You are on page 1of 11

This article has been accepted for publication in a future issue of this journal, but has not been

fully edited. Content may change prior to final publication. Citation information: DOI 10.1109/TPWRS.2021.3049324, IEEE
Transactions on Power Systems
1

Feedback Linearization Control of Converters with


LCL Filter For Weak AC Grid Integration
Javad Khazaei, Senior Member, IEEE, Zhenghong Tu, Student Member, IEEE, Arash Asrari, Member, IEEE.,
and Wenxin Liu, Senior Member, IEEE

Abstract—This paper investigates the design and analysis has complicated the application of this control algorithm. In
of a robust nonlinear controller for voltage source converters addition, the PLL gains in vector-controlled converters must
(VSCs) connected to weak AC grids. Feedback linearization be reduced to guarantee stability, which limits the operating
method is used to develop the control inputs that regulate the
active power and output voltage of the grid-connected converter. ranges of vector controlled converters. In another study, an
The frequency-domain model of an LCL filter in dq-frame is LCL filter was developed that could suppress high-frequency
used to develop the nonlinear control blocks for small-signal current harmonics resulted by weak AC grid integration of
stability analysis. Impact of phase-locked loop (PLL) is added converters [5]. Dynamics of PLL and power transfer capability
to the control block through feedforward loops. It is shown that of the converters in weak AC grids were not considered in that
unlike vector controlled converters, the proposed controller is
not negatively affected by the PLL gains and a broader range study.
of operation is achieved. In addition, the designed controller is Power synchronization control (PSC) is a recently proposed
robust against sudden active power commands from 0 to 1 p.u., approach to enable VSCs with maximum power transfer
and is capable of fault ride through when AC grid is very weak. capability in very weak AC grid integration using synchronous
The robustness of the proposed controller is evaluated using generator concept [7]. The PSC, however, fails to successfully
eigenvalue analysis of the linearized closed-loop model and time-
domain simulations of a grid-connected VSC in various cases. synchronize with the grid in case of faults and the converter
needs to switch back to conventional vector controller, which
reduces the reliability and flexibility of this control algorithm
Index Terms—Weak AC Grids, Nonlinear Control, Feedback
Linearization, Small-signal Stability Analysis. in weak AC grid integration.
Nonlinear control of converters is an alternative control
strategy that can guarantee the global stability and robust
I. I NTRODUCTION performance under uncertainties in the grid. Several studies
Despite the advantages of vector control strategy for inte- focused on application of nonlinear controllers for VSCs for
gration of renewable energy sources, stability and reliability of high voltage DC (HVDC) transmission systems [8], subsyn-
VSCs are not guaranteed when connected to a weak AC grid chronous resonance (SSR) mitigation of double-fed induction
through a long transmission line [1]. A few studies analyzed generator (DFID)-based wind farms [9], or stability enhance-
the weak AC grid integration of VSCs and developed new ment in microgrids [10]. Application of nonlinear controllers
control approaches to overcome the challenge [2]–[6]. For for weak AC grid integration has recently been considered for
example, the stability problems caused by the interactions a unified power quality conditioner (UPQC) that is installed
between the AC and DC voltage regulators of VSCs in a wind between weak AC grid and a VSC [11]. Neural network
energy conversion system were studied in [2]. It was concluded was then utilized to approximate the cost function related
that increasing the AC-bus voltage control gains could improve to the dynamic states of the grid and enhance the transient
the damping characteristics in weak AC grids. However, the stability. Although results showed improvements on transient
proposed method did not provide a solution for weak AC grid stability, short circuit ratio and power transfer capability of the
integration. Similar to [2], the sensitivity of control parameters converter in weak AC grid were not discussed. In addition,
and system design criteria for weak AC grid integration of use of flexible AC transmission systems (FACTS) devices
VSCs were analyzed in [3]; however, a solution was not significantly increases the cost, complexity, and maintenance
proposed. Some existing studies developed solutions to tackle of the proposed technology. Therefore, a robust nonlinear
the challenge of weak AC grid integration. An advanced control approach that can directly be applied to a VSC and
vector control method was proposed in [4] that introduced enhance its integration to weak AC grid has not been proposed
four additional feedforward gains and used gain scheduling yet. In the authors’ recent work, a nonlinear controller was
method to tune the feedforward parameters. However, tun- designed using feedback linearization technique to enhance
ing the parameters using a complex scheduling technique, integration of VSCs to weak AC grids [12]. The designed
controller was validated using time-domain simulations and it
J. Khazaei has a joint appointment with the electrical engineering de-
partment at Penn State Harrisburg and Architectural Engineering department was shown that the nonlinear controlled VSC can successfully
at Penn State University Park, PA, USA, Z. Tu and W. Liu are with the deliver up to 1 p.u. active power to a very weak AC grid with
Electrical Engineering department at Lehigh University, PA, USA, and Arash short circuit ratio (SCR) of 1. That study, however, was a pure
Asrari is with the school of electrical, computer, and biomedical engineering
at Southern Illinois University, Carbondale, IL. (E-mails:jxk792@psu.edu, simulation-based research and stability of the control design
zht218@lehigh.edu, arash.asrari@siu.edu, and wel814@lehigh.edu) was not investigated in [12]. This paper extends the nonlinear

0885-8950 (c) 2020 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission. See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
Authorized licensed use limited to: American University of Beirut. Downloaded on May 23,2021 at 10:13:32 UTC from IEEE Xplore. Restrictions apply.
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI 10.1109/TPWRS.2021.3049324, IEEE
Transactions on Power Systems
2

control design of converters in weak AC grids using small- A. Modeling LCL Filter Dynamics
signal stability analysis approach. Contributions of the paper Te LCL filter dynamics in dq frame can be expressed as
are listed in the following: vpd vpd  0 C f v pq  0 Ln i
*
v pd  *
v
i cdcd − v pd + ω 0 L c icq vcd
= (L 
c s + rc )i
1
cd i cd (1)
1
• A model-based nonlinear controller is designed using Gvd (s ) Gidω(sL) i = (L s + sL
feedback linearization technique that directly compen-
vcq − vpq −

