Overview of SOGI-Based Single-Phase Phase-Locked L

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/ACCESS.2021.3063774, IEEE
Access
Date of publication xxxx 00, 0000, date of current version xxxx 00, 0000.
Digital Object Identifier 10.1109/ACCESS.2017.Doi Number
Overview of SOGI-Based Single-Phase Phase-

Locked Loops for Grid Synchronization under
Complex Grid Conditions
JINMING XU1, (Member, IEEE), HAO QIAN1, YUAN HU1, SHENYIYANG BIAN1, and SHAOJUN
XIE1, (Member, IEEE)
1
College of Automation Engineering, Nanjing University of Aeronautics and Astronautics, Nanjing 211106, China
Corresponding author: Jinming Xu (xjinming01@ 163.com).
This work was supported in part by the National Natural Science Foundation of China under Grant 51807089 and Grant 51877104, and in part by the
Natural Science Foundation of Jiangsu Province under Grant BK20180432.
ABSTRACT Synchronization is the key part to ensure the high performance of grid-connected systems.
Phase-locked loop (PLL) is one of the most popular synchronizations due to its simple implementation and
robustness under certain grid variations. Particularly, in single-phase applications, PLL based on second-
order generalized integrator (SOGI-PLL) is widely used because of its simple structure, certain filtering
ability and frequency adaptability. The sensitivity of SOGI-PLL to the dc offset and low-order harmonics
has been studied a lot in the literature and many solutions have been proposed. However, as more and more
power electronic devices are integrated into the power grid recently, the grid condition becomes more
complex. As a result, the nonlinear loop coupling phenomenon of SOGI-PLL is more severe and obvious
and thus the performance of SOGI-PLL is degraded a lot, especially under the conditions of grid frequency
variations and weak grid. A popular method is to use the frequency-fixed SOGI-PLL (FFSOGI-PLL) while
how to eliminate the estimation error under frequency variations remain an important task. Though some
scattered methods have been proposed, a simple yet effective strategy is still missing. Besides, it has been
shown that the system stability margin will be reduced due to the negative-resistance behavior of PLL.
However, the models derived in the current documents cannot accurately reveal the instabilities caused by
the standard SOGI-PLL for missing the dynamics of the frequency feedback loop. Moreover, the PLL
parameters are usually designed according to the PLL bandwidth to guarantee system stability under a weak
grid. How to optimize the parameters is still unclear. In view of this, this paper further summarizes and
reviews the existing achievements of single-phase SOGI-PLL, and points out the problems to be solved and
the development direction to improve the SOGI-PLL under more complex and non-ideal grid conditions.
INDEX TERMS Grid synchronization, single-phase phase-locked loop, second-order generalized

integrator, complex grid condition.
I. INTRODUCTION stable operation and power quality of the utility grid is

Due to the increasing depletion of fossil fuels and the greatly challenged [1]-[3].
environmental problems caused by their large-scale usage, In grid-connected applications including active power
renewable energy power generation has received extensive filter, uninterruptible power supply (UPS), dynamic voltage
attention all over the world. As the penetration of renewable restorer (DVR), flexible ac transmission (FACT), high
energy sources (RESs) such as solar and wind energy voltage direct current (HVDC) transmission, controllable
continues to increase, more and more power electronic rectifier and distributed generation (DG), synchronization is
devices are integrated into the power grid. As a result, the one of the most important aspect [4]. Extracting the phase
grid tends to exhibit very complex non-ideal characteristics and frequency information from the grid voltage is of great
(e.g., dc offset, harmonics, voltage drop, phase jump, significance for the monitoring, control and protection [5]. At
frequency step or sweep and high grid impedance), and the present, scholars have proposed many synchronization
VOLUME XX, 2017 1
This work is licensed under a Creative Commons Attribution 4.0 License. For more information, see https://creativecommons.org/licenses/by/4.0/
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/ACCESS.2021.3063774, IEEE
Access
methods, such as zero-crossing detection, discrete Fourier frequency variations and harmonics [2]. Moreover, a
transform, adaptive notch filter, Kalman filter, recursive differential operator is used in the Deri-PLL, which will
weighted least squares method, artificial neural network, amplify high-frequency components and deteriorate the
frequency-locked loop and phase-locked loop (PLL) [6]-[8]. performance under harmonics and noises. The APF-PLL
Among them, the PLL technique is probably one of the most uses an all-pass filter to shift the phase of the fundamental
popular one used in grid-connected systems because of its signal by 90°, but it does not perform well in presence of the
relatively simple implementation and robustness under frequency changes and harmonics. The ideal Hilbert
different grid conditions. transform cannot be realized directly because it violates the
The PLL is a negative feedback closed-loop system, law of causality [3]. The practical Hilbert transform used is
which is mainly composed of three parts: 1) a phase detector an approximation with a finite response filter (FIR) or an
(PD), 2) a loop filter (LF), and 3) a voltage-controlled infinite response filter (IIR), and it is also affected by the
oscillator (VCO). The difference of various PLLs mainly lies changes in the grid frequency. The KF-PLL uses Kalman
in how the PD is constructed and how the phase error filtering technique to construct a pair of orthogonal signals at
information is yielded. For PLLs in three-phase applications, the cost of heavy computational burden. The Park-PLL and
a pair of orthogonal signals can be easily obtained through SOGI-PLL are two widely used single-phase PLLs that are
3s/2s coordinate transformation (i.e., Clarke transform) from immune to the change of the grid frequency, and have certain
the measured three-phase voltage signal. However, in single- filtering ability. The Park-PLL uses a positive and an inverse
phase applications, for there is only one-phase voltage signal Park transform and two low-pass filters (LPF) to form the PD,
available, the PD structure of single-phase PLLs is more but the two coordinate transformation modules not only
complicated, and many single-phase PLLs with different contribute to the difficulty of implementation, but also
performance are derived in the literature. greatly increase the nonlinearity of the PD and thus the
The simplest single-phase PLL uses a multiplier as the PD, difficulty in modeling. In addition, the cut-off frequency of
which is also known as the power-based PLL (pPLL) [9]- inner-loop LPFs of Park-PLL has significant impact on the
[12]. The PD output of the pPLL consists of a dc component dynamic performance, disturbance rejection capability, and
and an ac component. The first term is the phase error signal even the stability of the whole PLL. Hence, a compromise is
while the latter one is a double-frequency term. The double- needed [3], yet further improvement is difficult. The SOGI-
frequency component will generate oscillations in the pPLL PLL uses two first-order integrator to construct the OSG
and needs to be filtered out with an in-loop filter. However, based on second-order generalized integrator, which is easy
the introduction of the filter greatly reduces the PLL to be implemented digitally, and the nonlinearity is lower
bandwidth, resulting in a poor dynamic performance of this than that of Park-PLL.
type of PLLs. A well-known problem of the SOGI-PLL is the sensitivity
In order to solve the inherent double-frequency oscillation to the dc offset and low-order harmonics. These problems
phenomenon in the pPLL, scholars turn to the PLLs based on have been studied long ago and now many useful methods
the adaptive filter (AF-PLL) like enhanced PLL (EPLL) and are available. However, as the increasing integration of
the PLLs based on orthogonal signal generator (OSG-PLL). renewable energies and power electronic devices into the
The EPLL is a frequency-adaptive PLL and can track the power grid, some other non-ideal grid conditions like
amplitude, phase and frequency of the input signal [13]. In frequency variation and weak grid are more severe and these
the steady state, the double-frequency component is new issues can affect the performance of SOGI-PLL a lot as
counteracted by the opposite component generated by the well. To ensure the performance of SOGI-PLL under a more
EPLL. Then, extra in-loop filter is not needed, and the complex grid condition, it is necessary to comprehensively
dynamic performance is thus improved. However, the EPLL review recent advances and to point out the focus for future
has large transient fluctuations, and is highly nonlinear due to research. Note that the suppression strategies of dc offset and
the over use of sine and cosine operators, which can make low-order harmonics are mature, while the loop coupling
the optimization of parameters and system performances very phenomenon is not well addressed until now. Hence, the
difficult. In comparison, the OSG-PLLs have received more solutions about the dc and low-order harmonics are briefly
attention. talked about hereinafter, and more attentions will be paid to
For OSG-PLLs, common OSG construction methods the corresponding solutions under frequency variations and
include: T/4 (T is fundamental period) transfer delay (namely, weak grid conditions, which can be the main contribution of
the Delay-PLL) [14], first-order derivative operator (Deri- the paper.
PLL) [15], all-pass filter (APF-PLL) [16], Hilbert This study first introduces the working principles and
transformation (Hilbert-PLL) [17], Kalman filter (KF-PLL) basic characteristics of the SOGI-PLL. Then, the related
[18]-[19], inverse Park transformation (Park-PLL) [20]-[22] studies are summarized from three aspects—dc offset and
and second-order generalized integrator (SOGI-PLL) [8], low-order harmonics suppression, nonlinear loop coupling
[23]-[25]. Among them, the Delay-PLL and Deri-PLL are phenomenon and grid frequency adaptability, and robustness
simple, but their performances are greatly affected by grid in weak grid. The advantages and disadvantages of various
VOLUME XX, 2017 9
Access
SOGI Control loop
ug  ke 1 u ud  ff
ke
s  uq
u e 1  PLL
dq PI
PLL
1 s
s
PD LF VCO
Main research topics
DC offset Low-order harmonics Frequency variations Weak grid
FIGURE 1. Block diagram of the SOGI-PLL and its main research topics.
methods are also reviewed. Finally, some discussions on the Bode Diagram
20
Phase(deg) Magnitude(dB)
future research direction are presented.
0 ke=0.8 ke=2
ke=4
II. WORKING PRINCIPLES AND BASIC -20
CHARACTERISTICS OF SOGI-PLL
-40
The structure of SOGI-PLL is shown in Fig. 1, in which ug is
the input signal, ωff (=2π50 rad/s) is the central angular -60
90
frequency, ωPLL and θPLL are the estimated angular frequency 45
and phase angle, respectively. It can be seen from Fig. 1 that 0 ke=0.8
ke=2ke=4
the SOGI generates a pair of orthogonal signals (i.e., uα and -45
uβ), which are then sent to the Park transformation module -90 -1
(i.e., αβ→dq) to obtain ud and uq 10 100 101 102 103 104
Frequency(Hz)
ud   cos  PLL sin  PLL  u  (a)
    (1)
cos  PLL  u 
Bode Diagram
 uq    sin  PLL 50
Phase(deg) Magnitude(dB)
Assuming that the input signal is a pure sinusoidal in the 0 ke=0.8

