Received 22 June 2022; accepted 7 July 2022.

Date of publication 12 July 2022;

date of current version 22 July 2022. The review of this article was arranged by Associate Editor D. Dong.
Digital Object Identifier 10.1109/OJPEL.2022.3190103

Modeling Techniques and Stability Analysis

Tools for Grid-Connected Converters
QILIN PENG 1 , GIAMPAOLO BUTICCHI 1 (Senior Member, IEEE),
NADIA MEI LIN TAN 1,2 (Senior Member, IEEE), SANDRO GUENTER 1 (Member, IEEE),
JIAJUN YANG 3 (Member, IEEE), AND PATRICK WHEELER 4 (Fellow, IEEE)
1
Key Laboratory of More Electric Aircraft Technology of Zhejiang Province, University of Nottingham Ningbo China, Ningbo 315100, China
2
Institute of Power Engineering, Universiti Tenaga Nasional, Kajang 43000, Malaysia
3
China Beacons Institute, University of Nottingham Ningbo China, Ningbo 315000, China
4
University of Nottingham, Nottingham NG7 2RD, U.K.
CORRESPONDING AUTHOR: GIAMPAOLO BUTICCHI (e-mail: [email protected])
This work was supported in part by Ministry of Science & Technology under National Key R&D Program of China, under Grant 2021YFE0108600, in part by
Ningbo Key Technology Research and Development Programme under Grant 2021Z035, and in part by the Key International Cooperation of National Natural
Science Foundation of China under Grant 51920105011.

ABSTRACT Large-scale integration of renewable generation interfaced to the network through power elec-
tronic converters has led to drastic changes in power system dynamics. In islanded microgrids or weak
grids, different control concepts for the synchronization of converters have been proposed to provide virtual
inertia and improve their resilience against transient events, ensuring safe operation without heavy redundant
design. The complexity of these power-related control algorithms and their interaction with the inner control
loops causes problems in frequency components above the range of traditional studies which calls on
modeling techniques with a wider bandwidth. This work aims to provide an outline of modeling methods
for grid-connected converter dynamics from subsynchronous to switching sideband frequency range and
relevant analyzing tools. The major contributions of this work are: 1. Theoretical foundations and the
derivation processes are discussed for each of the modeling methods within the time domain, frequency
domain and harmonic domain. 2. Similarities and differences between these methods are highlighted and
recommendations are given regarding different grid situations. 3. A case study with an active front end
converter is shown and the analysis results are validated by simulation and Hardware-in-the-Loop (HiL) test,
illustrating the effectiveness of these methods.

INDEX TERMS Grid-connected converters, stability analysis, impedance-based analysis, harmonic stability.

I. INTRODUCTION counterparts, namely More Electric Aircraft (MEA) [3]. The

Conversion of electrical power from renewable power sources varying frequency and multiple kinds of on-board electri-
nowadays are accomplished mostly by power electronic sys- cal energy generation/utilization devices make it even more
tems [1]. This trend introduced faster dynamics and a larger complex than locomotive systems. This has been motivating
amount of harmonics. In particular, converters with high researchers for discovering an approach capable of character-
switching frequency and multiple nonlinear components in- izing the frequency coupling in both transient response and
herently generate harmonics and could interact with each steady-state operation. Stability analyzing methods are also
other, which are pretty common within electrified transporta- under great concern so as to couple with the ever-developing
tion systems. The case in Swiss locomotives shows how modeling methods.
the coupling of harmonics within different frequency ranges Small-signal stability is one of the most important factors
could damage the system severely [2]. As for the aircraft when designing the controller of a grid-connected converter.
industry, the recent trend is to substitute hydraulic and pneu- State-space-based time-domain (TD) modeling and eigen-
matic subsystems on conventional aircraft with their electric value analysis is a way to determine the oscillation mode