0 c cd c rcc)icqrc
(3) (2)sC f vpd

vpdicd− vgd + ω0 Ln inq =(L n s + rn )ind v gd
V0 cos(
sates the nonlinearities of the system and enables  *)
a robust 0 L c i cq
integration with broader range *of operating points for vpq − vgq − ω0 Ln ind = (Ln s + rn )inq (4)
P  * (a )
weak AC grids compared to vector controlledGp (s) converters, icd − ind + ω0 Cf vpq = Cf svpd (5)
P 
• Eigenvalue analysis and participation factor calculations
vpq icq −(sinq
Gipll ) − ω0 CGfvpll vpd(s)= Cfvsv pq (6)
revealed broad stability margins of the system under
V0 sin( * )
pq 0C f v pd  0 Ln i
various grid strengths or control parameters,  where s i=* d is the Laplace operator,  ω01 is theicqnominal  1 
cq dt
• Sensitivity analysis is conducted to analyze the impact of G vq (s ) Giq (s)i.e., 377 v 
 f vpq 
frequency of the system, rad/s, v sL , v r , i , and i sC
 c output
cd cq cd cq
v*pqgains are the dq-frame components of cqthe converter
c
LCL filter parameters, control parameters, and PLL
on stability of nonlinear controlled converter, 
and current, irespectively.
cq Also, vpd  ,0 Lvcpq
icd, vgd , and vgq are the
voltage vgq
• The proposed nonlinear controller can recover from tem-
dq-frame components of the voltage at the PCC and grid,
porary faults in the system and enables VSCs with fault ( b ) components of
respectively, and ind and inq are the dq-frame
ride through capabilities,
the grid current. As it can be seen in (1)-(6), there exists cross-
• The proposed controller provides total harmonic distor-
coupling effects due to the filter and grid inductance, as well
tion (THD) of less than 5% for point of common coupling
as filter capacitance. Although the cross-coupling components
(PCC) voltage and output current during weak AC grid
conditions.
• To evaluate the system-level performance of the proposed vpd  0 C f v pq  0 Ln inq
controller, a microgrid is developed and the proposed con- vcd  1
i cd
1 1
i nd
troller is connected in parallel with three solar generation
sL c  rc
 f vpd  sLn rn
sC
units to deliver 1 p.u. in a microgrid setup to a weak AC
grid.  0 L c icq v gd

The rest of the paper is organized as follows: Section II (a )


formulates the converter modeling and Section III discusses
the feedback linearization theory and the control design.
vpq 0C f v pd  0 Ln ind
 
Section IV elaborates the case studies and results, and section vcq
V concludes the paper. 1
i cq  1 1
i nq
sL c  rc
 
sC f
vpq  sL n  rn

 0 Lc icd vgq

II. M ODELING OF THE C ONVERTER (b )

Fig. 2: LCL filter dynamics in (a) d-axis and (b) q-axis.


Fig. 1 illustrates an interconnection of a voltage source
converter (VSC) to the grid through an LCL filter. The VSC related to the Lc can be compensated by the current controller
converts the DC source (e.g., battery energy storage or solar in VSCs, the impact of grid inductance, Ln , can result in
panel) to a usable AC signal. In this figure, Ln and rn are destabilizing the system in weak AC grids. The block diagram
the grid impedance components, and Lc , rc , and Cf are filter of the LCL filter in dq frame can be represented as Fig. 2.
components of the VSC. In the following, dynamic modeling If the resistances of the LCL filter are ignored, the transfer
of the converter is presented. function between the grid current and converter voltage can
then be expressed as
ind (s) inq (s) 1 1 1
= = (7)
DC
vc L c rc vp Ln r n vcd (s) vcq (s) Lc Ln Cf s
s2 +
Lc + Ln
DC
Source
Lc Ln Cf
AC ic iCf in
Cf vg Fig. 3 illustrates the Bode response of the transfer function
in (7) under various Ln values corresponding to different
SCRs.
q q seen that the resonant frequency, ωres =
It can be
Lc +Ln 1 1
Fig. 1: A typical interconnection of a three-phase VSC to the Lc Ln Cf = Ln Cf + Lc Cf , is inversely proportional to
grid. Ln , this means increasing Ln results in reducing the reso-
nant frequency. Since increasing Ln results in lower SCRs,
vpd C f v pq Ln inq therefore, the resonant frequency is lower for lower SCRs.

 1
i cd but republication/redistribution
0885-8950 (c) 2020 IEEE. Personal use is permitted, 1 1 i nd
requires IEEE permission. See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
sC v
Authorized licensed use limited to: American University of Beirut. Downloaded on May 23,2021 at 10:13:32 UTC from IEEE Xplore. Restrictions apply.
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI 10.1109/TPWRS.2021.3049324, IEEE
Transactions on Power Systems
3

Bode Diagram where Vpd denotes the amplitude of the PCC voltage. The
100 SCR=1
SCR=2
PLL effect can be added to the control structure of the
Magnitude (dB)
50 SCR=3 VSC by supplementing two extra terms including ∆Ipll (s) =
SCR=10
0 −∆θpll (s)Icd0 to the reference converter current in q frame,
-50
and ∆Vpll (s) = ∆θpll (t)Vpd0 to the reference converter
voltage in the q frame [13], [14]. By solving for ∆θpll (s) in
-100
(9) and replacing the results in these two feedforward items,
-90
dynamics of the PLL can be related to the PCC voltage as
Phase (deg)

-135

-180 ∆Ipll (s) = −Gpll


closed (s)Icd0 ∆vpq = Gipll (s)∆vpq (s) (10)

-225 ∆Vpll (s) = Gpll


closed (s)Vpd0 ∆vpq = Gvpll (s)∆vpq (s) (11)
-270
102 103 104 where Gipll = −Gpll pll
closed (s)Icd0 and Gvpll = Gclosed (s)Vpd0 .
Frequency (rad/s) These two feedforward items will be added to the overall
Fig. 3: Bode diagram of open-loop system under various control block diagram of the nonlinear controlled converter
SCRs.
DC
vc L c rc vp L n r n as shown with blue blocks in Fig. 5.
DC
urce
AC i in same impact on the loop
It can be shownc thatCLf n has the
III. C ONTROL D ESIGN
vg
gain. Since increasing Ln reduces the loop gain, it can be This section lays out a foundation of feedback linearization
concluded that lower the SCR, lower the loop gain. As it can and then will develop the control strategies. Detailed informa-
be seen, the system’s gain margin highly depends on the SCR, tion on theory and application of feedback linearization can
which highlights that Ln has a significant impact in stability be found in [15].
of the closed-loop system. In the next step, a controller will
be designed based on feedback linearization theory to stabilize
the system and minimize the impact of Ln on stability of the A. Feedback Linearization Theory
converter connected to a weak AC grid.
Consider a multi-input multi-output (MIMO) system to the
form of [15]
vpd
vpa abc ẋ = f (x) + g(x)u (12)
vpb
vpc dq Gpll (s) 1
 pll
y = h(x) (13)
vpq s
where x ∈ Rn is the state vector, f (x) is a smooth vector
0 field on Rn , g(x) is a smooth matrix field on Rn , h(x) is a
Fig. 4: Representation of PLL block diagram. smooth output function, x is an n-dimensional state vector,
u ∈ Rm is the vector of input control variables, and y is
the output vector. It is noted that the converter model in (1)-
(6) can be transformed to a standard MIMO model in (12)
B. Impact of Phase-locked Loop and (13). Moving forward, (12) and (13) will be used as the
Phase-locked loop (PLL) provides the transformation angle nonlinear converter model for brevity. Referring to feedback
((t)) for conversion from space-phasor to dq-frame by mea- linearization theory, the output y in (13) will be differentiated
suring the voltage at the PCC and forcing the q component until at least one input variable ui appears. Let us define
vpdof the PCC voltage to zero. The vblock pd diagramCoff v pqa PLL isLnL inqf h(x) and Lg h(x) as Lie derivatives of h(x) and g(x),
v*pd illustrated iin *
cd
Fig. 4, where (t) 
vcd = θpll (t) 1
= ωti cd is the phase
1 respectively,
1
i nd
which are defined as Lf h(x) =
∂h
f (x) and
G (s ) G (s ) ∂x
c  rc4, the small-signal
 vpd
 sLn  rn
anglevd of the PCC voltage. Referring to sL
id Fig. sC f