ke=2
ke=4
form of ug = Vm cos θg, and that the SOGI outputs a pair of
-50
signals with the same amplitude as the input signal and a
phase difference of 90°, (1) can then be rewritten as -100
0
ud  Vm cos( g   PLL ) 
 u   V sin(   )  (2) -45
ke=0.8
 q  m g PLL  -90 ke=2
ke=4
-135
It can be observed from (2) that ud includes the amplitude
information of the input signal while uq includes the phase -180 -1
10 100 101 102 103 104
error information. By using the PI regulator as the LF to Frequency(Hz)
(b)
control uq to zero and sending it to the VCO (represented by
an integrator), the estimated phase angle θPLL can be equal to FIGURE 2. Bode plots of the transfer functions of the SOGI block
that of the input. for different values of ke. (a) Gα(s); (b) Gβ(s).
According to the Mason formula, the transfer functions
from ug to uα and uβ can be deduced as where, in the steady state, ωPLL is 2π50 rad/s. According to (3)
u ( s ) kePLL s and (4), the Bode diagrams of Gα(s) and Gβ(s) when ke takes
G ( s)   (3) different values are drawn in Fig. 2.
u g ( s ) s 2  kePLL s   PLL
2
As can be observed from Fig. 2, Gα(s) behaves as a band-
u ( s ) kePLL
2
pass filter (BPF) whose center frequency is equal to ωPLL.
G (s )   (4)
ug ( s) s 2  ke PLL s  PLL
2 The amplitude gain of Gα(s) at ωPLL is unity and the phase lag
is zero. Besides, Gβ(s) appears as a low-pass filter whose cut-
off frequency is ωPLL as well. The amplitude gain of Gβ(s) at
VOLUME XX, 2017 9
Access
ωPLL is unity, and the phase lag is 90°. When ke takes When the input signal contains a dc component, using a filter
different values, the bandwidth of the system changes, but is probably the most direct and effective method. Possible
the amplitude gain and phase difference of Gα(s) and Gβ(s) at options are: sliding discrete Fourier transform (SDFT) [30],
ωPLL always remain the same. It can be concluded that the delayed signal cancellation (DSC) operator [31], moving
output of the SOGI module is a pair of quadrature signals average filter, notch filter (NF) [32], and complex coefficient
with the same amplitude as the input signal and a constant filter (CCF) [33].
phase difference of 90°, and that the SOGI-PLL has a certain Note that the core reason of the sensitivity to the dc offset
ability to suppress high-frequency disturbances. In addition, in SOGI-PLL lies in the low-pass feature of Gβ(s), implying
as shown in Fig. 1, the SOGI-PLL brings the estimated that reducing the dc gain of Gβ(s) may be another option to
angular frequency ωPLL back to the SOGI module to serve as suppress the dc offset [34]. In [35], two SOGI blocks are
the resonant frequency for SOGI (denoted as ωr). In steady connected in series (named the CSOGI-PLL) to reject the dc
state, ωPLL is equal to the frequency of the input signal, which offset, and the dc gain of the new Gβ(s) decreases notably.
makes the SOGI-PLL a frequency adaptive PLL. Similar to [35], an all-pass filter is added behind the SOGI to
However, it can also be found from Fig. 2(b) that Gβ(s) has obtain uβ [36]. In [37], a second-order SOGI-PLL (SO-SOGI-
almost no ability to suppress the dc components, indicating PLL) with a fourth-order transfer function is proposed, where
that the SOGI-PLL is sensitive to the dc offset in the input the two first-order integrators in the typical SOGI-PLL are
signal. Besides, reducing ke to attenuate the characteristic replaced by their second-order forms.
low-order harmonics in the power grid can lead to a very 2) DC ESTIMATION AND SUBTRACTION METHODS
poor dynamic response. To solve the tradeoff between the The second kind of methods to eliminate the dc component is
speed and accuracy, an extra filtering stage is usually utilized to estimate the dc offset and then send it back to either the
under distorted cases for the PLL. Moreover, the estimated input side [27], [38] or the SOGI output (i.e., uα and uβ) [39]-
angular frequency is fed back into the SOGI to ensure the [40] with a subtraction. The structures of three advanced
frequency adaptability, but an additional feedback loop is solutions proposed in [27], [39]-[40] are presented in Fig. 3.
also added in the SOGI-PLL. In other words, nonlinear
3) NON-DC SIGNAL EXTRACTION METHODS
coupling between the SOGI and the PLL control loop (from The third kind of dc elimination methods is to examine the
Park transform to θPLL) appears and makes the modeling and transfer function from each signal point to the output of the
parameter design complicated. Furthermore, in the weak grid, SOGI module, and then choose the signal that is not affected
the negative-resistance behavior of PLL can reduce the by the dc offset to create the new uβ.
system stability margin [26], while the unique loop coupling Two equivalent methods are proposed in [41] and [42] and
problem of SOGI-PLL will become more prominent so that their structures are shown in Fig. 4. As shown in Fig. 4, a
the estimation accuracy is seriously affected. The main signal is drawn from keε−uβ, which can be used as a new uβ
research topics of the SOGI-PLL are shown in Fig. 1 and after inverting. The transfer function of the new Gβ(s) is
details will be described below.
u (s ) ke s 2
G ( s)   (5)
III. SUPPRESSION OF THE DC OFFSET AND LOW- ug ( s) s 2  ke PLL s  PLL
2
ORDER HARMONICS
Two zeroes appear in the new Gβ(s), which enhance the dc
elimination by sacrificing the ability of suppressing the high-
A. SUPPRESSION OF THE DC OFFSET
frequency disturbances.
A dc offset can be easily generated by many reasons, like
grid faults, saturation phenomenon of the voltage/current
B. SUPPRESSION OF LOW-ORDER HARMONICS
sensor, dc injection of the distributed power generation
Due to the extensive use of various nonlinear loads, the grid
system, and the AD conversion process of fixed-point digital
voltage is often polluted with a large number of background
signal processor [27]-[28]. The presence of the dc component
harmonics, which cause certain fluctuations in the frequency
will result in oscillations at the fundamental frequency in the
and phase angle estimated by the PLL [43]. Correspondingly,
estimated frequency and phase angle of the PLL and thus
the current reference generated by the PLL as well as the grid
cause the dc injection problem in the grid current since the
current will be distorted.
current reference is generated by the PLL. Hence, many grid
The commonly-used solution to suppress harmonics is to
codes strictly restrict the dc value in the grid current. For
reduce the PLL bandwidth. However, this method sacrifices
example, in the standard IEC 61727, the dc offset is required
the dynamic performance and is not effective when the grid
to be less than 1% of the rated current [29].
voltage contains large amount of low-order harmonics [44].
To address the problem of the dc offset in single-phase
Existing solutions can be divided into three categories:
PLLs, the existing solutions in the literature can be divided
into three categories: 1) PRE-FILTERING METHODS
The first type of method is to place a filtering stage before
1) DIRECT FILTERING METHODS
the input signal passes into the PLL (i.e., pre-filtering) [45]-
[46]. This method, however, works well only when the grid
VOLUME XX, 2017 9
Access
1 ug ke  1 u
k  ke
s s
ug ke 1 u PLL u
 ke
s 1
u
PLL s
1
s 1
(a)
(a)
ug
ug  ke 1 u
 ke 1 u ke
ke s
s PLL u
PLL u
LPF
1
1
s
s
(b) (b)
FIGURE 4. SOGI structures based on signals without dc offset. (a)
1 u3 Scheme proposed in [41]; (b) Scheme proposed in [42].