This work is licensed under a Creative Commons Attribution 4.0 License. For more information, see https://creativecommons.org/licenses/by/4.0/
450 VOLUME 3, 2022
and the parameters which could trigger resonances [4]. How- Besides DPM, the Harmonic State-Space (HSS) modeling
ever, building the state-space model requires knowledge of the is also an effective method for modeling systems with multi-
structure and parameters of the system. Hence, when model- frequency coupling. The HSS method was first developed
ing a power electronics-based system, a change in the number for the analysis of helicopter blade dynamics [14], and was
of converters or the control loop may lead to a new state-space then utilized for motor drive systems in order to capture the
matrix. This problem is solved by the Component Connection frequency coupling of nonlinear and time-varying compo-
Method (CCM) to some extent [5], yet the exact parameters nents [2]. The DPM and HSS modeling methods are based on
of the whole system are still necessary. The focus of the the same initial hypotheses and they both utilize the method
frequency domain (FD) impedance modeling instead lies in of frequency segmentation to make the system time-invariant.
the extrinsic properties of the converter output port and the However, HSS takes the form of the exponential Fourier se-
power network input port [6]. Such a model can either be built ries while the DPM appears to be trigonometric, and thus
mathematically or measured in the real physical system [7]; leading to different mathematical representations of electrical
hence grey-box systems can be analyzed accordingly. Further, states. More importantly, for small-signal analysis, the HSS
according to [8], the linearized state-space transfer function model was derived directly from the time series Linear Time-
is equivalent to the impedance model when the voltage state Periodic (LTP) state-space equation. In contrast, the DPs are
variable is considered as the input and the current state vari- derived by integrating the system variables over a moving time
able as the output. window [4]. This leads to different properties and a different
However, in renewable power plants and microgrids, con- range of applications.
verters working in parallel could be using different switching Stability analysis also develops with the ever-improving
frequencies, modulation techniques and topologies with bidi- modeling methods. Basically, stability assessment can be
rectional operation [2]. Also, coupling between power-related categorized into signal frequency response-based [6] and
controllers, inner voltage/current loops, and PLLs becomes energy-based [15]. Energy-based methods have gained atten-
much more severer. This tends to bring unexpected harmonics tion in renewable integration to the power grid as they give
at frequencies ranging from hundreds of Hz to kHz due to the insight into the system from the perspective of power flow.
frequency coupling from the multiplication of several time- The whole power system is always stable as long as each
varying signals and bidirectional interaction [9]. In this case, component stays in the passivity region [16]. In contrast, a
the nonlinear time-variant characteristic of converters are so system oscillation mode is obtained from signal-based meth-
obvious and the coupling between DC and AC components ods so that specific harmonics damping techniques are utilized
are so severe that they can no longer be approximated as a to satisfy the grid code requirements. Moreover, with HD
time-invariant system. Modeling approaches such as multi- modeling methods, time- and frequency-domain analysis can
frequency averaging [10] and harmonic linearization [7] could be used in one unified model, and therefore more dedicated
solve this problem to some extent by introducing frequency controllers can be designed accordingly [17], [18]. Besides
shifting. However, the describing function and harmonic bal- these numerical analysis methods, transient stability analysis
ance approaches enforce the single input to a single output with a large signal non-linear model in phase plane provides a
map. Thus, the resultant model is actually part of the linear visualized way to analyze the stability and performance under
time-periodic system. To better analyze the effect of switch- larger disturbance [19]–[21].
ing modulation and resultant frequency coupling, harmonic- In this work, the derivation process and theoretical back-
domain (HD) models of converters need to be built [11]. ground of the above-mentioned modeling are discussed. Then,
The traditional phasor modeling approach was built on the the relevant stability analysis methods are derived. Some rep-
assumption of “Sinusoidal-Steady-State” or “Quasi-Steady- resentative work among recent research works are reviewed
State” whereby the dynamic process is relatively slow com- and a comparison is given on their properties and applicable
pared to the fundamental frequency. Therefore, it cannot scenarios. A case study on an active front end converter with
predict precisely the fast dynamic response of switched power these methods is presented in order to validate their accuracy
electronics-based systems. To address this issue, a method and provide an example of the application scenario. Analy-
based on the generalized averaging method (GAM) called sis results are validated by both PLECS simulation and HiL
the dynamic phasor modeling (DPM) was proposed by S.R. test.
Sanders and used to model a series resonant converter [12].
Dynamic phasor (DP)-based methods are actually a time- II. MODELING METHODS
variant Fourier decomposition with a sliding time window set Traditionally, there are two mainstream modeling routes,
according to the critical fundamental working period. Thus, namely, time-domain modeling and frequency-domain mod-
it has a bandwidth close to twice the fundamental frequency eling. Each methodology has its own strengths and is in-
and could predict the synchronous oscillation precisely with terchangeable due to the same physical theorem and circuit
a significant reduction in calculation time compared to the topology behind it. Recent studies have made the boundary
electromagnetic transient (EMT) simulation, as the complex of time and FD even vaguer with the combination of TD
calculation of the switching process is avoided and the simu- state equations and Fourier decomposition [17]. In this sec-
lation step size can be increased [13]. tion, three modeling methods are discussed regarding their

VOLUME 3, 2022 451

PENG ET AL.: MODELING TECHNIQUES AND STABILITY ANALYSIS TOOLS FOR GRID-CONNECTED CONVERTERS

of PLL. Then the impedance model of a 3-phase VSC with

current feedback controller can be derived:
    
Id Y (s) Ydq (s) Ud
= dd , (4)
Iq 0 Yqq (s) Uq
with:
Gff (s) − 1 Iq0 Lf sGpll (s)
Ydd (s) = ; Ydq (s) =
FIGURE 1. Thevenein equivalent representation of a subsystem with
sLf + Gi (s) sLf + Gi (s)
connected load. −Lf sGpll (s)Id0 − (1 − Gff (s))
Yqq (s) = . (5)
sLf + Gi (s)
derivation process and properties. A comparative review is Where Gi (s) and Gff (s) are the transfer function of the cur-
given based on the application range. rent controller and the feed forward compensator, L f is the
inductance of the filter inductor on the inverter side and Id0 ,
A. TRANSFER FUNCTION BASED SMALL-SIGNAL MODEL
Iq0 denotes the d and q-axis current at equilibrium. It is also
The fundamental concept of equivalent impedance modeling revealed that the impedance between Iq and Vq dominates
is to linearize the system around a stable operation point with the synchronization frequency due to the negative impedance
transfer functions in the FD and represent each subsystem of within the PLL control bandwidth. The simulation results are
a power system in the FD by an impedance connected to a compared with experimental results in [28], which offers a
voltage source (Thévenin equivalent). Based on this concept, new approach to testify the accuracy of the AC impedance
any kind of power distribution network can be considered as a model regarding stability analysis. In [20], it is further testified
power source and load, as depicted in Fig. 1, where: that the synchronous reference frame (SRF)-PLL is three-
ZL 1 phase balanced under asymmetrical faults when a pre-filter
VL = VS = VS . (1)
ZL + ZS 1 + (ZS /ZL ) is used. Thus dq-based model of the major components still
holds in this case. The model has been further developed to
The impedance of each part of the equivalent subsystem can
be able to deal with a larger phase perturbation (up to 38o)
be either calculated numerically by a transfer function, sim-
by utilizing a second order linearized model of the PLL [29].
ulation, or measured in the real system [22]. The advantages
Moreover, converters with Virtual Synchronized Generator
and basic properties of impedance-based modeling for AC and
(VSG)-based controller [30] and doubly-fed induction gener-
DC distribution networks have been reviewed in detail [23]–
ator (DFIG)-based wind power generation systems [31] could
[25]. Therefore, this paper mainly discusses recent progress
also be modeled accordingly.
and engineering practice, focusing on 3-phase AC distribution
systems and other emerging application scenarios. Regarding
2) SEQUENCE-DOMAIN IMPEDANCE MODEL
different linearization coordinate systems, impedance models
Based on the harmonic linearization principle, the system
can be divided into three major types, namely: dq-domain
impedance can be divided into positive and negative sequence
impedance, sequence-domain impedance and phasor-domain
components (Z p and Zn ) [7]. TD perturbation can be written
impedance model.
as:
 