open-loop transfer icd


 function of the  LPLL can be obtained by [13] v gdLg h(x) =
∂h
g(x). Therefore, referring to (12) and (13), the
c icq ∂x
pd (s) ∆θpll (s) Gpll (s) first derivative of output i, y˙i , can be written as [15]
Gpll
open (s) = = (8)
∆vpq (s) s m
(s) X
pq
vpq Gipll (s) Gvpllpll
(s) vpq y˙i = Lf hi (x) + (Lgi hi (x))ui (14)
k C f v pd  Ln ind
i cq* (s) = kp + s is aproportional  1 (PI) 
where Gpll ΔIpll pll i ΔVpll i nq
i=1
v*pq i cq integral 1
1
G (s
controller) G (s )
of the PLLiqto forcevthe q-component sL c  rc of PCC voltage In this case, if Lgi hj (x) = 0, no input will appear in the
 PLL vispq derivative 
sC f sL r
transfer function of the
vq

 cq n n
to zero. The closed-loop v function and one needs to differentiate again until
 L c icd
expressed asicq[13]
gq
ui appears. If the input appears after rm derivations, then,
∆θpll (s) Gpll
open (s)
m
Gpll
closed (s) = = (9) (rm ) (r ) (r −1)
X
yi = Lf m hi (x) + Lgj Lf m hi (x)uj , (15)
∆vpq (s) 1 + Gpll
open (s)Vpd
j=1

0885-8950 (c) 2020 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission. See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
Authorized licensed use limited to: American University of Beirut. Downloaded on May 23,2021 at 10:13:32 UTC from IEEE Xplore. Restrictions apply.
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI 10.1109/TPWRS.2021.3049324, IEEE
Transactions on Power Systems
4

vpd vpd  0 C f v pq  0 Ln inq


v *
pd
 i *
cd vcd  1
icd
1 1
i nd
Gvd (s ) Gid (s ) sL c  rc
 vpd
 sLn  rn
sC f

icd
  0 L c icq v gd
V0 cos( ) *

*
P *
(a )
Gp (s)
P  vpq Gipll (s) Gvpll(s) vpq
V0 sin( * ) 0C f v pd  0 Ln ind
 i *
 i cq   i nq
cq 1 1 1
G vq (s ) Giq (s) vcq sL c  rc
v*pq  
sC f
vpq  sL n  rn

i cq   0 Lc icd vgq

(b )

Fig. 5: Block diagram of VSC control with feedback linearization controller.

where i = {1, 2, . . . , m}. Equation (15) can be written in a B. Nonlinear Control Design
matrix form as
 vpd    0 C f v pq The control  0 Ln inq
objectives are active power and voltage regula-

 1.   f . vcd 
   (r )
(r1 ) 1
L h1 (x) tion at the point of common coupling (PCC). Therefore, the
y u1 icd i nd
 . = 1 . nonlinear
1 controller 1 is designed to regulate the output power
.  + E(x)  .   
(16) sC f
 .  
ym
(rm ) (rm )
Lf hm (x)
.  sL c . rc
u m
 through  loop
vpd an outer sL n  rncontrol input, and the voltage regulation
is carriedv out by the inner control loop. When connected to
 0 L c icq gd
weak AC grids, the VSCs suffer from voltage stability issues
where
 (r −1) (r −1)
 ( a ) and it was recommended in [16] to maintain the alternating
Lg1 Lf 1 h1 (x) . . . Lgm Lf 1 h1 (x) voltage level at the PCC instead of reactive power control.
E(x) = 
 .
.. . .. .
..
 Therefore, the main objective of the controller in this design is
v  0  0 Lthe
 
pq
C f v pd regulating n ind voltage at the point of common coupling instead
Lg1 Lf m hm (x) .v.cq. L
 of reactive
(r −1) (rm −1)
g m Lf hmi(x)
cq
power. The active i nq power is measured and passed
1 1 1
sL c  rc 2
through a low-pass filter. Dynamics of the measured active
 powervpqcan 
∂Lf h(x) sC f sL n  rn
and Lg Lf h(x) = g(x) and Lf Lf h(x) = Lf h(x) = then be represented by
∂Lf h(x)
∂x
 Lc icd vgq
3
f (x). The relative degree of0 the system is denoted Ṗ = −ωc P + ωc (vpd ind + vpq inq ) (18)
∂x 2
as the number of times that the output derivation needs to
(b )
be conducted until the input variable u appears. The MIMO where ωc is the bandwidth of the low-pass filter. The error
dynamics in (12) and (13) have a vector relative degree between the measured and reference active power is passed
{r1 , r2 , . . . , rm } at a point x0 if through a compensator to generate the converter angle. The
k −1 converter angle along with the nominal magnitude of voltage
• Lgj (Lf ) hi (x) = 0 for 1 ≤ j ≤ m, k ≤ ri − 1 ,
are used to generate the dq-frame reference voltages such that
1 ≤ i ≤ m, and all x in the neighborhood of x0 [15],  
0 ki
• Matrix E(x) is nonsingular at x = x [15].
δ ∗ = kp + (P ∗ − P ) = Gp (s)(P ∗ − P ) (19)
Therefore, a solution can be found for some choices of output s

functions h1 (x), P . . . , hm (x) if the total relative degree of vpd = V0 cos (δ ∗ ) (20)
m
the system, e.g., i=1 i r , is equal to the dimension of the ∗
vpq = V0 sin(δ ) ∗
(21)
state-space model (i.e., n). The input transformation can be
designed by [15] where δ ∗ is the reference converter angle for the active power
 (r )  control loop and is different than the PLL angle, θ, V0 is
Lf 1 h1 (x)
 
v1 the
 magnitude of the nominal PCC voltage, and Gp (s) =
.. .. 

u = −E −1 (x)   + E −1 (x)  k
 
 .   .  (17) kp +
i
is a PI controller for active power regulation.
(rm ) v s
Lf hm (x) m
| {z }
v

where v is the new input vector to be designed [15]. The C. Feedback Linearization-based Inner Control Design
system can be then converted to a linear and decoupled closed- To design the voltage regulation loops, the voltage tracking
(r ) ∗
loop model of yk k = vk , k ∈ {1, 2, ..., m} by replacing (17) errors are considered, e.g., evd = vpd − vpd and evq =

into (16). vpq −vpq . By taking a derivative of voltage tracking errors and

0885-8950 (c) 2020 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission. See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
Authorized licensed use limited to: American University of Beirut. Downloaded on May 23,2021 at 10:13:32 UTC from IEEE Xplore. Restrictions apply.
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI 10.1109/TPWRS.2021.3049324, IEEE
Transactions on Power Systems
5

vpd f (evd) / evd icd


* f (eid ) / eid Weak AC Grid
evd icd eid
E 1 K1 E id 1 Pulse
v*pd vd
K3
dq
i cdq v p
Generation
cos (  ) Unit PLL
P*
K2  VSC
kP vcd Pulse dq
abc
eP   ,
Pulse
vcq Gen icabc v abc
p
P ki  sin(  ) f (evq) / evq
f (eiq) / eiq Filter
v*pq *
icq PCC
1
eiq
E K1 E 1
evq vq iq
K3
v pq
K2  icq