s
SOGI-PLL is proposed to enhance the low-order harmonics
ug u rejection ability of the SOGI.
 ke  1 u1
ke At the same time, some schemes reviewed in Section III-A
s u
can also be applied here to suppress the low-order harmonics
PLL u2
1 for the SOGI-PLL [35]-[37].
s
IV. NON-LINEAR LOOP COUPLING PHENOMENON AND
(c) GRID FREQUENCY ADAPTABILTY
FIGURE 3. SOGI structures based on dc offset estimation. (a) As Fig. 1 reveals, the SOGI-PLL has two interdependent
Scheme proposed in [27]; (b) Scheme proposed in [39]; (c) Scheme feedback loops: one feeds the estimated phase angle θPLL
proposed in [40].
back to the Park transform to obtain the phase error signal uq,
while the other feeds the estimated angular frequency ωPLL
frequency is the nominal value. Otherwise, a phase delay is back to the SOGI to dynamically adjust the resonant
unavoidable and an additional phase angle compensation unit frequency ωr for the SOGI. In other words, the SOGI and the
or frequency adaptability link needs to be added [47]. control loop (also known as synchronous-rotating-frame PLL,
SRF-PLL) are coupled together by this frequency feedback
2) IN-LOOP FILTERING METHODS
path and thus the performance of SOGI will be affected by
The second type of methods is to include an appropriate filter
the SRF-PLL. That is, however, a disturbance in either the
within the control loop of the PLL (i.e., in-loop filtering).
SOGI or the SRF-PLL will influence each other. Note that
Combined with the PI regulator, the disturbance suppression
the Park transform is a highly nonlinear part in the PLL so
ability is improved a lot, but adverse effects are drawn on the
that the introduction of these feedback loops further increases
stability and dynamic performance of the PLL itself [48].
the nonlinearity of the SOGI-PLL. As a result, the frequency
3) OSG MODIFICATION METHODS adaptability is maintained with the feedback of the estimated
The third category is to enhance the filtering capability angular frequency ωPLL but at the cost of stability margin.
through the OSG modification of the PLL. For SOGI-PLL, Similar results can be found in [49] as well. Moreover, the
its filtering capability comes from both the SOGI and the PI SOGI and PLL control loop need to be considered together,
regulator. The PI regulator is located in the control loop of which increases the complexity and difficulty in both the
the PLL, which directly affects the stability and estimation modeling and parameter design.
accuracy. In other words, the PI ensures the PLL control loop It is noted that, in the aspect of digital implementation, for
to be a second-order system, which can achieve zero error in the fixed frequency-based SOGI-PLL (where a fixed value of
steady state under conditions of both phase and frequency grid frequency is sent back to SOGI, named as FFSOGI-
jump. Without changing the structure of PLL control loop, PLL), the computational burden is greatly reduced because
the SOGI module can be modified to enhance the ability of the frequency of SOGI is not updated [50]. Meanwhile, the
harmonics suppression. In [37], based on this idea, the SO- FFSOGI-PLL can achieve larger bandwidth, better stability
margin and faster dynamic performance.
VOLUME XX, 2017 9
Access
u ud  ff
ug 
SOGI u uq e PLL 1  PLL
dq PI
s
Frequency r
Detector
FIGURE 5. Block diagram of the FFSOGI-PLL based on parallel frequency detector.
ug  ke 1 u ud
 ff
s  uq
u e PLL 1  PLL
ke  ke dq PI
2cos 1 s
q
s
q
n n
n1/ q
FIGURE 6. Block diagram of the FFSOGI-PLL based on fractional-order conformal mapping.
In view of this, some interests are attached to the study of The main drawback of this method is the use of memory
FFSOGI-PLL. However, if the fixed resonant frequency of buffers to process the samples for the DFT operation.
the SOGI is not equal to the actual grid frequency, the In [52], Sun et al. combine the frequency-locked loop
estimated frequency of the PLL will contain a sinusoidal (FLL) together with the SOGI-PLL to realize the adjustment
ripple at twice the grid fundamental frequency. The content of the SOGI resonant frequency, named as SOGI with FLL
of this second-order harmonic is closely related to the PLL (SOGI-WFLL). It should be noted that the FLL is actually a
bandwidth [35]. If the PLL bandwidth is increased larger, the gradient estimator (GE) and three different GEs are proposed
total harmonic distortion (THD) of the current reference for the SOGI to estimate the fundamental frequency of the
generated by the PLL can be larger and the current quality of input in [53]. However, when the input signal contains the dc
the grid-connected converter is deteriorated. On the contrary, offset and harmonics, there will be certain fluctuations in the
if the bandwidth is small, the settling time may be long and detected frequency, so that the accuracy of the PLL is
the transient performance is poor. Such case is undesirable in affected. To reject the dc or harmonics, the methods in
some cases like the low-voltage ride through (LVRT). Section III or the multiple SOGI strategy can be used as a
In order not to affect the PLL bandwidth and to eliminate supplement. In addition, the proportional coefficient of the
or reduce the negative impact of the grid frequency variation FLL has to be carefully chosen by the tradeoff between the
on the FFSOGI-PLL, scholars have proposed many solutions, detection speed and disturbance suppression capability.
including parallel frequency detector based methods, phase In [54], a method based on the modulating function (MF)
error compensation methods and PLL structure modification is used to detect the frequency. Since the product of the MF
based methods. and the derivative of the input signal can be integrated into
the product of the derivative of the MF and the input signal
A. CATEGORY 1: PARALLEL FREQUENCY DETECTOR for integration, the high-frequency harmonics amplification
BASED PLL METHODS of the input can be avoided. Moreover, the integral operation
By removing the original frequency feedback and connecting has low-pass filtering effect naturally so that this method is
a frequency detector in parallel to the SOGI, the real-time not sensitive to harmonics and noise. What should be taken
frequency for the SOGI can be updated. Obviously, there is into serious consideration is the selection of the window
almost no coupling within the SOGI-PLL once the frequency length or the forgetting factor value of the MF.
feedback loop is replaced by the extra frequency detector. In [55], an approach based on fractional-order conformal
The basic structure of this method is shown in Fig. 5. mapping is adopted to construct the SOGI and is named as
In [51], the fixed window DFT is adopted to obtain the AFFSOGI-PLL. Its block is given in Fig. 6. The proportional
fundamental frequency. Though the DFT technique cannot gain ke and the resonant frequency ωr of the SOGI can be
get an accurate result under time-varying cases due to the adjusted in real time by changing the fractional-order q to
inherent spectral leakage and picket fence effect, such feature realize the frequency-adaptive feature for FFSOGI-PLL. The
of the DFT is used to track the frequency of ug adaptively. expression of φ / q is expressed as,
VOLUME XX, 2017 9
Access
 4  ke2 ker  g
   tan 1 ( ), 0  ke  2 G ( j g ) = (9)
 ke (6) (ker  g ) 2  (r2   g2 ) 2
q 
  , ke  2 r2  g2
G ( j g )=arc tan (10)
The AFFSOGI-PLL inherits the simplicity of FFSOGI- ke  r  g
PLL and at the same time does not increase the system order.
Seen from (7) and (8), if ωr = ωg, the estimated phase
However, these features depend on the accurate value of q
angle is exactly the same as the input. However, a steady
which is determined by the deviation of the grid frequency. It
is hence indicated that the AFFSOGI-PLL requires an extra state phase error is unavoidable in the PLL output if ωr  ωg.
frequency detection link, but is not detailed in [55]. In the following, two methods that compensate the phase
In [56], Shamim Reza et al. use a linear Kalman filter difference are discussed.
(LKF) to obtain a pair of orthogonal signals which are then Define
sent into the FLL to estimate the frequency and amplitude of u
 error  arc tan( )  ( g t    G ( j g ) (11)
the input. This method is not sensitive to noise. However, the u
method proposed in [56] needs to know the initial value of With some mathematical manipulations [59], θerror can be
the process noise, measurement noise and error covariance expressed as
matrix in advance, and the initial estimation state also needs
 error 
to be given manually. As a result, the implementation should
be carefully dealt with. g
 1) sin 2( g t    G ( j g )
(
The Teager energy operator (TEO) is introduced in [57] r (12)
with only five consecutive samples and is easy to implement arc tan
g
digitally. However, the TEO is shown to be sensitive to the 2(  1)[1  cos 2( g t    G ( j g )]
frequency variation and harmonics in the input signal. Hence, r
the input signal must be pre-filtered first, and the frequency Seen from (12), θerror occurs as a second-order harmonic
estimated by TEO is suggested to be fed back to the pre- oscillation if ωr  ωg. Hence, once the oscillation caused by
filtering stage to avoid any phase offset errors caused by grid θerror is filtered out, the fundamental frequency can be easily
frequency variations. estimated with a differential operator.
In summary, the parallel frequency detector based methods Hence, in [59], Shamim Reza et al. propose to use a SOGI
require a fast and accurate frequency detection. A simple yet whose resonant frequency is larger than the fundamental
effective solution is still ongoing. If the frequency detector frequency, and then compensate the errors of the estimated
has no filtering ability, the input signal must be pre-filtered, frequency and amplitude according to the frequency behavior
which means additional complexity in the implementation. It of SOGI. Indeed, the dynamics is improved, for the settling
should be also noted that if the frequency of the input signal time is inversely proportional to the product of ke and ωr.
changes too fast, the resonant frequency of SOGI can keep However, this method needs to use the phase unwrapping
on changing, which can eventually cause considerably large technique and a differential filter, which can be complicated.
transient fluctuations in the quadrature signal output by SOGI The computational burden is heavy. Furthermore, the phase
and distortion in the grid current reference [58]. angle cannot be estimated directly.
In [60], a different FFSOGI-PLL with using a phase error
B. CATEGORY 2: PHASE ERROR COMPENSATION compensation (SOGI-WPEC) is proposed and its structure is
BASED PLL METHODS shown in Fig. 7. It is analyzed that uα and uβ have a constant
The second kind for frequency adaptability of FFSOGI is to phase difference of 90° at different frequencies while the
take the grid frequency deviation into consideration, and then amplitude difference relates with the ratio of ωr/ωff. Then, if
to compensate for the generated phase difference under the the amplitude of uα and uβ is controlled the same, there will
off-nominal condition. be no second-order harmonic oscillations in the steady state.
Assuming a pure sinusoidal grid voltage, i.e., ug = Vm sin However, seen from (7) and (8), there is a phase difference
(ωgt+Φ) = Vm sin θg, with the help of (3) and (4), the steady between the real phase angle θg and that of uα, which can
state output of uα and uβ can be obtained cause errors in the estimated phase angle. Hence, the core
u  Vm G ( j g ) sin( g t    G ( j g )) (7) issue is to compensate the phase error at different frequencies
accordingly. Compared with the typical SOGI-PLL, this PLL
r
u  Vm G ( j g ) cos( g t    G ( j g )) (8) does not provided the significant improvement in the view of
g dynamic performance and harmonic suppression capability,
where but do have a better dc offset suppression capability [58]. In
addition, unlike the SOGI-PLL, the FFSOGI-PLL proposed
in [60] may be unable to estimate the grid voltage amplitude
accurately under off-nominal frequencies.
VOLUME XX, 2017 9
Access
ug u ud  ff
SOGI u

0 u 1 uq e PLL 1 1  PLL
dq PI
s

1  2
PLL  2
0
 ff kePLL0
FIGURE 7. Block diagram of the FFSOGI-PLL based on phase error compensation.
It is pointed out in [61] that an inverse compensation filter SOGI-PLL based on the differential elements (DE), named
can be added after the pre-filter to offset the error caused by the DE-PLL or SOGI with DE (SOGI-WDE). The structure
the frequency offset. However, the frequency response of this of DE-PLL is shown in Fig. 8. This PLL is not affected by
method near the fundamental frequency is relatively smooth, grid frequency changes and does not introduce a frequency
and the implementation of the inverse filter may be a little feedback loop into the OSG. It should be noted that the
difficult because it is usually not a canonical system [62]- amplitude of the OSG output (i.e., y1 and y2 in Fig. 8) is not
[63]. At present, there is no literature attempting to apply the equal to the amplitude of the input signal, but the amplitude
inverse filter into the FFSOGI-PLL under frequency of y1 and y1f, y2 and y2f are the same in the steady state,
variations. respectively. Consequently, the product of y1 and y2f, y2 and
To sum up, the key of the phase error compensation based y1f can be used to generate the phase error signal and the
methods is to find a proper way to accurately calculate the estimated phase angle. It is revealed that the DE-PLL can be
difference between the actual value and nominal value of the transformed into a typical FFSOGI-PLL form whose phase
estimated phase angle under different working frequencies. It angle feedback loop is equivalent to adding a SOGI into the
is noted that [58] and [60] examined the phase differences feedback path [58]. Hence, from the perspective of small
under some approximations (for instance, the assumption of signal modeling, the DE-PLL is equivalent to the traditional
ωPLL=ωff), which can have a great influence on the accurate SOGI-PLL.
compensation. If the bandwidth of FFSOGI-PLL is small
(e.g., in [60], the PLL bandwidth is only 32 Hz), it can be D. DISCUSSIONS ON FREQUENCY VARIATIONS
approximately considered that this assumption holds and the Generally speaking, in order for the FFSOGI-PLL to be
amplitude of uα is nearly the same as that of uβ1. In other frequency-adaptive, most of the methods mentioned above
words, the nonlinearity caused by feeding back the estimated have to add more algorithms which are usually a little
frequency ωPLL still exists but can be neglected under a small complex. Here, to understand the performances of different
bandwidth condition. But, for a higher bandwidth, ripples in methods, three relatively simple methods are selected for
ωPLL can turn larger and the amplitude of uβ1 can have some further comparisons, i.e., SOGI-WFLL [52], SOGI-WPEC
oscillations which affecting the compensation accuracy. [60] and SOGI-WDE [64]. Parameters follow the design
Therefore, under high bandwidth conditions, the phase angle methods in [58] and four cases are used to test their
error compensation still needs further study. performances:
1) Case I: grid frequency steps from 50 Hz to 51 Hz and
C. CATEGORY 3: PLL STRUCTURE MODIFICATION then to 55 Hz;
BASED METHODS 2) Case II: grid frequency sweeps from 50 Hz to 52 Hz at
Contrary to the methods mentioned in Section IV-A and IV- the rate of +10 Hz/s;
B, the third category adjusts the structure of SOGI to obtain 3) Case III: grid voltage sags from 1.0 p. u. to 0.6 p. u.;
the necessary phase error information. 4) Case IV: grid voltage contains 3rd (5%), 5th (4%), 7th (3%)
In [64], inspired by the idea of [60], Guan et al. propose a harmonics, and the total harmonic distortion is 7.07%.
The simulation is carried out in MATLAB/Simulink and
y1 ud  ff corresponding results are shown in Fig. 9. As can been seen
ug
DE
 uq from Figs. 9(a) and (c), the SOGI-WFLL and SOGI-WDE
y2 e PLL 1  PLL
dq PI behave larger overshoot than the SOGI-WDE. Besides, Fig.
s
9(d) indicates that both SOGI-WFLL and SOGI-WDE are
y1 f y2 f
sensitive to the harmonic contents in the grid voltage. As for
DE cos () the SOGI-WDE, it has a relatively nice performance but an
oscillation at twice the grid fundamental frequency occurs in
FIGURE 8. Block diagram of the FFSOGI-PLL based on derivative the PLL output, which is especially obvious if the difference
elements.
between the real grid frequency and the nominal frequency is
VOLUME XX, 2017 9
Access
(a)
(b)
(c)
(d)
FIGURE 9. Simulation results of (a) Case I, (b) Case II, (c) Case III, (d) Case IV.
large. One interesting thing is that these PLLs have almost In summary, the FFSOGI-PLL is superior to the typical
the identical performance in tracking a signal with frequency SOGI-PLL in terms of control design, dynamic performance
sweeping. and stability margin due to the absence of the coupled loops,
but it is greatly affected by the grid frequency variations. The
VOLUME XX, 2017 9
Access
TABLE 1. Advantages and disadvantages of different frequency adaptability strategies for the FFSOGI-PLL.
Type of the
Example Advantages Disadvantages Future direction
FFSOGI-PLL
 Resource-consuming
Ref. [51]
 High filtering effect  Trade-off of the window
DFT-based PLL
size
Ref. [52]  Easy to implement  Limited filtering effect
Combination of FLL and  Only one parameter is  Trade-off of the frequency
PLL introduced in the PLL estimation parameter design
 Inherent low-pass  Resource-consuming
Ref. [54]
characteristic of the MF  Selection of the forgetting
MF-based PLL Fast, accurate and relatively
Category 1  Fast tracking speed factor for the MF
simple frequency detector
 Need information of the
Ref. [55]  Very easy to implement
frequency deviation
AFFSOGI-PLL  Adaptively adjust SOGI
 No extra filtering effect
 Prior knowledge of the LKF
Ref. [56]  High filtering effect
initial conditions
LKF-based PLL  Immune to the dc offset
 Resource-consuming
Ref. [57]  Fast tracking speed  High computational burden
TEO-based PLL  Easy to implement  Sensitive to harmonics
Ref. [59]  No feedback loops  Cannot estimate the phase
SOGI with frequency  Unconditionally stable angle directly
error compensation  High filtering effect  Resource-consuming
Category 2 More accurate PLL model
Ref. [60]
 Not accurate under high
FFSOGI-PLL with phase  Easy to implement
PLL bandwidth conditions
error compensation
New simple OSG with high-
Ref. [64]  Low computational
Category 3  Sensitive to harmonics frequency disturbance
DE-based PLL burden
rejection
merits and drawbacks of three kinds of solutions listed in this With the rapid development of distributed power generations,
section are summarized in Table 1. The principle of the the proportion of REPG is increasing year by year. Since the
parallel frequency detector methods is simple. The additional RESs are usually located far away from the main grid, the
frequency detector does solve the influence of the coupled impedance of the transmission lines becomes non-negligible
feedback loops, but increases the implementation cost. And, and the utility grid in remote areas tends to exhibit weak grid
if the frequency detector does not have any filtering ability, characteristics. In weak grid cases, the PLL and the grid are
an additional pre-filter is needed, which further complicates coupled through the grid impedance at the point of common
the implementation. The solutions based on the phase error coupling (PCC). Usually, the PCC voltage contains a large
compensations do not require additional frequency detector, number of harmonics, which seriously affects the estimation
but the accuracy of its output phase angle highly relies on the accuracy of the PLL [65]-[69]. Meanwhile, previous studies
accuracy of modeling and error analysis. The third kind of have revealed that the PLL can exhibit negative-resistance
methods (i.e., the modified PLL structure), requires neither characteristics within its bandwidth, resulting in significant
the additional frequency detector nor the accurate calculation phase reductions of the inverter output impedance. Thereby,
of phase angle errors, so that it might be a very promising the negative impact introduced by PLL would deteriorate the
solution. But the key of utilizing this solution is to ensure that stability of the whole system [70]-[73]. However, the current
the output of the phase detector is a pair of signals with a research on the instability problems with PLL taken into
precise phase difference of 90° (note that the amplitude may consideration in the weak grid mainly focuses on the three-
be different from the input) and can suppress high-frequency phase systems, yet the research on single-phase application is
components as well. Further research on the PLL structure relatively few [74]-[75].
modification methods can be necessary.
A. MODELINGS OF THE SOGI-PLL
V. ROBUST PARAMETER DESIGN UNDER WEAK GRID To analyze the influence of PLL on system stability, the PLL
CONDITIONS model must be derived first. The traditional modeling method
VOLUME XX, 2017 9
Access
is small-signal linear modeling. A well-known model of the