1) dq-DOMAIN IMPEDANCE MODEL va (t ) = V1 cos (2π f1t ) + Vp cos 2π f pt + φvp
The dq-domain impedance-based stability analysis applied
+ Vn cos (2π fnt + φvn ) , (6)
with GNC was firstly carried out by Belkhayat in 1997 [26].
Impedance models under dq-reference frame have the advan- where V1 is the fundamental voltage amplitude with frequency
tage of simple structure and effectiveness for modeling the f1 , Vp , Vn , φvp and φvn is the magnitude and phase of the
electrical equipment with dq-reference controllers. Taking the sequence perturbation at frequency f p and fn , respectively.
grid-tied operation as an example, the major factor that could The sequence-domain impedance per phase can be expressed
trigger synchronous instability lies in the negative-valued as:

impedance introduced by the phase-locked loop (PLL) [27],  
the transfer function of which can be expressed as: Zp (s) = Gi (s) + sL f · 1 − Gff (s)Gv (s)
HPLL (s)   (7)
GPLL (s) = , (2) I1  
s + Vds HPLL (s) − Gi (s) ejϕ GPLL s − j2π f1 ,−1
2
with: 
 
1 Zn (s) = Gi (s) + sL f · 1 − Gff (s)Gv (s)
HPLL (s) = kp,pll + ki,pll . (3)
s   (8)
I1  
Where Vds is the steady state operation point of d-axes voltage, − Gi (s) e−jϕ GPLL s − j2π f1 ,−1
kp,pll and ki,pll are the proportional and integeral parameter 2

452 VOLUME 3, 2022

FIGURE 2. Closed-loop representation of a grid-tied VSC based power FIGURE 3. Reciprocal SISO circuit in the phasor domain.
system.

phasor domain can be described as [38]:

    
ϕ is the angle of the current vector I1 , and f1 denotes I YVmm (s) −YVmθ (s) U
= , (11)
the fundamental frequency. When the PLL is considered, a I0 ϕ YVmθ (s) YVθθ (s) U0 θ
negative-valued impedance is introduced, which will cause
where:
a low frequency oscillation when coupled with an inductive
weak grid. More specifically, due to the asymmetric nature
of the dq-axis control, a positive sequence disturbance at YVmm (s) = cos ϕ0YV3 (s); YVmθ (s) = − sin ϕ0YV3 (s)
f p will trigger responses at both f p and (2 f1 − f p ), which
is defined as mirror frequency decoupled (MFD) [32]. Such YVθθ (s) = I0YV1 (s) + cos ϕ0YV3 (s)
phenomena is also elaborated in [33] within the stationary- Gpll (s)Gi (s)
frame, which also reveals the influence of the steady-state YV1 (s) =
(1 + Ud0 GPLL (s))(sL + Gi (s))
voltage initial phase on the frequency coupling terms. Com-
Ycdd (s)
pared to the dq-domain model, the mathematical derivation YV2 (s) =
of the sequence-domain impedance model is more complex. 1 + Ud0 GPLL (s)
However, the impedance is not affected by reference frame Ycdd (s) + YV2 (s)
or system asymmetries. By taking the entire power system YV3 (s) =
2
as a closed-loop system, as depicted in Fig. 2, a single-input
It can be seen that the admittance matrix in this case takes
single-output (SISO) impedance model of each sequence is A(s) −C(s)
developed for 3-phase grid-tied VSCs in [34]. Besides, each the special form of: . Such matrices can be
C(s) D(s)
sequence is independent with real physical meaning and can transferred to a symmetric similar matrix as:
be easily measured by commercial equipment. Based on the    
properties above, it is then used for modeling of MMC [35], −1 a + jb 0 1 −1
B = TAT = + (d − a) ,
DFIG power generation systems [36], and etc. In [37], the 0 a − jb −1 1
process of modeling a 3-phase VSC into self immittances and   (12)
transfer immittances is discussed, which gives an insight of 1 j
where: T = √ 1
. Then, according to the principles of
the coupling of different frequency between AC and DC ports. 2 1 -j
the two-port network, the reciprocal two-port circuit of a sym-
metric admittance matrix can be built as shown in Fig. 3. Thus
3) PHASOR-DOMAIN IMPEDANCE MODEL far, the multi-input multi-output (MIMO) problem introduced
Phasors have been widely used in power system analysis so by the two-port equivalent circuit network can be transferred
that magnitudes and angles of electrical states can be repre- to a SISO problem. Different from two separate SISO paths
sented with a single variable. Considering the magnitude and in [34], all factors are compacted in one system so that the
angle of perturbations at the fundamental-frequency, phasors conventional FD analysis methods can be performed. Also,
can be linearized as: the output current limitation according to the gain margin
       could be calculated.
Ugd cos θ0 − sin θ0 U U According to the discussion above, regardless of the ref-
= =P ,
Ugq sin θ0 cos θ0 U0 θ U0 θ erence coordinates being used, the major challenges of
(9) impedance modeling lie in the frequency coupling effect
       in AC/DC combined network, which still calls on further
Igd cos ϕ0 − sin ϕ0 I I
= =P , exploration when applied to large-scale, power electronic
Igq sin ϕ0 cos ϕ0 I0 ϕ I0 ϕ dominated network modeling. The interchangeability between
(10) phasor domain admittance and dq-domain, sequence-domain
counterparts are discussed in [39]. Also, a three-by-three
where θ0 and ϕ0 denotes the angle of voltage U  and current I transfer matrix is built so as to capture the coupling ef-
at equilibrium, the subscripts g denotes the SRF of the power fect between AC and DC sides within the low-frequency
grid and P denotes the phasor domain transfer matrix of the range by separating the AC and DC component disturbance.
output current. Then the terminal impedance model in the Moreover, the impedance model of converters with multiple