Fig. 6: A VSC enhanced with feedback linearization control connected to a weak AC grid.

replacing v̇pd and v̇pq from (5) and (6), dynamics of voltage e.g., eid = i∗cd − icd and eiq = i∗cq − icq , are considered. By
tracking can be derived taking the derivative of current tracking error dynamics and
1 replacing i̇cd and i̇cq from (1) and (2), one can find
∗ ∗
ėvd = v̇pd − v̇pd = v̇pd + (ind − icd ) − ω0 vpq (22)
Cf rc 1
ėid = i̇∗cd + icd − ω0 icq + (vcd − vpd ) (31)
∗ ∗ 1 Lc Lc
ėvq = v̇pq − v̇pq = v̇pq + (inq − icq ) + ω0 vpd (23)
Cf rc 1
ėiq = i̇∗cq + icq + ω0 icd + (vcq − vpq ) (32)
1 Lc Lc

Let us define f (evd ) = ind − ω0 vpq + v̇pd , f (evq ) =
Cf rc 1
1 1 By defining f (eid ) = i̇∗cd + icd − ω0 icq − vpd , f (eiq ) =

inq + ω0 vpd + v̇pq , and E vd = E vq = − , the above Lc Lc
Cf Cf Rc 1 1
i̇∗cq + icq +ω0 icd − vpq , and E id = E iq = , equations
equation simplifies to (24) and (25) Lc Lc Lc
(31) and (32) simplify to (33) and (34)
ėvd = f (evd ) + E vd icd (24)
ėvq = f (evq ) + E vq icq (25) ėid = f (eid ) + E id vcd (33)
ėiq = f (eiq ) + E iq vcq (34)
which can be written as
  
ẏ1 L1 h1 (x)
  
i Then, using (17) and choosing a proportional gain controller,
= + E vd cd (26) the converter control inputs can be defined as
ẏ2 L1 h2 (x) icq
where ẏ1 = ėvd , ẏ2 = ėvq , L1 h1 (x) = f (evd ), and L1 h2 (x) = vcd = −E −1 −1
id f (eid ) + E id K3 eid = Gid (s)eid (35)
f (evq ). Considering icd and icq are inputs to be designed, vcq = −E −1 −1
iq f (eiq ) + E iq K3 eiq = Giq (s)eiq (36)
using (17) and (26), one can find the dynamics of the input
controller as where K3 is the proportional compensator gain and Gid (s)
and Giq (s) are calculated by:
i∗cd = E −1
vd [−f (evd ) + v1 ] = Gvd (s)evd (27)  
f (eid )
i∗cq = E −1
vq [−f (evq ) + v2 ] = Gvq (s)evq (28) −1
Gid (s) = Eid K3 − (37)
eid
where i∗cd and i∗ca are reference converter currents in dq 
f (eiq )

−1
reference frame, and v1 and v2 are two inputs to force the Giq (s) = Eiq K3 − (38)
eiq
voltage tracking error to zero. In addition, Gvd and Gvq are
expressed by: The designed controller inputs will then be supplemented to a
   VSC connected to a weak AC grid. The block diagram of the
−1 K2 f (evd )
Gvd (s) = Evd K1 + − (29) proposed controller is illustrated in Fig. 5 and Fig. 6 depicts
s evd the detailed implementation diagram. Parameter designs for
  
−1 K2 f (evq ) K1 , K2 , and K3 are included in the Appendix section.
Gvq (s) = Evq K1 + − (30)
s evq
A proportional integral (PI) controller can be chosen for v1 IV. C ASE S TUDIES
K2 K2
and v2 as v1 = (K1 + )evd and v2 = (K1 + )evq In this section, small-signal analysis is conducted to validate
s s
to force the voltage dynamics to zero. Therefore, v1 and v2 the effectiveness of the proposed controller in weak AC grid
will be replaced in (27) and (28), respectively. To design the integration of VSCs. Time-domain simulations are then carried
current controller, the dynamics of the current tracking error, out to support the small-signal analysis results.

0885-8950 (c) 2020 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission. See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
Authorized licensed use limited to: American University of Beirut. Downloaded on May 23,2021 at 10:13:32 UTC from IEEE Xplore. Restrictions apply.
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI 10.1109/TPWRS.2021.3049324, IEEE
Transactions on Power Systems
6

TABLE I: Eigenvalue analysis of the nonlinear controller under various SCRs.


SCR=1 SCR=2 SCR=3
Xpf λ f (Hz) η (%) λ f (Hz) η (%) λ f (Hz) η (%)
P, ω, vcd , icd λ1 = −4.3 + j1334 212 0.32 λ1 = −3.6 + j1732 275 0.2 λ1 = −3.4 + j2037 324 0.16
P, ω, vcd , icd λ2 = −4.3 − j1334 212 0.32 λ2 = −3.6 − j1732 275 0.2 λ2 = −3.4 − j2037 324 0.16
P, ω λ3 = −2.2 + j581 92.4 0.37 λ3 = −1.9 + j978 60 0.19 λ3 = −1.9 + j1283 204 0.14
P, ω λ4 = −2.2 − j581 92.4 0.37 λ4 = −1.9 − j978 60 0.19 λ4 = −1.9 − j1283 204 0.14
θ, γP λ5 = −1.3 + j377 60 0.34 λ5 = −1.3 + j377 60 0.34 λ5 = −1.3 + j377 60 0.34
θ, γP λ6 = −1.3 − j377 60 0.34 λ6 = −1.3 − j377 60 0.34 λ6 = −1.3 − j377 60 0.34
vcd λ7 = −56.6 0 100 λ7 = −56.6 0 100 λ7 = −56.6 0 100
icq λ8 = −4.6 0 100 λ8 = −6.55 0 100 λ8 = −7.13 0 100
inq λ9 = −0.1 0 100 λ9 = −0.1 0 100 λ9 = −0.1 0 100
ind λ10 = −8.5e−6 0 100 λ10 = −3.9e−6 0 100 λ10 = −2.5e−6 0 100
vpq λ11 = −8.5e−6 0 100 λ11 = −3.9e−6 0 100 λ11 = −2.5e−6 0 100
vpd λ12 = −8.5e−6 0 100 λ12 = −3.9e−6 0 100 λ12 = −2.5e−6 0 100

1500 1500

1000 1000 1000

500 500 500

0 0 0

-500 -500 -500

-1000 -1000 -1000

-1500 -1500
-6 -5 -4 -3 -2 -1 0 -10 -8 -6 -4 -2 0
-20 -15 -10 -5 0 5 10

Fig. 7: Impact of (1) PLL gains (left subplot), (2) outer loop gains (middle subplot), and (3) converter impedance (right subplot)
on stability of the system.