 ff
SOGI-PLL is proposed in [8] and the structure is shown in
Fig. 10, where kpd = Vm is the static PD gain of the SOGI- g k pd e  PLL 1  PLL
PI
PLL and τp= 2/(ke*ωPLL) is the time constant [76]. The  ps 1 s
transfer function of the PI regulator is kp + ki/s.
In Fig. 10, the open-loop and closed-loop transfer function
FIGURE 10. The conventional small-signal model of the SOGI-PLL.
from θg to θPLL can be expressed as
k pd ( k p s  ki ) FFSOGI-PLL, so the theoretical analysis obtained using the
Gol (s )  (13)
s ( p s  1)
2
existing model (ignoring the frequency feedback loop) may
be inconsistent with the actual results, which is especially
Gol ( s) k pd (k p s  ki )
Gcl ( s )  = (14) obvious in case of large grid impedance. As can be seen from
1+Gol (s ) ( p s  1)[ s 2  k pd ( k p s  ki )] the work presented in [49], the SOGI-PLL and FFSOGI-PLL
Seen from (13), two poles appear at the origin, making the do exhibit quite different characteristics under the weak grid.
model shown in Fig. 10 a typical type-II control system. However, at present, the research on the accurate modeling
In early studies, for the sake of design simplicity, the PD of SOGI-PLL is still scarce.
dynamic of SOGI-PLL is usually neglected. Correspondingly, It is pointed out in [80] that when the grid impedance or
τp is viewed as zero, which makes the model shown in Fig. the PLL bandwidth is large, the frequency coupling effect of
10 a typical second-order system. Hence, (14) is rewritten as the PLL cannot be ignored anymore and the small-signal
a typical second-order system transfer function, i.e. model at the fundamental frequency is no longer applicable.
k pd (k p s  ki ) Therefore, in [80], a new modeling method of FFSOGI-PLL
2n s  n2
Gcl ( s )   (15) based on the extended harmonic linearization method has
s  k pd ( k p s  ki )
2
s  2n s  n2
2
been proposed by considering the frequency coupling effect.
where ξ is the damping ratio and ωn is the natural frequency. Though the precision in stability prediction is improved, this
For SOGI-PLL, the typical type-II control system model is model is a multi-input and multi-output (MIMO) system. As
more accurate than the typical second-order system model. a result, some secondary components need to be ignored and
However, both models only establish the transfer function the matrix order should be selected appropriately. A similar
from the phase angle of the input signal (i.e., θg) to the study using harmonic signal-flow graph has been done in [81]
estimated phase angle of the PLL (i.e., θPLL). The relationship to investigate the frequency coupling under different
between the input signal and the PLL output (i.e., the current bandwidth for the FFSOGI-PLL.
reference) is not revealed, which is of course more important In addition, there are also some studies concerning the
in the modeling and stability analysis of the entire inverter- modeling methods under the large disturbance (for the three-
grid system in the presence of large grid impedance. In view phase SRF-PLL, while studies for the single-phase SOGI-
of the above drawback, the perturbation and linearization PLL are lacking). The existing methods include the
method (also known as harmonic linearization) is proposed equivalent area criterion (EAC), the phase portrait, the
and the transfer function from the input signal perturbation to Lyapunov stability analysis and the quasi-static large-signal
the PLL output perturbation can be established [74]. Then, by analysis based methods [82]. These methods mainly take care
applying the impedance-based stability criterion, the of the characteristics of the PLL loop, while the inverter is
influence of the PLL on the system stability can be analyzed. modeled as a current source and the dynamic characteristics
Based on the perturbation and linearization method, the of the current loop and the output filter are ignored [83].
small-signal model of the Delay-PLL is derived in [74], and Hence, deficiencies are apparent of the modeling methods
the impedance-based stability criterion is used to analyze the mentioned above in comprehensively studying the stability.
influence of current reference amplitude, power factor angle
and PLL bandwidth on the system performance. In [77], the B. PARAMETER DESIGN IN WEAK GRID CASES
SOGI-PLL is regarded as the cascaded connection of SOGI The parameters of PLLs are often designed under ideal grid
and Delay-PLL, and the authors admit that the model might conditions. Three parameters (i.e., ke, kp and ki) of SOGI-PLL
only be accurate when the considered frequency is smaller should be designed. The traditional trial-and-error method is
than the grid frequency. Then, in [78], Zhang et al. uses the very simple, yet a lot of trials and errors are needed to obtain
perturbation and linearization method to further derive the a relatively satisfactory result. Currently, the parameters are
small-signal models of the Delay-PLL and SOGI-PLL. mainly designed depending on the established PLL models as
However, the studies mentioned above have all ignored reviewed above.
the loop coupling characteristics of SOGI-PLL. In another The typical second-order system model in (15) are widely
word, the influence of the frequency feedback loop is not used to design parameters for PLLs, by ignoring the dynamic
considered, and the feedback frequency ωPLL is considered to characteristics of SOGI. Then, only two parameters need to
be its steady-state value during the modeling process [79]. In be designed: one is the damping coefficient ξ and the other is
other words, the SOGI-PLL is equivalently treated as the the natural frequency ωn. Usually, choose ξ as 0.707 to
VOLUME XX, 2017 9
Access
balance the settling time and overshoot. In [84], Harnefors et disturbance in the PLL control loop. Under large disturbance
al. recommend choosing ωn to be one-tenth of the cut-off cases, the estimation accuracy of the phase angle is
frequency of the current control loop from the perspective of deteriorated, or even causes oscillations. On the other hand,
avoiding interaction between the PLL and the current control. the integral coefficient ki has low-pass characteristic in nature,
In [85], since a pre-filtering DSC is adopted, the in-loop which is useful to suppress high-frequency disturbances and
disturbance suppression ability is not the main concern to is conducive to the system stability. However, a very small
design the PLL bandwidth. Hence, according to the design value of ki may reduce the damping factor of the PLL itself,
goal of optimal dynamic performance, ωn is recommended to and also have a negative effect on the system stability. In
be ten times of the fundamental frequency. Then, based on other words, a too large or too small PLL bandwidth is both
the above guidelines, the PI parameters are obtained, while harmful to the system stability, but the relevant mechanism
the proportional coefficient ke for the SOGI (which is ignored needs further research.
in the PLL model) can be selected according to either the
minimum settling time criterion (ke = 1.57) [50] or the best C. DISCUSSIONS ON WEAK GRID APPLICATIONS
damping ratio (ke = 1.414) [32]. In summary, the existing modeling and parameter design
To be more accurate, as can be seen from (13), the model methods for the SOGI-PLL still have some shortcomings, as
of Fig. 9 is a typical type-II control system. The commonly summarized in Table 2. Traditional PLL model shown in Fig.
used design method for such system is the symmetrical 10 cannot describe the PLL characteristics under the weak
optimum method whose main idea is to obtain the maximum grid condition and large signal disturbances. The perturbation
phase margin at the cut-off frequency [86]-[88]. The and linearization model does fill in the gap of modeling, but
parameters are usually determined by the requirements of such method needs to be further revised when either the grid
stability margin, transient speed and harmonic suppression impedance or the PLL bandwidth is large. Moreover, there
ability in [8], [12]. In particular, in [50], ke and the PI are few studies that consider the loop coupling characteristics
parameters of FFSOGI-PLL are designed according to the of SOGI-PLL in the process of modeling, and the accuracy of
criterion of the shortest SOGI settling time and the output these models are difficult to guarantee if the PLL bandwidth
unit vector (i.e., sin θPLL and cos θPLL) distortion less than 1%. is large. On the other hand, for the design, the existing
However, when the grid impedance is non-negligible, the research mainly focuses on the selection of PLL bandwidth
parameters designed above may not meet the requirements of under the weak grid, and only qualitatively stays at the level
robustness. Some documents have noticed this problem and of designing the appropriate bandwidth to ensure the stable
conducted useful research on the parameter design of PLL in operation of the system under the weak grid. There is still
the weak grid. In [66], Zhu et al. use the type-II control lacking of unified approach for the selection of robust
system model of the PLL to establish the small-signal model parameters for SOGI-PLL (with accurate modeling) under
of grid-connected converter, and delineate the satisfactory the weak grid condition, and the mechanism of the influence
region to select parameters for the current loop and PLL by of each parameter on the system stability should be further
considering the gain margin (GM) and phase margin (PM) of explained. Therefore, the accurate modeling and parameter
the converter loop gain and the short circuit ratio (SCR). In design considering the loop coupling of the SOGI-PLL will
[89], in the case that the current loop bandwidth is firstly be the focus of further research.
chosen and the expected stability margins are set, the energy One thing should be noted is how to guarantee the system
system analysis consortium (ESAC) criterion is applied to stability with SOGI-PLL as the synchronization unit under
determine the forbidden area of parameters in order to select a weak grid. Reducing the PLL bandwidth is a conventional
the appropriate PLL bandwidth for large grid impedance. method but leads to a sluggish dynamic response. Actually,
The studies mentioned here all consider the influence of the FFSOGI-PLL shows a more robust performance than
the PLL on the current loop, but they are all for the three- the SOGI-PLL [49]. However, the current FFSOGI-PLL
phase SRF-PLL. In [68], the small-signal model of a single- can be very complex if complex grid conditions (including
phase PLL (i.e., Delay-PLL) is established with the frequency variation, harmonics and grid impedance) are
perturbation and linearization method and the parameters of considered simultaneously. Another possible option to
the Delay-PLL is designed. In [77], it is pointed out that with improve the robustness is to modify the traditional SOGI-
the same PI parameters, the SOGI-PLL is more robust with PLL. Adding a delay or inertial unit in the frequency
higher stability margin than the Delay-PLL. But, due to the feedback path to obtain a slow frequency adaptation has
loop coupling of SOGI-PLL, the current research on been confirmed effective in [81]. Hence, improvements are
parameter optimization for SOGI-PLL under the weak grid is needed and will be one of the future research direction.
still challenging.
Furthermore, in [59], it is found that the two parameters (kp VI. BRIEF REVIEW OF SOME OTHER GRID
and ki) of the PI regulator in SOGI-PLL have opposite effects SYNCHRONIZATION METHODS
on estimation accuracy and system stability. The proportional
coefficient kp has no filtering effect and can amplify the A. SOME RECENT GENERALIZED INTEGRATORS
VOLUME XX, 2017 9
Access
TABLE 2. Merits and drawbacks of different models for the SOGI-PLL.
Model type Merits Drawbacks Design method
 The OSG is considered as a