VOLUME 3, 2022 453

PENG ET AL.: MODELING TECHNIQUES AND STABILITY ANALYSIS TOOLS FOR GRID-CONNECTED CONVERTERS

TABLE 1. Comparison of Impedance Modeling in Different Domains 1) CONJUGATIVE PROPERTY

Phasors can be expressed in the form of complex value, thus,
except for zeroth-order DPs, a single DP is represented by
real and imaginary variables. By separating the two parts of
the dynamic equations, a nonlinear real-valued DP model has
the property of:
 ∗
xk = xRk + jxIk = x∗−k = xR−k + jxI−k . (15)

2) DIFFERENTIAL PROPERTY
The derivative of kth -order phasor can be expressed as:
dXk (t )/dt = dx/dtk (t ) − jkωs Xk (t ). (16)
As the state-based dynamic models can generally be ex-
pressed as dtd x(t ) = f {x(t ), u(t )}, where u(t ) is the time-series
input signal. The DP derivative can be expressed as:
d
xk = −jkωs xk +  f (x, u)k , (17)
dt
substituted with the electrical states of the basic circuit com-
ponents, the state equation of which can be derived as shown
in Table 2.

3) PRODUCT PROPERTY
For the two time series waveform x1 (t ) and x2 (t ), the product
of their phasors is:

x1 x2 k = x1 k−i x2 i . (18)

i
• stands for the advantages and  stands for the drawbacks.
That is, the time series waveform can be transformed to the
inner coupling effects is being extended to the switching fre- convolution of their counterparts in phasor domain. According
quency range with the HSS model [40], which will be further to (18), the voltage waveform at the switch terminals of a
discussed in section II-C. The comparison of properties of pulse-width modulated (PWM) system can be expressed as:

small-signal impedance modeling methods is summarized as vs qk = vs k−i qi , (19)
shown in Table 1. i
where vs and q denote the voltage waveform of the power
B. DYNAMIC PHASOR MODELING source and PWM switching function, respectively. Therefore,
Different from the transfer function dominant methods men- a link between the source and load component can be estab-
tioned above, the GAM to obtain the DPM is based on the lished accordingly.
property that any TD waveform x(t ) can be represented inside As discussed above, the essence of DPM is to retain the
the interval τ ∈(τ − T, τ ] by the following Fourier series [41]: dominant Fourier coefficients term to describe the major
characteristics of the system. For instance, when analyzing
∞

high-frequency PWM switching circuits, the DC component
x(τ ) = Xk (t )e jkωs τ , (13) should be addressed to capture the low-frequency behavior.
k=−∞ Then the result would be the same as the state-space averaged
model. For resonant converters with some predominantly fun-
with ωs = 2π fs and Xk (t ) being the complex Fourier coeffi- damental frequency, first-order and the dominant harmonic
cients, also named phasors. The dynamic kth -order phasor at DPs should be retained to capture the major sinusoidal re-
time t, Xk (t ), can be expressed as: sponse and the zeroth-order coefficients of those states that
 t
exhibit slowly varying behavior [12]. As the DPM is a large
1
Xk (t ) = x(τ )e−jkωs τ dτ = xk (t ). (14) signal model that varies relatively slowly over time, a larger
T t−T simulation step size can be used compared to EMT mod-
els. In [42], the stability criterion for interfacing between
Where xk (t ) denotes the average kth -order phasor over the DPM and EMT is given. However, as the amount of DPs
period T . The major advantages of the DPM over traditional being considered increases, the calculation complexity of the
methods come from the three important properties. model raises significantly, which defeats the purpose of saving

454 VOLUME 3, 2022

TABLE 2. The Basic Element Dynamic Equations of the Three Methods

computing resources. The multi-frequency DPM is for small = Ak−m xm e(jkωs +s )t + Bk−m km e(jkωs +s )t
signal analysis when the system is linearized to form a Linear k,m∈Z k,m∈Z
Time-Invariant (LTI) system. Also, it is worth mentioning that (23)
when the initial phase is extracted from the DP model and  

expressed in the form of a generalized dq (GDQ) transforma- jkωs + s xk = Ak−m xm + Bk−m um . (24)
tion, a GDQ model is then derived, which puts an emphasis m∈Z m∈Z
on the phase difference so that the power transmission is more
Consequently, the steady state equilibrium can be derived
convenient to calculate accordingly [43]. Combined with the
as:
concept of sequence component, unbalance faults analysis can
be performed [44]. Ẋ = ( [A] − N) X + [B]U , (25)
It follows that the DPM could be used for both large-signal
transient simulation [45]–[49] and steady-state stability as- where:
sessment [50]–[52]. In both cases, it can be coupled with other ⎡ ⎤
.. .. ·
modeling methods by an appropriate interface. The trade-
⎢ . . ·· ⎥
off between model precision and system complexity can be ⎢ A0 −1 A−2
A ⎥
⎢ ⎥ same with
achieved by proper selection of the phasor sequence. [A] = ⎢
⎢ · · · A 1 A0 A−1 ···⎥ ⎥ , [B] (26)
⎢ A 2 A1 A0
⎥
⎣ ⎦
· .. ..
··
C. HARMONIC STATE-SPACE
. .
Similar to the DPM, the HSS is based on the principle that TD
variables can be transferred to the FD by utilizing (13). Thus,  T same with
X = · · · x−1 x0 x1 · · · , (27)
when a small signal disturbance est appears, the state matrix U
of the system becomes [53]:  
N = diag · · · , −j2ωs , −jωs , 0, jωs , j2ωs , · · · . (28)