A. Sensitivity Analysis the PLL gains were multiplied by a factor K. By modifying


K from 1 to 5, the trajectory of the eigenvalues were plotted,
The block diagram of the system with the propsoed con- as shown in Fig. 7 (left subplot). It can be observed that
troller (shown in Fig. 5) was developed in MATLAB/Simulink by increasing the PLL gains, λ1 to λ4 move towards the
for the small-signal analysis. MATLAB’s “LINMOD” function origin, and the system becomes unstable when PLL gains are
was used to derive the state-space linearized A, B, C, D ma- increased by 1.5 times. Unlike vector controlled approaches
trices of the closed-loop system. The readers are encouraged that need to reduce PLL gains in weak AC grid integration
to refer to [17] to learn the basics of linearization process in to ensure stability [19], nonlinear controlled converter allows
MATLAB. Once the state-space matrices are derived, eigen- a reasonable margin for the PLL gains to increase without
value analysis can be conducted to examine the stability of the destabilizing the system in very weak AC grid condition. This
closed-loop system. Eigenvalues of the VSC with the propsoed provides the control system enough bandwidth for stability
controller under various SCRs are shown in Table I. As it can purposes.
be seen, the system has 12 eigenvalues. Participation factor
analysis is also carried out to relate the modes to the state 2) Impact of Nonlinear Controller Gains: To analyze the
variables of the system, as listed in the Xpf column in the sensitivity of outer loop power controller gains on stability of
table. The frequency and damping of each mode have also the nonlinear controlled converter in weak AC grids, the SCR
been shown in the table. It is noted that, if λi = αi +jβi is the was set to 1 and the outer loop gains (kp and ki in (19)) were
eigenvalue of ith under-damped oscillating mode, p the damping multiplied by a factor K. The gain, K, was modified from 1 to
ratio for this mode is calculated by ηi = −αi / αi2 + βi2 [18]. 50 and the trajectory of eigenvalues were plotted, as depicted
As can be seen from Table. I, all the modes reside in the left in Fig. 7 (middle subplot). It is observed that by increasing the
half-plane (LHP) and the system is stable even under very outer loop gains, λ1 and λ2 move towards the origin, but they
weak AC grid condition with SCR=1. In addition, λ1 to λ6 stay on the left half plane even after increasing the gains by 50
provide low damping and λ9 to λ12 reside very close to origin. times. In addition, λ3 to λ6 move towards the left hand plane
Impact of parameters of the system on these eigenvalues will (LFP) and the stability of the system is improved. Therefore,
be investigated in the case studies. the designed controller is not sensitive towards changes in the
1) Impact of PLL: To analyze the impact of PLL gains on outer loop gains and the controller bandwidth can be chosen
stability of the designed controller, the SCR was set to 1 and without impacting stability.

0885-8950 (c) 2020 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission. See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
Authorized licensed use limited to: American University of Beirut. Downloaded on May 23,2021 at 10:13:32 UTC from IEEE Xplore. Restrictions apply.
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI 10.1109/TPWRS.2021.3049324, IEEE
Transactions on Power Systems
7

3) Impact of Converter Impedance: This scenario inves- to deliver more than 0.6 p.u. in SCR=1 [20], the developed
tigates the impact of converter impedance on stability of nonlinear-based controller delivers 1 p.u. successfully. In ad-
the system in very weak AC grid conditions. The converter dition, the converter could regulate the voltage and frequency
impedance, i.e., rc + jω0 Lc , is increased via multiplying it to 1 p.u. It is worth noting that the converter is required to
by a gain K and increasing the gain from 1 to 3. Eigenloci be supported by 0.5 p.u. reactive power when the SCR is 1
of the system for this condition can be shown in Fig. 7 (see the last subplot). Therefore, the VA rating of the converter
(right subplot). It is observed that by increasing the converter needs to be higher than 1 p.u. in order to support the required
impedance by 1.9 times, λ8 which is related to the converter reactive power. The results confirm the stability analysis results
current (referring to participation factor analysis in Table I) shown in Table I as the system remains stable under very
moves towards the right half plane (RHP) and the system weak AC grid condition. In addition, the fast Fourier transform
becomes unstable. (FFT) analysis results of the PCC voltage and converter’s
output current for the feedback linearization controlled VSC
at SCR=1 for 1 p.u. active power transfer is shown in Figure
1 SCR=1 9. It is observed that the proposed controller provides 4.25%
P [p.u.]

Pref current THD and 3.87% voltage THD, which is acceptable.


0.5
Pmeasured
0 THD= 4.25% THD= 3.87%
0 2 4 6 8 10
1.02 100 100

Mag (% of Fundamental)

Mag (% of Fundamental)
f [p.u.]

80 80
1
Current Voltage
fref 60 60
0.98 f
0 2 4 6 8 measured 10 40 40
Nonlinear
0
Q [p.u.]

20 20
-0.2
-0.4 0 0
0 200 400 600 800 1000 0 200 400 600 800 1000
-0.6 Frequency (Hz) Frequency (Hz)
0 2 4 6 8 10
Vref Fig. 9: THD analysis of voltage an current waveforms of
nonlinear-based converter for SCR=1.
V [p.u.]

1.1 Vmeasured
1
0.9 2) Sensitivity of Outer Loop Gains: This case study inves-
0 2 4 6 8 10 tigates the sensitivity of outer loop control gains on the per-
0.4 formance of the grid-connected VSC system. Three different
Vpq [p.u.]

0.2 gains are selected for the outer loop and simulation results are
0 illustrated in Fig. 10. As it can be shown in the first subplot,
Time [sec] the dynamic response of the system for active power transfer
-0.2
0 2 4 6 8 10 is better with high control gains (this confirms the eigenvalue
analysis results in Fig. 7). In addition, as it is depicted in the
third subplot in Fig. 10, dynamic response of the converter’s
Fig. 8: Simulation results for a nonlinear-based converter for
reactive power is similar to the active power and the response
SCR=1.
improves with high control gains. Compared to the existing
methods for weak AC grid integration of the VSCs, that
are very sensitive to changes in the control parameters [1],
B. Time-domain Simulations [21], [22], the designed controller is robust against parameter
To validate the effectiveness of the designed feedback changes (this validates the small-signal analysis results in Fig.
linearization-based nonlinear controller, time-domain simula- 7).
tions on a grid-connected converter (as shown in Fig. 1) are 3) Fault Ride Through Capability: This case examines the
carried out using MATLAB Simscape power system toolbox. capability of proposed controller in fault ride through after
Average converter model was used for the nonlinear controlled transients for SCR=1. The reference power for the VSC is set
VSC. Four case studies are considered: (1) converter response to 1 p.u. A balanced three-phase short circuit occurs at the
when the AC grid is very weak with SCR=1, (2) sensitivity PCC after 3 seconds and is cleared after 3 cycles. The fault
of output power regulation gains when SCR=1, (3) fault ride impedance is set to 0.2 Ω to cause a voltage dip at the PCC.
through capability of the proposed controller when SCR=1, Simulation results for this case study are shown in Fig. 11. As
and (4) sudden active power change when SCR=1. Parameters it can be observed, the converter successfully recovers after the
of the system are listed in the Appendix section. fault and restores the voltage and frequency to their nominal
1) Very Weak AC grid with SCR=1: The capability of the settings. It is also seen that the converter successfully delivers
developed controller in transferring 1 p.u. active power in a 1 p.u. in very weak AC grid after the fault is cleared.
very weak AC grid with SCR=1 is evaluated in this case as 4) Black Start Capability: In this case, the capability of
illustrated in Fig. 8. Unlike vector controlled VSCs that fail the proposed controller in responding to sudden active power

0885-8950 (c) 2020 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission. See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
Authorized licensed use limited to: American University of Beirut. Downloaded on May 23,2021 at 10:13:32 UTC from IEEE Xplore. Restrictions apply.
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI 10.1109/TPWRS.2021.3049324, IEEE
Transactions on Power Systems
8

1.2
1
1
P [p.u.]

