. Our future work will focus on the control and protection algorithms during fault operation,
e.g., voltage-dependent current order limits (VDCOLs) for LCC-HVDC and low voltage ride-through
(LVRT) for WECS.

Author Contributions: Conceptualization, X.H. and H.G.; Methodology, X.H.; Validation, X.H, H.G. and G.Y.;
Writing—Original Draft Preparation, X.H.; Writing—Review & Editing, H.G., G.Y. and X.Z.; Supervision, H.G.
and X.Z.; Project Administration, G.Y.
Funding: This work was supported by the National Natural Science Foundation of China (Nos. 61722307, U1510208,
and 51711530235).
Conflicts of Interest: The authors declare no conflict of interest.

References
1. Mendoza-Vizcaino, J.; Sumper, A.; Galceran-Arellano, S. PV, Wind and storage integration on small islands
for the fulfilment of the 50-50 renewable electricity generation target. Sustainability 2017, 9, 905. [CrossRef]
2. Korkas, C.D.; Baldi, S.; Michailidis, I.; Kosmatopoulos, E.B. Occupancy-based demand response and thermal
comfort optimization in microgrids with renewable energy sources and energy storage. Appl. Energy 2016,
163, 93–104. [CrossRef]
3. Tavakoli, M.; Shokridehakia, F.; Marzband, M.; Godina, R.; Pouresmaeil, E. A two stage hierarchical control
approach for the optimal energy management in commercial building microgrids based on local wind power
and PEVs. Sustain. Cities Soc. 2018, 41, 332–340. [CrossRef]
4. Zhou, H.; Yang, G.; Wang, J. Modeling, analysis, and control for the rectifier of hybrid HVdc systems for
DFIG-based wind farms. IEEE Trans. Energy Convers. 2011, 26, 340–353. [CrossRef]
5. Brenna, M.; Foiadelli, F.; Longo, M.; Zaninelli, D. Improvement of wind energy production through HVDC
systems. Energies 2017, 10, 157. [CrossRef]
6. Smailes, M.; Ng, C.; Mckeever, P.; Shek, J.; Theotokatos, G.; Abusara, M. Hybrid, multi-megawatt HVDC
transformer topology comparison for future offshore wind farms. Energies 2017, 10, 851. [CrossRef]
7. Halawa, E.; James, G.; Shi, X.R.; Sari, N.H.; Nepal, R. The prospect for an Australian-Asian power grid:
A critical appraisal. Energies 2018, 11, 200. [CrossRef]
8. Pena, R.; Clare, J.C.; Asher, G.M. Doubly fed induction generator using back-to-back PWM converters and
its application to variable-speed wind-energy generation. IEE Proc. Electr. Power Appl. 1996, 143, 231–241.
[CrossRef]
9. Ma, S.; Geng, H.; Liu, L.; Yang, G.; Pal, B.C. Grid-Synchronization stability improvement of large scale wind
farm during severe grid fault. IEEE Trans. Power Syst. 2018, 33, 216–226. [CrossRef]
10. Zhou, H.; Yang, G.; Wang, J.; Geng, H. Control of a hybrid high-voltage DC connection for large doubly fed
induction generator-based wind farms. IET Renew. Power Gener. 2011, 5, 36–47. [CrossRef]
11. Zhou, H.; Yang, G.; Geng, H. Grid integration of DFIG-based offshore wind farms with hybrid HVDC
connection. In Proceedings of the 2008 International Conference on Electrical Machines and Systems, Wuhan,
China, 17–20 October 2008.
12. Bozhko, S.V.; Blasco-Gimnez, R.; Li, R.; Clare, J.C.; Asher, G.M. Control of offshore DFIG-based wind farm
grid with line-commutated HVDC connection. IEEE Trans. Energy Convers. 2007, 22, 71–78. [CrossRef]
13. Bozhko, S.V.; Asher, G.; Li, R.; Clare, J.; Yao, L. Large offshore DFIG-based wind farm with line-commutated
HVDC connection to the main grid: Engineering studies. IEEE Trans. Energy Convers. 2008, 23, 119–127.
[CrossRef]
14. Xiang, D.; Ran, L.; Bumby, J.R.; Tavner, P.J.; Yang, S. Coordinated control of an HVDC link and doubly fed
induction generators in a large offshore wind farm. IEEE Trans. Power Deliv. 2006, 21, 463–471. [CrossRef]
15. Li, R.; Bozhko, S.; Asher, G. Frequency control design for offshore wind farm grid with LCC-HVDC link
connection. IEEE Trans. Power Electron. 2008, 23, 1085–1092. [CrossRef]
Energies 2018, 11, 2207 19 of 19