x(t ) = est xk ejkωs t , (20) where [·] denotes the Toeplitz matrix or diagonal-constant
k∈Z matrix which is used for convolution calculation. According
the differentiation of (20) could be expressed as: to (C), the HSS method provides a unified way to construct
a multi-frequency model which is linearized around operating

  trajectories in the harmonic basis of the period Ts .
ẋ(t ) = jkωs + s xk e(jkωs +s )t , (21)
In fact, harmonic linearization can be regarded as a combi-
k∈Z
nation of FD stabilization and perturbation linearization [25].
for an arbitrary LTP model: Thus, power electronic converters of any topology, which
 can be expressed by a state-space model, can be easily
ẋ(t ) = A(t )x(t ) + B(t )u(t ) transferred to an HSS model: single-phase converters [54],
(22)
y(t ) = C(t )x(t ) + D(t )u(t ) transformers [55], three-phase converters under both balanced
and unbalanced situation [43], [56] and especially modu-
the HSS variable update equation can be derived as (23). lar multilevel converters that have the nature of nonlinear
The coefficients of same frequency on each side of the inter-resonance between sub-modules [40], [57]. Similar to
equation should be equal, so the exponents can be eliminated, the DPM, the selection of harmonics under consideration
then the kth-order HSS equation is obtained: has been a critical concern for HSS regarding computational

 

burden. The specific simulation time comparison of different
jkωs + s xk e(jkωs +s )t = Ak ejkωs t · xm e(jmωs +s )t orders of harmonics of a single-phase transformer has been
k∈Z k∈Z m∈Z made in [58], indicating that the computation duration of


the model increases quadratically with the harmonic order.
+ Bk ejkωs t · um e(jmωs +s )t
An effective solution to this problem is to select the critical
k∈Z m∈Z


harmonics according to the impedance/admittance map [59].
= Ak xm ej(k+m)ωs t+st + Bk um ej(k+m)ωs t+st By considering only the harmonic components with steady-
k,m∈Z k,m∈Z state impedance larger than a certain threshold or significantly

VOLUME 3, 2022 455

PENG ET AL.: MODELING TECHNIQUES AND STABILITY ANALYSIS TOOLS FOR GRID-CONNECTED CONVERTERS

different kinds of disturbance. The corresponding stability

assessment methods will be discussed in this section.

A. EQUIVALENT IMPEDANCE-BASED ANALYSIS

1) IMPEDANCE RATIO CRITERIA
The equivalent impedance ratio-based analysis was intro-
duced by Middlebrook to analyze DC-DC systems [6], which
provides a sufficient criterion for the small-signal stability of
SISO systems. Due to its over-conservative nature, modified
criteria are proposed on its basis. Five of the most used meth-
FIGURE 4. Flow chart of the DPM and HSS derivation process. ods ranked from high to low according to conservatism are:
Middlebrook criterion, Gain Margin Phase Margin (GMPM)
criterion [65], Opposing Argument criterion [66], Energy
larger than its adjacency, the order of the HSS model can be Source Analysis Consortium(ESAC) criterion [67] and per-
effectively reduced without causing an unacceptable error. turbation based criterion [68].
In [60], the HSS model of a single-phase Second-Order However, the criterion used most for combined AC, DC
Generalized Integrator (SOGI)-PLL is built, improving the networks is the Nyquist criterion in which the loop gains
accuracy of the controller behavior compared to LTI models. (ZS /ZL ) of the system described in (1) should satisfy the
Further, the nonlinear PLL dynamic is considered in state- Nyquist stability while the converter system being studied is
space modeling in [61], which leads to a nonlinear modular internally stable. Moreover, when analyzing a system with
state-space model. Time-domain simulations show compa- multiple converters, this stability assessment should be per-
rable accuracy with the detailed switching model in both formed on each one with others considered as loads, so that
transient and steady-state. Though this method is still based all could satisfy the stability criterion. In [38], due to the
on traditional state-space thus high-frequency ripples caused special structured reciprocal SISO circuit model built in the
by switching dynamics are not present, the way of processing polar-frame, MIMO inverter systems could also be analyzed
the low-frequency nonlinearity is also constructive to HSS with the Nyquist criterion.
modeling. In addition, the Generalized Nyquist Criterion (GNC) can
By reason of the foregoing, the derivation process of the be adopted in order to access the stability of the multi-port
two methods follows the same steps with different order and impedance system. Namely, the total numbers of clockwise
variable representation [43], thus leading to real states with and counterclockwise encirclements of the eigenvalues of
the DPM while complex states with the HSS model [62]. The the impedance ratio around the complex plane critical point
flow charts of the derivation processes are shown in Fig. 4. (−1, j0) should be equal to the number of its open-loop right-
The linearization process of DPM is encircled by a dashed half-plane (RHP) poles. Consequently, for minimum phase
line as it is not performed when modeling transient behavior. systems, the stability is guaranteed as long as the eigenloci
Similarly, if the state-space model used for the HSS is not don’t encircle (−1, j0). However, it is pointed out that for
linearized for small-signal analysis, the consequent HD model weak grid where non-minimum phase systems are becoming
is also accurate with the dynamic behavior, which is known dominant, Bode plots of eigenvalues tend to fail for system
as the dynamic harmonic (DHD) model [11], [63]. Due to stability analysis [69]. This problem was addressed in [70]
the difference in linearization principle, the HSS model is by identifying the open-loop RHP poles through magnitude
linearized around the time-periodic trajectories, whereas the slope and phase change rate variation. Moreover, a design-
DP model is based on linearization around the time-invariant oriented impedance specification procedure utilizing a Bode
points. Consequently, as the HSS model is the Fourier decom- plot of individual impedances with a special focus on crossing
position of an existing state-space model without changing direction is proposed to stabilize interconnect systems. Note
the model topology, it is intrinsically more programmable and that the determination of the crossing number at 0 Hz is very
could be executed automatically by computer programming. important in this method, which is discussed in [71].
In general, the DPM has the advantage that the TD simula- Inspired by the participation factor (PF) analysis in the
tion is either independently or interfacing with power system state-space model, the eigenvectors of the MIMO impedance
EMT models. However, the HSS is relatively less complex ratio (or return ratio) matrix are used to identify the PFs of
for impedance-based stability analysis and a larger amount of each converter in a multiconverter system when instability
harmonics can be taken into account [64]. appears [72]. The key converters that trigger the instability can
be found accordingly. The installation locations of converters
can also be chosen to ensure better stability margins. In [73],
III. STABILITY ASSESSMENT METHODS the residue of each node admittance is instead utilized to
The above-discussed modeling methods serve as the foun- define the PF. Based on the chain rule, a three-layer gray-box
dation to evaluate the system dynamics and stability against method is developed to estimate the critical node and the