 Simple pure gain (steady state gain) Bandwidth-oriented method (the
Typical second-order system
 Can be used to design the PI  Effect of the proportional damping ratio is usually selected
model [84], [85]
regulator parameters coefficient ke in SOGI is not as 0.707)
included
 The OSG dynamics is
 Can be used for systematically
Typical type-II control simplified as an inertia unit
parameter design Symmetrical optimum method
system model [8], [50]  Cannot be used in system
 Effect of ke is considered
impedance modeling
 Ignore the frequency feedback Bandwidth-oriented method (the
 OSG dynamics is considered
Perturbation and linearization effect PLL model is simplified in the
model [74], [77], [78]  Can be used to analyze the PLL low-frequency range, i.e.,
 Ignore the high-frequency
effect in weak grid cases several hundred Hz)
coupling components
 System order is too high
Extended perturbation and  Take frequency coupling effect Not mentioned
linearization model [80] into consideration  Still do not consider the
frequency feedback effect
The concept of generalized integrator (GI) is introduced in The first method extracts the phase angle, frequency and
[90] for current control of the active power filter. Focusing amplitude information from the grid voltage at PCC, just like
on the grid synchronization, the dc and harmonic components what most grid synchronization methods do [99]-[102]. As
need to be well addressed. the adaptive observer is the combination of identification and
In this case, high-order GIs are more attractive than the observation, the grid fundamental frequency is viewed as an
standard SOGI [91]-[92]. As mentioned in Section III-A, unknown parameter. The adaptation law to estimate the
three modifications are made: CSOGI-PLL [35], SO-SOGI- frequency should be carefully designed to satisfy Lyapunov
PLL [37] and MSTOGI-PLL [40]. Recently, even a fifth stability criterion while the selection of Lyapunov function
order GI is proposed [93]. Though the dc and harmonic usually depends on the researchers’ experience. Moreover,
attenuation is better, the design and implementation for high- even some existing studies do consider the dc, fundamental
order GIs are complicated. Another option to deal with the component and main harmonics contained in the grid voltage
harmonics is to collaborate the SOGI together, such as the (e.g. in [100], 5th, 7th, 11th, 13th harmonics are considered),
multiple-SOGI (MSOGI), where each SOGI is tuned at a error is unavoidable for neglecting higher-order harmonics
specified harmonic frequency [94]. An equivalent structure and inter-harmonics. If more harmonics are included in the
of MSOGI is shown in [95] and it is demonstrated that the estimation model, the observer order and implementation
computational burden is reduced compared with the original complexity will increase a lot. That is, a balance between the
MSOGI-PLL. A prior knowledge of the harmonic orders is estimation accuracy and implementation complexity should
one of the drawbacks of this method. be made.
Apart from adding new paths or higher-order integrators The second method takes the system output filter into the
into the standard SOGI-PLL, some researchers are trying to model to reconstruct the whole inverter system and to realize
modify the SOGI recently [96]-[98]. The aim is to obtain a the ‘sensorless’ goal [103]-[105]. For instance, in [104], only
faster dynamic and simpler tuning process. Actually, it can one sensor for the grid current is needed. The grid voltage for
be used to improve the performance of MSOGI or DSOGI. synchronization and the capacitor current for active damping
of the LCL filter resonance can be estimated by the observer.
B. ADAPTIVE OBSERVERS Apparently, the system works well if the parameters are all at
In recent years, the adaptive observers (AO) are also widely their nominal values. However, the estimation accuracy may
used for the purpose of grid synchronization. With the help be low due to the parameter uncertainties.
of adaptive observer, the grid frequency can be obtained by If the grid voltage harmonics and system parameter
the arc-tangent operation of a pair of orthogonal signals that variations are considered, the system control with an adaptive
the adaptive observer estimates. Hence, no complex filtering observer technique is rather complex. To mitigate the
or OSG is needed. The fast dynamic response is another complexity of AO and to take advantage of fast dynamic of
advantage of the adaptive observer compared with the PLL. AO (that is, to improve the dynamics for the PLL by AO) at
Generally speaking, the adaptive observer used in the grid- the same time, some attempts are made to combine the AO
connected inverters can be divided into two categories. and GI-based PLL together. The results in [106]-[107] seem
promising and further improvements are needed.
VOLUME XX, 2017 9
Access
TABLE 3. Current techniques and future direction for the SOGI-PLL under complex grid conditions.
Specific grid
Current techniques and the classification Research status Future direction
condition
 Direct filtering methods
 DC estimation and subtraction methods Select one method for specific
DC offset Mature
application
 Non-dc signal extraction methods
 Pre-filtering methods
Low-order  In-loop filtering methods Select one method for specific
Mature
harmonics application
 OSG modification methods
 Parallel frequency detector methods A fast, simple, accurate
Frequency  Phase error compensation methods Complex and inaccurate in frequency estimator;
variations some conditions More accurate compensation;
 PLL modification methods New phase detector.
 Typical second order system model
 Typical type-II system model Accurate under small
bandwidth condition; Taking the loop coupling into
modeling  Perturbation and linearization model
Nonlinear loop coupling consideration
Weak grid  Extended perturbation and linearization phenomenon is not considered
condition model
Design under ideal grid
 Bandwidth-oriented method condition; Insight of each PI parameter
design
 Symmetrical optimum method Has to reduce the bandwidth design
under weak grid condition
4) Currently, some good attempts are made to combine the