16. Zhang, M.; Yuan, X.; Hu, J.; Wang, S.; Ma, S.; He, Q.; Yi, J. Wind power transmission through LCC-HVDC
with wind turbine inertial and primary frequency supports. In Proceedings of the 2015 IEEE Power & Energy
Society General Meeting, Denver, CO, USA, 26–30 July 2015.
17. Cardiel-Álvarez, M.Á.; Rodriguez-Amenedo, J.L.; Arnaltes, S.; Montilla-DJesus, M.E. Modeling and control
of LCC rectifiers for offshore wind farms connected by HVDC links. IEEE Trans. Energy Convers. 2017, 32,
1284–1296. [CrossRef]
18. Yin, H.; Fan, L.; Miao, Z. Coordination between DFIG-based wind farm and LCC-HVDC transmission
considering limiting factors. In Proceedings of the IEEE 2011 EnergyTech, Cleveland, OH, USA, 25–26
May 2011.
19. Yin, H.; Fan, L. Modeling and control of DFIG-based large offshore wind farm with HVDC-link integration.
In Proceedings of the 41st North American Power Symposium, Starkville, MS, USA, 4–6 October 2009.
20. Blasco-Gimenez, R.; Añó-Villalba, S.; Rodríguez-D’Derlée, J.; Morant, F.; Bernal-Perez, S. Distributed voltage
and frequency control of offshore wind farms connected with a diode-based HVdc link. IEEE Trans. Power
Electron. 2010, 25, 3095–3105. [CrossRef]
21. Yu, L.; Li, R.; Xu, L. Distributed PLL-based control of offshore wind turbine connected with diode-rectifier
based HVDC systems. IEEE Trans. Power Deliv. 2018, 33, 1328–1336. [CrossRef]
22. Cardiel-Álvarez, M.Á.; Arnaltes, S.; Rodriguez-Amenedo, J.L.; Nami, A. Decentralized control of offshore
wind farms connected to diode-based HVDC links. IEEE Trans. Energy Convers. 2018, 33, 1233–1241.
[CrossRef]
23. Xi, X.; Geng, H.; Yang, G. Enhanced model of the doubly fed induction generator-based wind farm for
small-signal stability studies of weak power system. IET Renew. Power Gener. 2014, 8, 765–774. [CrossRef]
24. Zhang, Y.; Ooi, B.T. Stand-Alone doubly-fed induction generators (DFIGs) with autonomous frequency
control. IEEE Trans. Power Deliv. 2013, 28, 752–760. [CrossRef]
25. Pena, R.; Clare, J.C.; Asher, G.M. A doubly fed induction generator using back-to-back PWM converters
supplying an isolated load from a variable speed wind turbine. IEE Proc. Electr. Power Appl. 1996, 143,
380–387. [CrossRef]
26. Abdoune, F.; Aouzellag, D.; Ghedamsi, K. Terminal voltage build-up and control of a DFIG based stand-alone
wind energy conversion system. Renew. Energy 2016, 97, 468–480. [CrossRef]
27. Li, D.; Shen, Q.; Liu, Z.; Wang, H.; Ding, M.; Liu, F. Auto-disturbance rejection control for the stator voltage
in a stand-alone DFIG-based wind energy conversion system. In Proceedings of the 2016 35th Chinese
Control Conference (CCC), Chengdu, China, 27–29 July 2016.
28. Fazeli, M.; Asher, G.; Klumpner, C.; Yao, L. Novel integration of DFIG-based wind generators within
microgrids. IEEE Trans. Energy Convers. 2011, 26, 840–850. [CrossRef]
29. Fazeli, M.; Bozhko, S.V.; Asher, G.M.; Yao, L. Voltage and frequency control of offshore DFIG-based wind
farms with line commutated HVDC connection. In Proceedings of the 2008 4th IET Conference on Power
Electronics, Machines and Drives, York, UK, 2–4 April 2008.
30. Huang, L.; Xin, H.; Zhang, L.; Wang, Z.; Wu, K.; Wang, H. Synchronization and frequency regulation of
DFIG-based wind turbine generators with synchronized control. IEEE Trans. Energy Convers. 2017, 32,
1251–1262. [CrossRef]
31. Zhong, Q.-C.; Nguyen, P.-L.; Ma, Z.; Sheng, W. Self-synchronized synchronverters: Inverters without a
dedicated synchronization unit. IEEE Trans. Power Electron. 2014, 29, 617–630. [CrossRef]
32. Kalman, R.E. Phase-plane analysis of automatic control systems with nonlinear gain elements. Trans. AIEE
1955, 73, 383–390. [CrossRef]
33. Atighechi, H.; Chiniforoosh, S.; Jatskevich, J.; Davoudi, A.; Martinez, J.A.; Faruque, M.O.; Sood, V.;
Saeedifard, M.; Cano, J.M.; Mahseredjian, J.; et al. Dynamic average-value modeling of CIGRE HVDC
benchmark system. IEEE Trans. Power Deliv. 2014, 29, 2046–2054. [CrossRef]

© 2018 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access
article distributed under the terms and conditions of the Creative Commons Attribution
(CC BY) license (http://creativecommons.org/licenses/by/4.0/).