456 VOLUME 3, 2022

 output admittance of two converters in parallel. It can be seen
−1 −1 −1
that the parallel admittance Zinv = Zinv1 + Zinv2 still lies
within the expected passivity region. This provides a decen-
tralized way of identifying the overall system stability [16].
Moreover, the source and load concept is no longer necessary
to be defined. Consequently, with the purpose of increasing
the overall system passivity [80], the control structure could
be simplified while maintaining a high control bandwidth.
However, as proven in [16], the passivity margin criterion is a
sufficient and unnecessary condition. Thus, a trade-off has to
FIGURE 5. Schematic diagram of the passivity region for two converters in be made between the stability objective and the conservation
parallel.
of the controller design.

exact passive component or control loop of it that affects the B. STATE-SPACE EIGENVALUE ANALYSIS
stability of the whole system in a perturbation and observation The eigenvalue analysis has been a widely used method for
(P&O) manner. Further, the PF analysis is extended from accessing the stability and oscillation mode of the traditional
the oscillation frequency points to critical frequency ranges power systems. The state matrix A of a small-signal linearized
where the eignloci are close to or tend to encircle (−1, j0) LTI model can be extracted to form the characteristic equa-
[74]. Active damping techniques can be adapted accordingly tion:
to enlarge the PM and GM of the critical converters within
critical frequency ranges. det(sI − A) = 0, (30)
The norm-based stability criterion has been proposed to
simplify analyses and indicate the stability margin levels [75]– the resultant eigenvalues could be an indication of the dy-
[77]. Nevertheless, the resultant stability criteria are actually namic modes of the power system. And the participation rate
sufficient and unnecessary conditions because the system can calculated by the corresponding left and right eigenvectors
be stable even if its stability requirements are not met. To can also reveal the impact of the grid-connected converters on
further reduce the conservatism, the relevant stability margin the instability [4]. However, the PFs of the state-space model
was regulated according to Gerschgorin’s Theorem when the are all scalars instead of transfer functions with frequency-
infinity-norm is used [77]. domain impedance ratio matrices. Thus, they can only predict
the PFs of each node under oscillation modes instead of
2) PASSIVITY-BASED CRITERIA port dynamic over a frequency range [74]. That is, the PF
Passivity is mathematically defined when a one port network analysis with state-space models offers limited guidance in
system only absorbs energy from the outside physical world, controller design to optimize the oscillation damping. This
namely: problem is solved by nonlinear modular state-space modeling
 +∞  +∞ of power-electronics systems [61], in which the VSCs models
1   are based on reduced-order impedance model and nonlinear
v(t )i(t )dt = V jω I ∗ (jω)dω
−∞ 2π −∞ PLL dynamic is considered. By tuning different parameters,
 the nonlinear model can be linearized at different equilibrium
1 +∞  
= Re Z (jω) I (jω) 2 jω ≥ 0. points with the resultant Jacobian matrix. The eigenvalues
π 0 distribution trend gives information on the system damping
(29) ability of different oscillation modes in response to the pa-
The stability of a passive system is inherently guaranteed be- rameter change.
cause the energy dissipation process always drives the system With the increase of control complexity and power elec-
back to its equilibrium point [78]. Consequently, the passivity tronic converter penetration rate, the order of the system
criterion of the one port system can be defined as [79]: becomes larger. Thus, the difficulty of building and deriving
r Z (s) does not have right half plane (RHP) poles and the the system model increases significantly. This problem was
entire Nyquist curve is within the RHP solved by the CCM, which defines each critical part as a sub-
r Z (s) has a non-negative real part for every ω system and connects them according to their linear algebraic
Compared with impedance-ratio-based methods, passivity- relation. Its modular structure also improves the extensibility
based methods are more suitable for multi-converter systems. of the model [5].
When the impedance of the converters within the system all
satisfy the passivity margin criterion, their summation will C. FLOQUET-THEORY-BASED METHOD
also be in between the passivity region with an even larger The Floquet theory illustrates the relationship between LTP
passivity margin. For example, as depicted in Fig. 5, a pas- and LTI systems [81]. By introducing the time-periodic ma-
sivity margin angle of α is required for the system to make trix P(t, t0 ), any LTP system can be transformed into an LTI
−1 −1
sure possible resonance is well damped. Zinv1 and Zinv2 are the model. According to [82], the solution of an LTP system (22)