VII. CONCLUSION advantages of PLL and some other grid synchronization
The SOGI-PLL can be a high-performance single-phase PLL methods, such as the adaptive observer technique. This
used in a wide range of applications. It has many merits like method can be another way to improve the SOGI-PLL
the simple structure, good filtering capability and frequency performance and further investigations are needed.
adaptability. However, as the grid is becoming more power As a result, for more complex grid conditions (especially
electronics characterized, the SOGI-PLL is still required to for the grid frequency variation and high grid impedance),
be further improved, given that such PLL can be sensitive to the effective single-phase PLL solutions still require further
the dc offset, low-order harmonics and high grid impedance. explorations, detailed of which can be found in the main text
Hence, this article collects and reviews the existing and tables of this study.
solutions and a summary is shown in Table 3. It is found that
1) The solutions for the dc offset and low-order harmonics REFERENCES
suppression are relatively mature, and readers can select [1] F. Blaabjerg, R. Teodorescu, M. Liserre and A. V. Timbus,
proper solutions based on specific requirements. "Overview of Control and Grid Synchronization for Distributed
Power Generation Systems," in IEEE Transactions on Industrial
2) The FFSOGI-PLL shows better dynamics and less Electronics, vol. 53, no. 5, pp. 1398-1409, Oct. 2006.
computational burden than the standard SOGI-PLL. One [2] R. R. Behera and A. N. Thakur, "An overview of various grid
thing should be well addressed is the estimation error synchronization techniques for single-phase grid integration of
renewable distributed power generation systems," 2016 International
under grid frequency variations. The parallel frequency
Conference on Electrical, Electronics, and Optimization Techniques
detector methods are intuitive in principle but a simple (ICEEOT), Chennai, 2016, pp. 2876-2880.
yet effective frequency estimator is not found. The phase [3] Y. Yang and F. Blaabjerg, "Synchronization in single-phase grid-
error compensation methods are sufficient for PLLs with connected photovoltaic systems under grid faults," 2012 3rd IEEE
International Symposium on Power Electronics for Distributed
low bandwidth but may not be accurate enough if the Generation Systems (PEDG), Aalborg, 2012, pp. 476-482.
bandwidth turns larger. The PLL structure modification [4] V. D. Bacon, S. A. Oliveira da Silva, L. B. Garcia Campanhol and B.
methods also show fast dynamic response, but the A. Angélico, "Stability analysis and performance evaluation of a
single-phase phase-locked loop algorithm using a non-autonomous
filtering ability is poor and the transient overshoot can be adaptive filter," in IET Power Electronics, vol. 7, no. 8, pp. 2081-
high. For these methods, how to design the phase 2092, August 2014.
detector remains a tough task. [5] Y. Yang, F. Blaabjerg and Z. Zou, "Benchmarking of Grid Fault
Modes in Single-Phase Grid-Connected Photovoltaic Systems," in
3) Because of the loop coupling, the existing modeling is IEEE Transactions on Industry Applications, vol. 49, no. 5, pp.
not accurate enough under the conditions of high grid 2167-2176, Sept.-Oct. 2013.
impedance and large PLL bandwidth. The conventional [6] Z. Zhang, Y. Yang, F. Blaabjerg and R. Ma, "Challenges to grid
design methods for the SOGI-PLL may be insufficient in synchronization of single-phase grid-connected inverters in Zero-
Voltage Ride-Through Operation," 2016 IEEE 2nd Annual Southern
a weak grid and how to optimize the PLL parameters Power Electronics Conference (SPEC), Auckland, 2016, pp. 1-6.
still needs further exploration. [7] D. Yazdani, M. Pahlevaninezhad and A. Bakhshai, "Single-phase
grid-synchronization algorithms for converter interfaced distributed
VOLUME XX, 2017 9
Access
generation systems," 2009 Canadian Conference on Electrical and single-phase shunt active power filters under adverse grid
Computer Engineering, St. John's, NL, 2009, pp. 127-131. conditions," 4th Annual International Power Electronics, Drive
[8] S. Golestan, M. Monfared, F. D. Freijedo and J. M. Guerrero, Systems and Technologies Conference, Tehran, 2013, pp. 510-517.
"Dynamics Assessment of Advanced Single-Phase PLL Structures," [25] M. Ciobotaru, R. Teodorescu and F. Blaabjerg, "A new single-phase
in IEEE Transactions on Industrial Electronics, vol. 60, no. 6, pp. PLL structure based on second order generalized integrator," 2006
2167-2177, June 2013. 37th IEEE Power Electronics Specialists Conference, Jeju, 2006, pp.
[9] R. M. Santos Filho, P. F. Seixas, P. C. Cortizo, L. A. B. Torres and A. 1-6.
F. Souza, "Comparison of Three Single-Phase PLL Algorithms for [26] Y. Gui, X. Wang, H. Wu and F. Blaabjerg, "Voltage-Modulated
UPS Applications," in IEEE Transactions on Industrial Electronics, Direct Power Control for a Weak Grid-Connected Voltage Source
vol. 55, no. 8, pp. 2923-2932, Aug. 2008. Inverters," in IEEE Transactions on Power Electronics, vol. 34, no.
[10] A. Elrayyah, Y. Sozer and M. Elbuluk, "Robust phase locked-loop 11, pp. 11383-11395, Nov. 2019.
algorithm for single-phase utility-interactive inverters," in IET Power [27] M. Karimi-Ghartemani, S. A. Khajehoddin, P. K. Jain, A. Bakhshai
Electronics, vol. 7, no. 5, pp. 1064-1072, May 2014. and M. Mojiri, "Addressing DC Component in PLL and Notch Filter
[11] I. Carugati, P. Donato, S. Maestri, D. Carrica and M. Benedetti, Algorithms," in IEEE Transactions on Power Electronics, vol. 27, no.
"Frequency Adaptive PLL for Polluted Single-Phase Grids," in IEEE 1, pp. 78-86, Jan. 2012.
Transactions on Power Electronics, vol. 27, no. 5, pp. 2396-2404, [28] S. Golestan, J. M. Guerrero, F. Musavi and J. C. Vasquez, "Single-
May 2012. Phase Frequency-Locked Loops: A Comprehensive Review," in
[12] S. Golestan, M. Monfared, F. D. Freijedo and J. M. Guerrero, IEEE Transactions on Power Electronics, vol. 34, no. 12, pp. 11791-
"Design and Tuning of a Modified Power-Based PLL for Single- 11812, Dec. 2019.
Phase Grid-Connected Power Conditioning Systems," in IEEE [29] Photovoltaic (PV) Systems-Characteristics of the Utility Interface,
Transactions on Power Electronics, vol. 27, no. 8, pp. 3639-3650, IEC Standard 61727, 2004.
Aug. 2012. [30] H. Liu, Y. Sun, H. Hu and Y. Xing, "A new single-phase PLL based
[13] M. Karimi-Ghartemani and M. R. Iravani, "A nonlinear adaptive on discrete fourier transform," 2015 IEEE Applied Power Electronics
filter for online signal analysis in power systems: applications," in Conference and Exposition (APEC), Charlotte, NC, 2015, pp. 521-
IEEE Transactions on Power Delivery, vol. 17, no. 2, pp. 617-622, 526.
April 2002. [31] M. Xie, H. Wen, C. Zhu and Y. Yang, "DC Offset Rejection
[14] Y. Yang, K. Zhou and F. Blaabjerg, "Virtual Unit Delay for digital Improvement in Single-Phase SOGI-PLL Algorithms: Methods
frequency adaptive T/4 Delay Phase-Locked Loop system," 2016 Review and Experimental Evaluation," in IEEE Access, vol. 5, pp.
IEEE 8th International Power Electronics and Motion Control 12810-12819, 2017.
Conference (IPEMC-ECCE Asia), Hefei, 2016, pp. 2910-2916. [32] S. Golestan, J. M. Guerrero and J. C. Vasquez, "Single-Phase PLLs:
[15] I. Galkin and M. Vorobyov, "Optimizing of sampling in a low-cost A Review of Recent Advances," in IEEE Transactions on Power
single-phase instantaneous AC-grid synchronization unit with Electronics, vol. 32, no. 12, pp. 9013-9030, Dec. 2017.
discrete calculation of derivative function," IECON 2015 - 41st [33] S. Prakash, J. K. Singh, R. K. Behera and A. Mondal,
Annual Conference of the IEEE Industrial Electronics Society, "Comprehensive Analysis of SOGI-PLL Based Algorithms for
Yokohama, 2015, pp. 004538-004543. Single-Phase System," 2019 National Power Electronics Conference
[16] Rae-Young Kim, See-Young Choi and In-Young Suh, (NPEC), Tiruchirappalli, India, 2019, pp. 1-6.
"Instantaneous control of average power for grid tie inverter using [34] S. Golestan, J. M. Guerrero and G. B. Gharehpetian, "Five
single phase D-Q rotating frame with all pass filter," 30th Annual Approaches to Deal With Problem of DC Offset in Phase-Locked
Conference of IEEE Industrial Electronics Society, 2004. IECON Loop Algorithms: Design Considerations and Performance
2004, Busan, South Korea, 2004, pp. 274-279 Vol. 1. Evaluations," in IEEE Transactions on Power Electronics, vol. 31,
[17] P. Hao, W. Zanji and C. Jianye, "A Measuring Method of the Single- no. 1, pp. 648-661, Jan. 2016.
Phase AC Frequency, Phase, and Reactive Power Based on the [35] A. Kulkarni and V. John, "Design of a fast response time single-
Hilbert Filtering," in IEEE Transactions on Instrumentation and phase PLL with DC offset rejection capability," 2016 IEEE Applied
Measurement, vol. 56, no. 3, pp. 918-923, June 2007. Power Electronics Conference and Exposition (APEC), Long Beach,
[18] K. De Brabandere et al., "Design and Operation of a Phase-Locked CA, 2016, pp. 2200-2206.
Loop with Kalman Estimator-Based Filter for Single-Phase [36] B. Liu et al., "A Simple Approach to Reject DC Offset for Single-
Applications," IECON 2006 - 32nd Annual Conference on IEEE Phase Synchronous Reference Frame PLL in Grid-Tied Converters,"
Industrial Electronics, Paris, 2006, pp. 525-530. in IEEE Access, vol. 8, pp. 112297-112308, 2020.
[19] A. de Jesús Chica Leal, C. L. T. Rodríguez and F. Santamaria, [37] Z. Xin, X. Wang, Z. Qin, M. Lu, P. C. Loh and F. Blaabjerg, "An
"Comparison of synchronization techniques for a grid-tied Improved Second-Order Generalized Integrator Based Quadrature
photovoltaic system," 2016 IEEE 36th Central American and Signal Generator," in IEEE Transactions on Power Electronics, vol.
Panama Convention (CONCAPAN XXXVI), San Jose, 2016, pp. 1-5. 31, no. 12, pp. 8068-8073, Dec. 2016.
[20] S. M. Silva, B. M. Lopes, B. J. C. Filho, R. P. Campana and W. C. [38] S. Hwang, L. Liu, H. Li and J. Kim, "DC Offset Error Compensation
Bosventura, "Performance evaluation of PLL algorithms for single- for Synchronous Reference Frame PLL in Single-Phase Grid-
phase grid-connected systems," Conference Record of the 2004 IEEE Connected Converters," in IEEE Transactions on Power Electronics,
Industry Applications Conference, 2004. 39th IAS Annual Meeting., vol. 27, no. 8, pp. 3467-3471, Aug. 2012.
Seattle, WA, USA, 2004, pp. 2259-2263 vol.4. [39] M. Ciobotaru, R. Teodorescu and V. G. Agelidis, "Offset rejection
[21] H. Fadil, D. Yousfi, Y. Aite Driss and A. R. Nasrudin, for PLL based synchronization in grid-connected converters," 2008
"Synchronization techniques benchmarking of grid fault modes in Twenty-Third Annual IEEE Applied Power Electronics Conference
single-phase systems," 2014 International Renewable and and Exposition, Austin, TX, 2008, pp. 1611-1617.
Sustainable Energy Conference (IRSEC), Ouarzazate, 2014, pp. 191- [40] C. Zhang, X. Zhao, X. Wang, X. Chai, Z. Zhang and X. Guo, "A
196. Grid Synchronization PLL Method Based on Mixed Second- and
[22] M. Cacciato, A. Consoli, V. Crisafulli, G. Scarcella and G. Scelba, Third-Order Generalized Integrator for DC Offset Elimination and
"Robustness evaluation of phase-locked loop algorithms for single- Frequency Adaptability," in IEEE Journal of Emerging and Selected
phase distributed generation systems," SPEEDAM 2010, Pisa, 2010, Topics in Power Electronics, vol. 6, no. 3, pp. 1517-1526, Sept. 2018.
pp. 914-919. [41] S. Shinnaka, "A Robust Single-Phase PLL System with Stable and
[23] L. N. Arruda, S. M. Silva, and B. J. C. Filho, “PLL structures for Fast Tracking," in IEEE Transactions on Industry Applications, vol.
utility connected systems,” in Conference Record of the 2001 IEEE 44, no. 2, pp. 624-633, March-April 2008.
Industry Applications Conference, the 36th IAS Annual Meeting 2001, [42] G. Fedele and A. Ferrise, "Non Adaptive Second-Order Generalized
vol. 4, pp. 2655–2660, Sept. 30–Oct. 4, 2001. Integrator for Identification of a Biased Sinusoidal Signal," in IEEE
[24] S. Golestan, M. Monfared and J. M. Guerrero, "Second order Transactions on Automatic Control, vol. 57, no. 7, pp. 1838-1842,
generalized integrator based reference current generation method for July 2012.
VOLUME XX, 2017 9
Access
[43] S. Golestan, F. D. Freijedo and J. M. Guerrero, "A Systematic [60] F. Xiao, L. Dong, L. Li and X. Liao, "A Frequency-Fixed SOGI-
Approach to Design High-Order Phase-Locked Loops," in IEEE Based PLL for Single-Phase Grid-Connected Converters," in IEEE
Transactions on Power Electronics, vol. 30, no. 6, pp. 2885-2890, Transactions on Power Electronics, vol. 32, no. 3, pp. 1713-1719,
June 2015. March 2017.
[44] S. Golestan, M. Monfared and F. D. Freijedo, "Design-Oriented [61] P. Liu and S. Duan, "An Open-Loop Synchronization Technique
Study of Advanced Synchronous Reference Frame Phase-Locked With Simple Structure for Phase Error Compensation and Frequency
Loops," in IEEE Transactions on Power Electronics, vol. 28, no. 2, Estimation," in IEEE Transactions on Industrial Electronics, vol. 67,
pp. 765-778, Feb. 2013. no. 10, pp. 8936-8940, Oct. 2020.
[45] S. Golestan and J. M. Guerrero, "An Analysis of Modified [62] S. Golestan, A. Vidal, A. G. Yepes, J. M. Guerrero, J. C. Vasquez
Demodulation-Based Grid Voltage Parameter Estimator," in IEEE and J. Doval-Gandoy, "A True Open-Loop Synchronization
Transactions on Power Electronics, vol. 30, no. 12, pp. 6528-6533, Technique," in IEEE Transactions on Industrial Informatics, vol. 12,
Dec. 2015. no. 3, pp. 1093-1103, June 2016.
[46] S. Golestan, J. M. Guerrero, A. Vidal, A. G. Yepes and J. Doval- [63] S. Golestan, J. M. Guerrero and J. C. Vasquez, "An Open-Loop Grid
Gandoy, "PLL With MAF-Based Prefiltering Stage: Small-Signal Synchronization Approach for Single-Phase Applications," in IEEE
Modeling and Performance Enhancement," in IEEE Transactions on Transactions on Power Electronics, vol. 33, no. 7, pp. 5548-5555,
Power Electronics, vol. 31, no. 6, pp. 4013-4019, June 2016. July 2018.
[47] S. Gude and C. Chu, "Single-Phase Enhanced Phase-Locked Loops [64] Q. Guan, Y. Zhang, Y. Kang and J. M. Guerrero, "Single-Phase
Based on Multiple Delayed Signal Cancellation Filters for Micro- Phase-Locked Loop Based on Derivative Elements," in IEEE
Grid Applications," in IEEE Transactions on Industry Applications, Transactions on Power Electronics, vol. 32, no. 6, pp. 4411-4420,
vol. 55, no. 6, pp. 7122-7133, Nov.-Dec. 2019. June 2017.
[48] F. D. Freijedo, J. Doval-Gandoy, O. Lopez and E. Acha, "Tuning of [65] S. Zhou et al., "An Improved Design of Current Controller for LCL-
Phase-Locked Loops for Power Converters Under Distorted Utility Type Grid-Connected Converter to Reduce Negative Effect of PLL
Conditions," in IEEE Transactions on Industry Applications, vol. 45, in Weak Grid," in IEEE Journal of Emerging and Selected Topics in
no. 6, pp. 2039-2047, Nov.-dec. 2009. Power Electronics, vol. 6, no. 2, pp. 648-663, June 2018.
[49] J. Xu, H. Qian, S. Bian, Y. Hu and S. Xie, "Comparative study of [66] D. Zhu, S. Zhou, X. Zou and Y. Kang, "Improved Design of PLL
single-phase phase-locked loops for grid-connected inverters under Controller for LCL-Type Grid-Connected Converter in Weak Grid,"
non-ideal grid conditions," in CSEE Journal of Power and Energy in IEEE Transactions on Power Electronics, vol. 35, no. 5, pp. 4715-
Systems. 4727, May 2020.
[50] A. Kulkarni and V. John, "A novel design method for SOGI-PLL for [67] D. Zhu, S. Zhou, X. Zou, Y. Kang and K. Zou, "Small-Signal
minimum settling time and low unit vector distortion," IECON 2013 Disturbance Compensation Control for LCL-Type Grid-Connected
- 39th Annual Conference of the IEEE Industrial Electronics Society, Converter in Weak Grid," in IEEE Transactions on Industry
Vienna, 2013, pp. 274-279. Applications, vol. 56, no. 3, pp. 2852-2861, May-June 2020.
[51] M. S. Reza, M. Ciobotaru and V. G. Agelidis, "Tracking of time- [68] Y. Han, H. Chen, Z. Li, P. Yang, L. Xu and J. M. Guerrero,
varying grid voltage using DFT based second order generalized "Stability Analysis for the Grid-Connected Single-Phase
integrator technique," 2012 IEEE International Conference on Power Asymmetrical Cascaded Multilevel Inverter With SRF-PI Current
System Technology (POWERCON), Auckland, 2012, pp. 1-6. Control Under Weak Grid Conditions," in IEEE Transactions on
[52] Q. Sun, J. M. Guerrero, T. Jing, J. C. Vasquez and R. Yang, "An Power Electronics, vol. 34, no. 3, pp. 2052-2069, March 2019.
Islanding Detection Method by Using Frequency Positive Feedback [69] X. Chen, Y. Zhang, S. Wang, J. Chen and C. Gong, "Impedance-
Based on FLL for Single-Phase Microgrid," in IEEE Transactions on Phased Dynamic Control Method for Grid-Connected Inverters in a
Smart Grid, vol. 8, no. 4, pp. 1821-1830, July 2017. Weak Grid," in IEEE Transactions on Power Electronics, vol. 32, no.
[53] J. Matas, H. Martín, J. de la Hoz, A. Abusorrah, Y. A. Al-Turki and 1, pp. 274-283, Jan. 2017.
M. Al-Hindawi, "A Family of Gradient Descent Grid Frequency [70] L. Huang et al., "Grid-Synchronization Stability Analysis and Loop
Estimators for the SOGI Filter," in IEEE Transactions on Power Shaping for PLL-Based Power Converters With Different Reactive
Electronics, vol. 33, no. 7, pp. 5796-5810, July 2018. Power Control," in IEEE Transactions on Smart Grid, vol. 11, no. 1,
[54] G. Fedele, C. Picardi and D. Sgro, "A Power Electrical Signal pp. 501-516, Jan. 2020.
Tracking Strategy Based on the Modulating Functions Method," in [71] Z. Xie et al., "Modeling and Control Parameters Design for Grid-
IEEE Transactions on Industrial Electronics, vol. 56, no. 10, pp. Connected Inverter System Considering the Effect of PLL and Grid
4079-4087, Oct. 2009. Impedance," in IEEE Access, vol. 8, pp. 40474-40484, 2020.
[55] M. A. Akhtar and S. Saha, "An Adaptive Frequency-Fixed Second- [72] Y. Huang, X. Yuan, J. Hu and P. Zhou, "Modeling of VSC
Order Generalized Integrator-Quadrature Signal Generator Using Connected to Weak Grid for Stability Analysis of DC-Link Voltage
Fractional-Order Conformal Mapping Based Approach," in IEEE Control," in IEEE Journal of Emerging and Selected Topics in
Transactions on Power Electronics, vol. 35, no. 6, pp. 5548-5552, Power Electronics, vol. 3, no. 4, pp. 1193-1204, Dec. 2015.
June 2020. [73] X. Zhang, D. Xia, Z. Fu, G. Wang and D. Xu, "An Improved
[56] M. S. Reza, M. Ciobotaru and V. G. Agelidis, "Grid voltage offset Feedforward Control Method Considering PLL Dynamics to
and harmonics rejection using second order generalized integrator Improve Weak Grid Stability of Grid-Connected Inverters," in IEEE
and kalman filter technique," Proceedings of The 7th International Transactions on Industry Applications, vol. 54, no. 5, pp. 5143-5151,
Power Electronics and Motion Control Conference, Harbin, 2012, pp. Sept.-Oct. 2018.
104-111. [74] H. Wu, X. Ruan, D. Yang, "Research on the Stability Caused by
[57] A. Sahoo, K. Mahmud, M. Ciobotaru and J. Ravishankar, "Adaptive Phase-Locked Loop for LCL-Type Grid-Connected Inverter under
Grid Synchronization Technique for Single-phase Inverters in AC Weak Grid Condition," in Proceedings of the CSEE, vol. 34, no. 30,
Microgrid," 2019 IEEE Energy Conversion Congress and Exposition pp. 5259-5268, Dec 2014.
(ECCE), Baltimore, MD, USA, 2019, pp. 4441-4446. [75] X. Wang and F. Blaabjerg, "Harmonic Stability in Power Electronic-
[58] S. Golestan, S. Y. Mousazadeh, J. M. Guerrero and J. C. Vasquez, Based Power Systems: Concept, Modeling, and Analysis," in IEEE
"A Critical Examination of Frequency-Fixed Second-Order Transactions on Smart Grid, vol. 10, no. 3, pp. 2858-2870, May
Generalized Integrator-Based Phase-Locked Loops," in IEEE 2019.
Transactions on Power Electronics, vol. 32, no. 9, pp. 6666-6672, [76] C. Zhang, X. Wang, F. Blaabjerg and W. Wang, "Benchmarking of
Sept. 2017. small-signal dynamics of single-phase PLLs," 2015 9th International
[59] M. S. Reza, M. Ciobotaru and V. G. Agelidis, "Estimation of single- Conference on Power Electronics and ECCE Asia (ICPE-ECCE
phase grid voltage fundamental parameters using fixed frequency Asia), Seoul, 2015, pp. 1420-1427.
tuned second-order generalized integrator based technique," 2013 4th [77] J. Xu, Q. Qian, B. Zhang and S. Xie, "Harmonics and Stability
IEEE International Symposium on Power Electronics for Distributed Analysis of Single-Phase Grid-Connected Inverters in Distributed
Generation Systems (PEDG), Rogers, AR, 2013, pp. 1-8. Power Generation Systems Considering Phase-Locked Loop
VOLUME XX, 2017 9
Access
Impact," in IEEE Transactions on Sustainable Energy, vol. 10, no. 3, Tradeoff of Voltage Sequence Detection Methods Under Adverse
pp. 1470-1480, July 2019. Grid Conditions," in IEEE Transactions on Industrial Electronics,
[78] C. Zhang, X. Wang, F. Blaabjerg, W. Wang and C. Liu, "The vol. 61, no. 5, pp. 2139-2151, May 2014.
influence of phase-locked loop on the stability of single-phase grid- [95] J. Geng, X. Li, Q. Liu, J. Chen, Z. Xin and P. C. Loh, "Frequency-
connected inverter," 2015 IEEE Energy Conversion Congress and Locked Loop Based on a Repetitive Controller for Grid
Exposition (ECCE), Montreal, QC, 2015, pp. 4737-4744. Synchronization Systems," in IEEE Access, vol. 8, pp. 154861-
[79] S. Shah and L. Parsa, "On impedance modeling of single-phase 154870, 2020.
voltage source converters," 2016 IEEE Energy Conversion Congress [96] N. Kumar, B. Singh, B. K. Panigrahi, "Grid synchronisation
and Exposition (ECCE), Milwaukee, WI, 2016, pp. 1-8. framework for partially shaded solar PV-based microgrid using
[80] S. F⊘yen, C. Zhang, O. Fosso and M. Molinas, "Frequency Domain intelligent control strategy," in IET Generation, Transmission &
Modelling for Assessment of Hilbert and SOGI Based Single-Phase Distribution, vol. 13, no. 6, pp. 829-837, 2019.
Synchronisation," IECON 2019 - 45th Annual Conference of the [97] N. Hui, D. Wang and Y. Li, "A Novel Hybrid Filter-Based PLL to
IEEE Industrial Electronics Society, Lisbon, Portugal, 2019, pp. Eliminate Effect of Input Harmonics and DC Offset," in IEEE Access,
1780-1785. vol. 6, pp. 19762-19773, 2018.
[81] S. Shah, P. Koralewicz, V. Gevorgian and L. Parsa, "Small-Signal [98] F. U. Nazir, N. Kumar, B. C. Pal, B. Singh and B. K. Panigrahi,
Modeling and Design of Phase-Locked Loops Using Harmonic "Enhanced SOGI Controller for Weak Grid Integrated Solar PV
Signal-Flow Graphs," in IEEE Transactions on Energy Conversion, System," in IEEE Transactions on Energy Conversion, vol. 35, no. 3,
vol. 35, no. 2, pp. 600-610, June 2020. pp. 1208-1217, Sept. 2020.
[82] M. G. Taul, X. Wang, P. Davari and F. Blaabjerg, "An Overview of [99] I. Askarian, S. Eren, M. Palevani and A. Knight, "New frequency
Assessment Methods for Synchronization Stability of Grid- and amplitude estimation techniques for grid-connected DC/AC
Connected Converters Under Severe Symmetrical Grid Faults," in inverters," 2017 IEEE Energy Conversion Congress and Exposition
IEEE Transactions on Power Electronics, vol. 34, no. 10, pp. 9655- (ECCE), Cincinnati, OH, 2017, pp. 5738-5743.
9670, Oct. 2019. [100] R. Bimarta and K. Kim, "A Robust Frequency-Adaptive Current
[83] Y. Xia, M. Yu, X. Wang and W. Wei, "Describing Function Method Control of a Grid-Connected Inverter Based on LMI-LQR Under
Based Power Oscillation Analysis of LCL-Filtered Single-Stage PV Polytopic Uncertainties," in IEEE Access, vol. 8, pp. 28756-28773,
Generators Connected to Weak Grid," in IEEE Transactions on 2020.
Power Electronics, vol. 34, no. 9, pp. 8724-8738, Sept. 2019. [101] H. Ahmed, M. Ahsan, M. Benbouzid, A. Albarbar, M. Shahjalal and
[84] L. Harnefors, M. Bongiorno and S. Lundberg, "Input-Admittance S. Biricik, "Coordinate Transformation-Free Observer-Based
Calculation and Shaping for Controlled Voltage-Source Converters," Adaptive Estimation of Distorted Single-Phase Grid Voltage Signal,"
in IEEE Transactions on Industrial Electronics, vol. 54, no. 6, pp. in IEEE Access, vol. 8, pp. 74280-74290, 2020.
3323-3334, Dec. 2007. [102] Z. Dai, W. Lin and H. Lin, "Estimation of Single-Phase Grid Voltage
[85] Y. F. Wang and Y. Wei Li, "Analysis and Digital Implementation of Parameters With Zero Steady-State Error," in IEEE Transactions on
Cascaded Delayed-Signal-Cancellation PLL," in IEEE Transactions Power Electronics, vol. 31, no. 5, pp. 3867-3879, May 2016.
on Power Electronics, vol. 26, no. 4, pp. 1067-1080, April 2011. [103] T. V. Tran and K. Kim, "Frequency Adaptive Grid Voltage
[86] S. Golestan, J. M. Guerrero, J. C. Vasquez, A. M. Abusorrah and Y. Sensorless Control of LCL-Filtered Inverter Based on Extended
Al-Turki, "Modeling, Tuning, and Performance Comparison of Model Observer," in IEEE Transactions on Industrial Electronics, vol.
Second-Order-Generalized-Integrator-Based FLLs," in IEEE 67, no. 9, pp. 7560-7573, Sept. 2020.
Transactions on Power Electronics, vol. 33, no. 12, pp. 10229-10239, [104] R. A. Fantino, C. A. Busada and J. A. Solsona, "Observer-Based
Dec. 2018. Grid-Voltage Sensorless Synchronization and Control of a VSI-LCL
[87] Y. Han, M. Luo, X. Zhao, J. M. Guerrero and L. Xu, "Comparative Tied to an Unbalanced Grid," in IEEE Transactions on Industrial
Performance Evaluation of Orthogonal-Signal-Generators-Based Electronics, vol. 66, no. 7, pp. 4972-4981, July 2019.
Single-Phase PLL Algorithms—A Survey," in IEEE Transactions on [105] M. Su et al., "Single-Sensor Control of LCL-Filtered Grid-Connected
Power Electronics, vol. 31, no. 5, pp. 3932-3944, May 2016. Inverters," in IEEE Access, vol. 7, pp. 38481-38494, 2019.
[88] C. Tang, "Four Design Methods for Proportional-Integral Controller [106] Y. Park, H. Kim and S. Sul, "Frequency-Adaptive Observer to
of Grid-Connected Inverter with LCL Output Filter," in Power Extract AC-Coupled Signals for Grid Synchronization," in IEEE
System Technology, vol. 37, no. 11, pp. 3268-3275, Nov 2013. Transactions on Industry Applications, vol. 53, no. 1, pp. 273-282,
[89] X. Li and H. Lin, "A Design Method of Phase-Locked Loop for Jan.-Feb. 2017.
Grid-Connected Converters Considering the Influence of Current [107] H. Ahmed and M. Benbouzid, "On the Enhancement of Generalized
Loops in Weak Grid," in IEEE Journal of Emerging and Selected Integrator-Based Adaptive Filter Dynamic Tuning Range," in IEEE
Topics in Power Electronics. Transactions on Instrumentation and Measurement, vol. 69, no. 10,
[90] X. Yuan, W. Merk, H. Stemmler, and J. Allmeling, “Stationary- pp. 7449-7457, Oct. 2020.
frame generalized integrators for current control of active power
filters with zero steady-state error for current harmonics of concern
under unbalanced and distorted operating conditions,” in IEEE
Transactions on Industry Applications, vol. 38, no. 2, pp. 523-532,
March-April 2002.
[91] P. Shah, I. Hussain, B. Singh, A. Chandra and K. Al-Haddad, "GI-
Based Control Scheme for Single-Stage Grid Interfaced SECS for
Power Quality Improvement," in IEEE Transactions on Industry
Applications, vol. 55, no. 1, pp. 869-881, Jan.-Feb. 2019.
[92] M. Z. Hasan, I. F. Shiam, T. T. Nova and M. S. Reza, "A Modified
PLL Based on Second Order Generalized Integrator for Single-Phase
Voltage System," 2019 International Conference on Electrical,
Computer and Communication Engineering (ECCE), Cox'sBazar,
Bangladesh, 2019, pp. 1-6.
[93] N. Kumar, I. Hussain, B. Singh and B. K. Panigrahi,
"Implementation of Multilayer Fifth-Order Generalized Integrator-
Based Adaptive Control for Grid-Tied Solar PV Energy Conversion
System," in IEEE Transactions on Industrial Informatics, vol. 14, no.
7, pp. 2857-2868, July 2018.
[94] J. Matas, M. Castilla, J. Miret, L. García de Vicuña and R. Guzman,
"An Adaptive Prefiltering Method to Improve the Speed/Accuracy
VOLUME XX, 2017 9
Access
JINMING XU (S’13–M’18) was born in Xuzhou, SHENYIYANG BIAN was born in Jiangsu
Jiangsu Province, China, in 1987. He received the Province, China, in 1996. He received the B.S.
B.S. degree in electrical engineering and the Ph.D. degree in electrical engineering in 2018 from
degree in power electronics from Nanjing Nanjing University of Aeronautics and
University of Aeronautics and Astronautics Astronautics (NUAA), Nanjing, China, where he
(NUAA), Nanjing, China, in 2009 and 2017, is currently working towards the M.S. degree in
respectively. electrical engineering.
In 2017, he joined the College of Automation His current research interests mainly include
Engineering, NUAA, where he is currently an the grid-connected inverters and control
Associate Professor. He has authored and strategies.
coauthored over 80 technical papers in many international journals and
conference proceedings. His current research interests include the grid-
connected inverters and control, and renewable power generations.
Dr. Xu was a recipient of the 2015 IET Power Electronics Premium
Award. He received the Excellent Doctoral Dissertations Award from
Jiangsu Province, China, in 2018.
HAO QIAN was born in Huai’an, Jiangsu