1 0.8
0.5
2.9 3 3.1 3.2 3.3 3.4 3.5 3.6
Kp =0.02, Ki =0.001 0.8 1.02
0 Kp =0.2, K i =0.005
0 2 4 6 8 10
Kp =0.4, K i =0.01 5 6 1
1.02
0.98
2.9 3 3.1 3.2 3.3 3.4 3.5 3.6
f [p.u.]

1
0
-0.5
1.02 -1
0.98
0 2 4 6 8 10 -1.5
1 2.9 3 3.1 3.2 3.3 3.4 3.5 3.6
0

5 5.5 1
Q [p.u.]

-0.2
0.95
-0.4
0.9
2.9 3 3.1 3.2 3.3 3.4 3.5 3.6
0 2 4 6 8 10

1 Fig. 11: Simulation results for fault ride through capability of


nonlinear-based controller when SCR=1.
V [p.u.]

1
0.9

0.9 1.5
0.8
0 2 4 6 8 10 1
Time [sec] 0.8 Pref
4.999 5 5.001 5.002 5.003 0.5
Pmeasured
Fig. 10: Simulation results for a nonlinear-based converter for 0
0 1 2 3 4 5
sensitivity of control gains.
1

commands in SCR=1 is tested. The reference active power is 0.96


set to 0 initially and is increased to 1 p.u. after 1 second. 0 1 2 3 4 5

As it can be seen in Fig. 12, the proposed controller can 1


successfully achieve the control objectives even with a sudden 0
step change in the active power demand in very weak AC grid -1
condition, however, the controller gains had to be re-tuned. -2
New controller gains include Kp = 0.25 and Ki = 0.0001. 0 1 2 3 4 5

5) Sudden SCR Change: This case evaluates the perfor- 1.2


1
mance of the proposed controller when the SCR changes
0.8
from 3 to 1 during the operation. To apply the SCR change 0.6
in time-domain simulations, two parallel impedances with 0.4
two breakers at the sending and receiving ends of the grid 0 1 2 3 4 5

impedance are considered. The first impedance is set for


SCR=3 for the first three seconds and the parallel impedance Fig. 12: Simulation results for black start capability.
is set for SCR=1. After three seconds, the top impedance (with
SCR=3) is isolated from the system (by breaker operations)
and the second impedance (with SCR=1) will be connected to multiple VSCs, a microgrid model was developed. The model
the system. The reference active power is set to 1 p.u. for both was composed of a VSC with the proposed controller and three
scenarios. Simulation results for this case study are depicted solar photovoltaic (PV) farms connected to a weak AC grid
in Fig. 13. As it can be seen, the proposed controller can through the point of common coupling (PCC) as illustrated in
successfully operate in strong grid conditions with SCR=3 (the Figure 14. The solar array is connected to the PCC through a
first 3 seconds in the plots) and can successfully respond to three-phase DC to AC converter (inverter) and a filter. Standard
an operating condition change when the SCR changes from 3 vector control with maximum power point tracking (MPPT) is
to 1 after 3 seconds. used to control the solar inverter in a synchronous reference
6) Operation in a Microgrid Setup: To study the perfor- frame. The PV systems operate at a constant solar irradiance
mance of the proposed controller in grid integration with input of 1000W/m2 supplemented with a random noise to

0885-8950 (c) 2020 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission. See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
Authorized licensed use limited to: American University of Beirut. Downloaded on May 23,2021 at 10:13:32 UTC from IEEE Xplore. Restrictions apply.
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI 10.1109/TPWRS.2021.3049324, IEEE
Transactions on Power Systems
9

1.5 0.5

1 P
0.25 ref
P measured
0.5
0
Pref 0 1 2 3 4 5 6 7 8 9P 10
0 PV
0 1 2 3 4 5P 6 0.6
measured

1.015
0.5
1

0.985 0.4
0 1 2 3 4 5 6 7 8 9 10
0 1 2 3 4 5 6 1.02
1
1
0
0.98
0 1 2 3 4 5 6 7 8 9 10
-1
0 1 2 3 4 5 6 1
1.01
0

-1
1
0 1 2 3 4 5 6 7 8 9 10

1.1
0.99
0 1 2 3 4 5 6 1

0.9

Fig. 13: Simulation results for SCR change during operation 0 1 2 3 4 5 6 7 8 9 10

of the system.
Fig. 15: Simulation results for microgrid integration.
PCC
Nonlinear AC
Control grid is 1 p.u. after 5 seconds. As it can be seen in the second
DC
subplot, the output of PV farms is highly uncertain due to
the changes in the irradiance. The intermittency of solar farm
DC Weak Grid
resulted in some spikes in voltage and frequency at the PCC,
which are negligible. The proposed controller can successfully
AC
operate when connected in parallel to other VSCs and can
Solar Farm Inverter
1 deliver up to 1 p.u. in very weak AC grid in a microgrid
setup.
DC
AC Grid
Solar Farm Inverter V. C ONCLUSION
2
A nonlinear controller was developed in this paper using
DC
feedback linearization theory for VSCs to improve their dy-
AC namic performance in very weak AC grid conditions. The
Solar Farm Inverter objective of the controller was to regulate the active power and
3
voltage of the converter in dq frame. Small-signal analysis was
Fig. 14: A microgrid setup to test the grid integration of the conducted to analyze the stability of the designed controller in
proposed controller. SCR=1. It was demonstrated through eigenvalue analysis that
control gains have minimum impact on stability of the nonlin-
ear controlled converter. In addition, the filter parameters have
account for the uncertain behavior of output PV power. The minimum impact on stability of modes that reside very close
information on developing the solar photovoltaic system can to the origin. Unlike vector controlled converters that should
be found in [23], [24]. have reduced PLL gains for weak AC grid integration, the
Simulation results for microgrid integration of the nonlinear proposed controller is stable even when 50% increase in PLL
controlled VSC is demonstrated in Fig. 15. The total power gains occurs in very weak AC grid conditions. Time-domain
output of PV farms averages around 0.5 p.u., therefore, the simulation results verify the effectiveness of the proposed
VSC should support up to 0.5 p.u. in SCR=1 such that the controller in very weak AC grid conditions with SCR=1. It
microgrid (PV farms and VSC) delivers 1 p.u. to weak AC was also found that, the proposed nonlinear-based controller
grid. Hence, the VSC is initially set to deliver 0.25 p.u. active could recover from temporary faults and was capable of black
power and the setpoint changes to 0.5 p.u. after 5 seconds. As a start even in very weak AC grid conditions. Finally, it was
result, the total power sent from the microgrid to the weak AC demonstrated that the proposed controller operates well when