VOLUME 3, 2022 457

PENG ET AL.: MODELING TECHNIQUES AND STABILITY ANALYSIS TOOLS FOR GRID-CONNECTED CONVERTERS

can be expressed as:

x(t ) =  (t, t0 ) x (t0 )
 t
+ (t, τ )B(τ )u(τ )dτ, t ≥ t0 , (31)
t0

where the matrix (t, t0 ) is a state transition matrix and is a

periodic time-varying matrix with period Ts . If P(t, t0 ) can be
numerical obtained, the original time-varying state transition
matrix (STM) A(t ) can be transferred into a time-invariant
matrix Q by: FIGURE 6. Phase plane global stable region and disturbance trajectory of
the SRF-PLL.
P (t, t0 ) =  (t, t0 ) e−Q(t−t0 ) . (32)
The eigenvalues of Q are called Floquet exponents (FEs)
and can be used to illustrate the system dynamics [83]. If
P(t, t0 ) can not be numerical obtained (which is true in most
cases), the stability of an LTP system is instead analyzed
mainly by (TA , 0), where TA is the minimum period [84].
The eigenvalues of (TA , 0) are called Floquet multipliers
(FMs). In [85], a complete model is derived for an asymmetric
single-phase cascaded H-bridge inverter. Different steady-
state operating trajectories are considered, loci and moduli
of resultant FMs are used to evaluate the system stability.
Compared to FEs, FMs can only indicate the system stabil-
ity instead of characterizing the oscillation frequency in the FIGURE 7. Classification of the methods discussed in this paper.
fundamental strip [14]. The relationships between different
analysis methods are summarized in [86], where an LTP-
MMC model is built and a numerical solution of the constant
matrix Q in the TD is used to calculate the FEs and analyze
the interaction between different controllers under unbalanced
grid conditions. Further, it is proved that FEs provide a physi-
cally sound way for truncation harmonic number selection of
HSS models for eigenvalue analysis [87].

D. PHASE PORTRAIT FIGURE 8. Application range of different modeling methods.

Unlike the above-mentioned methods, which access the small-

signal performance with linearized models, phase portrait stable equilibrium points. Disturbances within each small
methods focus on the large-signal stability (transient sta- zone will converge to the unique stable point. By that, the
bility). Namely, the ability of grid-connected converters to phase portrait offers a visualized way to access the dynamic
maintain synchronism when severe grid disturbances occur or performance with both frequency and phase jump, while oscil-
when the fundamental frequency varies [88]. Simultaneously, lation can also be confirmed when under specific disturbance.
a large signal model needs to be built in order to capture the In [21], based on the DPM, the time-domain simulation and
dynamics under serious phase jumps. Taking the model of phase plane trajectory of a grid-forming converter is given
the SRF-PLL as an example, different from the small-signal for different control strategies. Combined with the impedance
model as shown in (2), which is based on the assumption model and eigenvalue analysis, the coupling effect between
that the phase angle difference is sufficiently small, the large- the power-related control and inner closed-loop control is
signal model is expressed as: further revealed.

θ
˙ = ω
(33) IV. DISCUSSION
˙ = −ki,pllV sin θ − kp,pllV ω cos θ
ω
Thus far, the above discussed modeling and stability analysis
Where θ and ω denotes the difference between the real methods can be categorized based on the modeling domain as
value of the phase angle and the frequency and their estima- shown in Fig. 7, and the corresponding application range [89]
tion, V is the amplitude of the grid voltage. Based on (33), the is depicted in Fig. 8. A comparison of some main features
global stability of the SRF-PLL was studied in [19]. of the modeling approaches is given in table. 3. For stiff
The resultant phase portrait is shown as Fig. 6, which sep- grids with large inertia provided by synchronous generators,
arates the operation range into multiple small zones around PWM modulation of converters usually won’t affect the whole

458 VOLUME 3, 2022

TABLE 3. Comparison of the Modeling Approaches

system stability. Also, power-related controller or synchro-

nization unit and inner controller can be considered separately
because the coupling is neglectable. So the common approach
is to analyze the large-signal synchronous stability (frequency
dynamics) with TD EMT equations [20], [21] and simplified
symmetrical small-signal models for aiding the inner con-
troller design [90].
However, in weak grids like renewable power plants and
microgrids, the large impedance and frequency multiplication
of different converters enhance the coupling between different FIGURE 9. Structure diagram of the grid-connected converter and its
control loops and synchronization units. In this case, different control.
control loops should be considered comprehensively when
building the model. In particular, the transfer function models
TABLE 4. Model Parameters of the Grid-Connected Converter
can be used to identify the frequency response of the terminal
currents and voltages usually caused by the PLL, the inner
voltage and current control loop and passive filter components
(voltage dynamics). Thus, the bode plots of different ports
can be used as guidance for controller design. Also, both
impedance ratio and passivity-based methods allow black-box
operation, so a detailed system configuration is not neces-
sary. The major difference between these two methods is that
passivity-based methods are decentralized, which means the
system is stable as long as all units are passive. However, the Fourier inversion and FD with Harmonic Transfer Function
passivity-based criteria are over-conservative and thus may (HTF). Besides, Floquet theory for analyzing LTP systems
lead to poor dynamic performance. Impedance-based methods can also be used for stability analysis and oscillation mode
instead could identify the critical node that may cause insta- identification [85], [87]. Compared to classic TD and FD
bility and give constrictive information on parameter tuning methods, it gives an insight into the switching behavior and
[73], [74]. harmonic coupling. The frequency of interest is then expended
Time-domain state-space model instead can also provide to switching dynamics which is extremely important in power
insights into the coupling path based on the participation electronics dominate microgrids with different and a synchro-
factors [21]. Note that the TD state-space model is actually a nized switching frequencies, making it a solid supplement to
series of differential equations. Thus, it can be both linear [48] traditional modeling. However, the above-mentioned methods
and nonlinear [61] based on different modeling assumptions. for stability assessment still need transform or indirect execu-
The nonlinear modular state-space models can also be lin- tion. In [17], based on Harmonic Riccati Equation, globally
earized around different operating points, the elements of the and asymptotically stabilizing periodic feedback control laws
resultant Jacobian matrix can also be transformed to transfer are proposed which directly come from HSS models without
function and FD methods can be used. Also, the transient sta- invoking the Floquet theory.
bility of the synchronization units or power-related controllers
can be analyzed with the help of the nonlinear model in the
phase plane. However, a major drawback of TD modeling is V. CASE STUDY
that detailed data is needed which is not feasible in a large- In this section, the above-mentioned modeling and stability
scale interconnected network. assessment methods will be utilized to analyze a two level 3-
The harmonic modeling is a combination of TD and FD. phase grid-connected converter with proportional-integral (PI)
Thus, it can be analyzed with some of the existing methods in controller for the DC-link voltage control and the dq-domain
both tracks. More specifically, it can be transferred to TD with current controller as shown in Fig. 9.