Province, China, in 1997. He received the B.S. SHAOJUN XIE (M’05) was born in Hubei
degree in electrical engineering in 2019 from Province, China, in 1968. He received the B.S.,
Nanjing University of Aeronautics and M.S., and Ph.D. degrees in electrical engineering
Astronautics (NUAA), Nanjing, China, where he from Nanjing University of Aeronautics and
is currently working towards the M.S. degree in Astronautics (NUAA), China, in 1989, 1992, and
electrical engineering. 1995, respectively.
His current research interests mainly include In 1992, he joined the Faculty of Electrical
the grid-connected inverter and control strategies. Engineering Teaching and Research Division,
NUAA, where he is currently a Professor with
the College of Automation Engineering. In recent
five years, he has authored and coauthored over 100 technical papers
published in many international journals and conference proceedings. His
YUAN HU was born in Jing’men, Hubei main interests include the aviation electrical power supply and power
Province, China, in 1996. He received the B.S. electronic conversion.
degree in electrical engineering and automation Dr. Xie was a recipient of the 2015 IET Power Electronics Premium
in 2019 from Northeast Normal University Award.
(NENU), Changchun, China. He is currently
working towards the M.S. degree in electrical
engineering in Nanjing University of Aeronautics
and Astronautics (NUAA), Nanjing, China.
His current research interests mainly include
the grid-connected inverter and control strategies.
VOLUME XX, 2017 9