0885-8950 (c) 2020 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission. See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
Authorized licensed use limited to: American University of Beirut. Downloaded on May 23,2021 at 10:13:32 UTC from IEEE Xplore. Restrictions apply.
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI 10.1109/TPWRS.2021.3049324, IEEE
Transactions on Power Systems
10

a VSC is connected in parallel with other VSCs in a microgrid


setup and could deliver up to 1 p.u. to a very weak AC grid.

v*pd K2 i cd K3 vpd
1 1
K1 
s Lc C f s Lc s  rc
vpd
(a)
(a )
i cd* K3 vcd i cd
1
Lc L c s  rc
i cd

(b )
Fig. 16: Simplified models for tuning the control parameters
K1 and K2 (a) and K3 (b).
(b)

Fig. 17: Pole-zero map of simplified block diagrams for tuning


vpd f (evd) / evd icd control parameters K3 (a), and K1 and K2 (b).
A PPENDIX
* f (eid ) / eid Weak AC Grid
e icd
To tunevdthe control
1
E vd K1 1 2
parameters K , K , and K3 ,eidsimplified
E id 1 TABLE II: Parameters of the System
* Pulse dq
vcontrol
pd blocks are derived from Fig. 5. Fig. 16 (a) shows the K3 i cdq v p
Component Generation
Parameter and Value
simplified model for tuning K1 and K2 , which is derived from
cos5(by Ssc = 10 Unit
PLL
 Similar approach is

Fig. ) ignoring the feedforward
K2 items. Grid MVA, Vb = 100kV, f = 60 Hz, X/R = 7
used to find the simplified control diagram for tuning K3 , as Zb = 1000 Ω,VSC
Ln = 2.62 H
vcd Sb = 10MVA,Pulse dq
Vb = 100kV, f = 60 Hz, abc

depicted in Fig. 16 (b). The closed-loop transfer functions of VSC
Pulse  ,
Lc = 0.1Ln , rc = 35 mΩ, kp = 0.27, kabc
i = 0.001abc
the system for Fig. 16 are expressed as: vcqK1 = 0.1,
Gen ic v
rad/secp
 sin(  ) vpd

Kf (1eK
vq3) / evq 2 3
s+K K
K2 = 20, K3 = 10, ωc = 56.6
Kppll = 180, kipll = 3200, Cf = 2.5 µ F
G1 (s) = = (39)
*
vpq v (C L2 )s2 + (K K + r )s + K K
fPV
(eiqFarms
) / eiq Sb = 0.5 MVA (each), Vb = 5kV,Filter
f = 60 Hz,
pd f c 1 3 c *
icq
2 3
eiq PCC
i∗cd  1 K3 K 1 Filter (R = 10 mΩ, L = 0.25 mH, C = 1.5 µF )
G2 (s) = E E
evq icd =vqL2 s + (K3 + r1c Lc ) (40)
iq Transformer (5/100 kV, 1 MVA, L1 = L2 = 0.002 p.u.)
c K3
Thevtuning starts with K3 by plotting the pole-zero
icq map of

pq
K2
closed-loop transfer function G 2 (s) and checking the eigen-
[2] Y. Huang, X. Yuan, J. Hu, P. Zhou, and D. Wang, “DC-bus voltage
values of the system. Once K3 is tuned, the final value will be control stability affected by AC-bus voltage control in VSCs connected
used in closed-loop transfer function of G2 (s) for parameter to weak AC grids,” IEEE J. Emerg. Topics Power Electron., vol. 4, no. 2,
tuning for K1 and K2 using pole-zero map of G2 (s). Fig. 17 pp. 445–458, June 2016.
[3] A. Yogarathinam, J. Kaur, and N. R. Chaudhuri, “Impact of inertia
(a) shows the pole-zero map of closed-loop transfer function and effective short circuit ratio on control of frequency in weak grids
G2 (s) for increasing K3 , and Fig. 17 (b) depicts the pole-zero interfacing LCC-HVDC and DFIG-based wind farms,” IEEE Trans.
map of G1 (s) for increasing K1 and K2 , indicating larger K1 , Power Del., vol. 32, no. 4, pp. 2040–2051, Aug 2017.
[4] A. Egea-Alvarez, S. Fekriasl, F. Hassan, and O. Gomis-Bellmunt,
K2 , and K3 gains result in movement of poles more to the “Advanced vector control for voltage source converters connected to
right-half-plane. However, one should not increase the gains weak grids,” IEEE Trans. Power Syst., vol. 30, no. 6, pp. 3072–3081,

too much to avoid violating the maximum limits on vpd or i∗cd . Nov 2015.
[5] J. Fang, X. Li, X. Yang, and Y. Tang, “An integrated trap-LCL filter with
Similar approach can be used for developing a simplified block reduced current harmonics for grid-connected converters under weak
diagram for active power controller loop and tuning kp and ki grid conditions,” IEEE Trans. Power Electron., vol. 32, no. 11, pp. 8446–
for active power regulator. More information on developing 8457, Nov 2017.
[6] S. Sang, N. Gao, X. Cai, and R. Li, “A novel power-voltage control
simplified control block diagrams for active power controller strategy for the grid-tied inverter to raise the rated power injection level
tuning can be found in [21]. in a weak grid,” IEEE J. Emerg. Topics Power Electron., vol. 6, no. 1,
Parameters of the microgrid system in Figure 14 and the pp. 219–232, March 2018.
[7] P. Mitra, L. Zhang, and L. Harnefors, “Offshore wind integration to a
VSC are provided in Table II. It is noted that Ssc is the short weak grid by VSC-HVDC links using power-synchronization control: A
circuit capacity of the grid. case study,” IEEE Trans. Power Del., vol. 29, no. 1, pp. 453–461, Feb
2014.
[8] Z. Li, Q. Hao, F. Gao, L. Wu, and M. Guan, “Nonlinear decoupling
R EFERENCES control of two-terminal MMC-HVDC based on feedback linearization,”
IEEE Trans. Power Del., vol. 34, no. 1, pp. 376–386, 2018.
[1] L. Harnefors, M. Bongiorno, and S. Lundberg, “Input-admittance cal- [9] M. Chowdhury and G. Shafiullah, “SSR mitigation of series-
culation and shaping for controlled voltage-source converters,” IEEE compensated FIG wind farms by a nonlinear damping controller using
Trans. Ind. Electron., vol. 54, no. 6, pp. 3323–3334, 2007. partial feedback linearization,” IEEE Trans. Power Syst., vol. 33, no. 3,