Overview of SOGI-Based Single-Phase Phase-Locked L

Uploaded by

Copyright:

Available Formats

Overview of SOGI-Based Single-Phase Phase-Locked L

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Overview of SOGI-Based Single-Phase Phase-Locked L

Uploaded by

Copyright:

Available Formats

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

Overview of SOGI-Based Single-Phase Phase-

INDEX TERMS Grid synchronization, single-phase phase-locked loop, second-order generalized

I. INTRODUCTION stable operation and power quality of the utility grid is

VOLUME XX, 2017 1

VOLUME XX, 2017 9

SOGI Control loop

Main research topics

DC offset Low-order harmonics Frequency variations Weak grid

Assuming that the input signal is a pure sinusoidal in the 0 ke=0.8

VOLUME XX, 2017 9

VOLUME XX, 2017 9

1 u3 Scheme proposed in [41]; (b) Scheme proposed in [42].

VOLUME XX, 2017 9

FIGURE 6. Block diagram of the FFSOGI-PLL based on fractional-order conformal mapping.

VOLUME XX, 2017 9

VOLUME XX, 2017 9

VOLUME XX, 2017 9

VOLUME XX, 2017 9

VOLUME XX, 2017 9

is small-signal linear modeling. A well-known model of the

VOLUME XX, 2017 9

VOLUME XX, 2017 9

TABLE 2. Merits and drawbacks of different models for the SOGI-PLL.

Model type Merits Drawbacks Design method

 The OSG is considered as a

VOLUME XX, 2017 9

4) Currently, some good attempts are made to combine the

VOLUME XX, 2017 9

VOLUME XX, 2017 9

VOLUME XX, 2017 9

VOLUME XX, 2017 9

HAO QIAN was born in Huai’an, Jiangsu

VOLUME XX, 2017 9

You might also like