0885-8950 (c) 2020 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission. See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
Authorized licensed use limited to: American University of Beirut. Downloaded on May 23,2021 at 10:13:32 UTC from IEEE Xplore. Restrictions apply.
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI 10.1109/TPWRS.2021.3049324, IEEE
Transactions on Power Systems
11

pp. 2528–2538, 2017. Zhenghong Tu (Student Member, IEEE) received


[10] M. Davari, W. Gao, and F. Blaabjerg, “A fault-tolerant, passivity- the B.S. degree in electrical engineering from the
based controller enhanced by the equilibrium-to-equilibrium maneuver Huazhong University of Science and Technology,
capability for the DC-voltage power port VSC in multi-infeed AC/DC Wuhan, China, in 2015, and the M.S. degree in
modernized grids,” IEEE Trans. Emerg. Sel. Topics Power Electron., electrical engineering from the Shanghai Jiao Tong
2019. University, Shanghai, China, in 2018. Currently, he
[11] H. Nazaripouya and S. Mehraeen, “Modeling and nonlinear optimal is working toward the Ph.D. degree in electrical
control of weak/islanded grids using FACTS device in a game theoretic engineering with the Lehigh University, Bethlehem,
approach,” IEEE Trans. Control Syst. Technol., vol. 24, no. 1, pp. 158– PA, USA. His research interests include microgrid,
171, Jan 2016. power electronics, and controls.
[12] J. Khazaei, Z. Tu, A. Asrari, and W. Liu, “Nonlinear control design for
voltage source converters in weak AC grids,” in 2020 IEEE Texas Power
and Energy Conference (TPEC), College Station, TX, 2020, pp. 1–6.
[13] B. Wen, D. Boroyevich, R. Burgos, P. Mattavelli, and Z. Shen, “Analysis
of DQ small-signal impedance of grid-tied inverters,” IEEE Trans. Power
Electron., vol. 31, no. 1, pp. 675–687, 2015.
[14] Y. Tang, J. Fang, X. Li, and H. Li, “Reshaping quadrature-axis Arash Asrari (Member, IEEE) received the Ph.D.
impedance of three-phase grid-connected converters for low-frequency degree in electrical engineering from the University
stability improvement,” in 2018 International Power Electronics Con- of Central Florida, Orlando, FL, USA, in 2015. From
ference (IPEC-Niigata 2018-ECCE Asia). IEEE, 2018, pp. 3910–3915. 2015 to 2017, he was a senior Consulting Engineer
[15] A. Isidori, Nonlinear control systems. Springer Science & Business with Phasor Engineering, LLC, located in the state of
Media, 2013. Florida. He is currently an Assistant Professor with
[16] L. Zhang, L. Harnefors, and H.-P. Nee, “Power-synchronization control the School of Electrical, Computer, and Biomedical
of grid-connected voltage-source converters,” IEEE Trans. Power Syst., Engineering, Southern Illinois University Carbon-
vol. 25, no. 2, pp. 809–820, 2009. dale (SIUC), IL, USA. His current research interests
[17] E. Figueres, G. Garcerá, J. Sandia, F. Gonzalez-Espin, and J. C. Rubio, include power systems optimization and smart grid
“Sensitivity study of the dynamics of three-phase photovoltaic inverters security. He was a recipient of the Outstanding
with an LCL grid filter,” IEEE Trans. Ind. Electron., vol. 56, no. 3, pp. Teacher of the Year Award in the Electrical and Computer Engineering
706–717, 2009. Department of SIUC in 2019. He is currently an Associate Editor of IEEE
[18] P. Kundur, N. J. Balu, and M. G. Lauby, Power system stability and Access.
control. McGraw-hill New York, 1994, vol. 7.
[19] J. Z. Zhou, H. Ding, S. Fan, Y. Zhang, and A. M. Gole, “Impact of
Short-Circuit Ratio and Phase-Locked-Loop Parameters on the Small-
Signal Behavior of a VSC-HVDC Converter,” IEEE Trans. Power Del.,
vol. 29, no. 5, pp. 2287–2296, 2014. Wenxin Liu (Senior Member, IEEE) received the
[20] L. Zhang, L. Harnefors, and H. Nee, “Power-synchronization control B.S. degree in industrial automation and the M.S. de-
of grid-connected voltage-source converters,” IEEE Trans. Power Syst., gree in control theory and applications from North-
vol. 25, no. 2, pp. 809–820, May 2010. eastern University, Shenyang, China, in 1996 and
[21] J. Khazaei, Z. Miao, and L. Piyasinghe, “Impedance-model-based 2000, respectively, and the Ph.D. degree in electrical
MIMO analysis of power synchronization control,” Elect. Power Syst. engineering from the Missouri University of Science
Res, vol. 154, pp. 341–351, 2018. and Technology (formerly University of Missouri–
[22] A. Asrari, M. Mustafa, M. Ansari, and J. Khazaei, “Impedance analysis Rolla), Rolla, MO, USA, in 2005. From 2005 to
of virtual synchronous generator-based vector controlled converters for 2009, he was an Assistant Scholar Scientist with
weak AC grid integration,” IEEE Trans. Sustain. Energy, vol. 10, no. 3, the Center for Advanced Power Systems, Florida
pp. 1481–1490, 2019. State University, Tallahassee, FL, USA. From 2009
[23] J. Khazaei, Z. Miao, L. Piyasinghe, and L. Fan, “Real-time digital to 2014, he was an Assistant Professor with the Klipsch School of Electrical
simulation-based modeling of a single-phase single-stage PV system,” and Computer Engineering, New Mexico State University, Las Cruces, NM,
Electric Power Systems Research, vol. 123, pp. 85–91, 2015. USA. He is currently an Associate Professor with the Department of Electrical
[24] J. Khazaei, Z. Tu, and W. Liu, “Small-Signal Modeling and Analysis of and Computer Engineering, Lehigh University, Bethlehem, PA, USA. His
Virtual Inertia-based PV Systems,” IEEE Trans. Energy Convers., 2020. research interests include power systems, power electronics, and controls. Dr.
Liu is an Editor of the IEEE TRANSACTIONS ON SMART GRID, IEEE
TRANSACTIONS ON POWER SYSTEMS, and Journal of Electrical Engi-
neering Technology, and an Associate Editor for the IEEE TRANSACTIONS
ON INDUSTRIAL INFORMATICS.

Javad Khazaei (S’10 M’16 SM’20) received the


Ph.D. degree in Electrical Engineering from Uni-
versity of South Florida (USF) in 2016 with focus
on power and energy systems. He is currently an
Assistant Professor at the School of Science, En-
gineering & Technology at Penn State Harrisburg
and is also affiliated with the Architectural Engi-
neering Department at Penn State University Park.
His research interests include cyber-physical power
system modeling, control, and dynamics, microgrid
optimization, smart grid security, and power elec-
tronics applications in smart grids and shipboard microgrids.

0885-8950 (c) 2020 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission. See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
Authorized licensed use limited to: American University of Beirut. Downloaded on May 23,2021 at 10:13:32 UTC from IEEE Xplore. Restrictions apply.

You might also like