VOLUME 3, 2022 459

PENG ET AL.: MODELING TECHNIQUES AND STABILITY ANALYSIS TOOLS FOR GRID-CONNECTED CONVERTERS

FIGURE 10. Small-signal model of the grid-connected converter system.

Theoretical analysis will be compared against the simula-

tion results. The parameter is shown in Table 4 . The design
reference of the controller can be found in [91].

A. TRANSFER FUNCTION-BASED SMALL-SIGNAL ANALYSIS

The αβ-domain model is widely used in the modeling of FIGURE 11. dq-domain converter input admittance.
VSCs [33] and has been proved to be interchangeable with
dq-domain model [92]. In this paper, dq-domain model is
carried out as the controller is designed within SRF. The dq-
domain small-signal model of the converter system is shown
in Fig. 10. Gdc and Gi are the PI controller of the DC-link
voltage and dq-domain current, with Gdei denotes the decou-
pling term. Gdel and GPW M are the first-order delay and PWM
drive coefficient, respectively, the subscript of electrical state s
denotes the real signal with the physical system and c denotes
the signal detected by the controller. G pv is the linearized
transfer function between the active power disturbance and
the DC-link voltage response [92], which can be expressed FIGURE 12. PLL frequency error signal. (a) Bandwidth of 3 Hz (b)
Bandwidth of 10 Hz.
as:
dV dc 1
G pv = = . (34)
dP sCdc Vdc0 + Idc0 Fig. 12 illustrates the variation in the PLL frequency error
signals, with the disturbance of the same amplitude and wave-
The dashed line parts illustrate the small-signal disturbance
form, when the bandwidth is 3 Hz and 10 Hz. Thus, the
introduced by the PLL, where the small-signal transfer func-
measured impedance in the coupling admittance may have a
tion is shown in (2). The d denotes the small-signal distur-
relatively larger bias.
bance, and subscript 0 denotes the steady-state value. As the
Combined with the LC-type impedance of the transmis-
system is a SISO system but has two state variables in d
sion line, GNC-based analysis can be performed as shown in
and q-channel respectively, expanding matrices are used as
Fig. 13, where the two eigenvalues of the system are depicted
depicted in the figure in order to generate the small-signal
in both Nyquist and Bode plots. The zoom-in plot in Fig. 13(a)
variable on q-channel. The derivation process of each block
shows that the characteristic root locus plots of the system
and the impedance measurement algorithm has expatiated
do not encircle (−1, 0) point, the system is thus stable. How-
in [28], [92].
ever, combined with Fig. 13(b) the root locus of λ2 intersects
The bode plot of the derived admittance by this model is
the unit circle at approximately 715 Hz, which could trigger
shown in Fig. 11 and compared with the measurement re-
convergent oscillation around this frequency under transient
sults obtained by a simulation conducted with the software
event.
Plexim PLECS. It can be observed that the d-d channel ad-
mittance Ydd of the converter system has a positive increment
in the low-frequency range, which is known as the constant B. TRANSIENT STABILITY ANALYSIS WITH PHASE
power load (CPL) behavior. Such phenomena are introduced PORTRAIT
by the DC-link voltage control loop and could cause low- As the bandwidth of the inner current loop is much larger than
frequency oscillation and voltage overshoot. In contrast, the the bandwidth of the PLL, the dynamic of which is always
low-frequency phase jump in q-axis is mainly caused by the neglected when studying the effect of the PLL. Considering
PLL as the input current reference from Gv is always 0. Note the grid-PLL interaction in the above-mentioned system and
that the disturbance signal within the bandwidth of the PLL assuming that the impedance is static, a second-order nonlin-
may trigger a nonlinear response in the error signal output. ear model of the dynamic around the fundamental frequency

460 VOLUME 3, 2022

 FIGURE 15. Simulated phase A capacitor voltage of the HSS model.

with dynamic phasor is derived in [21]. However, the model

FIGURE 13. Eigenvalue of the simulated impedance from the small-signal
model. is so complicated that it cannot be described by a first- or
second-order non-linear differential equation. Therefore, the
stability analysis is based on a reduced-order time-domain
simulation framework rather than a design-oriented graphical
analysis with the trajectory in the phase plane.

C. HARMONIC DOMAIN MODELING AND STABILITY

ANALYSIS
In order to access the HD dynamics of the converter system,
the HSS model, having flexibility in harmonics order, is used
to develop the system model in this paper. The open-loop HSS
FIGURE 14. PLL frequency error signal under frequency drop.
model is first built for the converter, which is also referred to
as the topology model [93]. Time-domain discrete Forward-
can be derived [20]: Euler iteration of the model is performed so as to illustrate
ki,pll   
the accuracy in time-domain simulation. The simulation result
ω̇ = Id ωgn + ω Xline + Iq Rline and comparison with PLECS are shown in Fig. 15. It can
1 − kp,pll Id Ẋline be seen that the major harmonic components in both models
 kp,pllVamp cos δ agree with each other during the steady-state of the open-loop
− Vamp sin θ − ω. (35) operation.
1 − kp,pll Id Ẋline
To include the controller behavior, the model needs to be
Where Vamp is the amplitude of the grid voltage, Rline and Xline linearized around the time-periodic trajectories calculated by
are the line resistance and reactance. According to the global the open-loop topology model. Assuming the switching in-
stability zone discussed in [19], the phase domain stability stant of the controller output signal is very small, the relation
zone of the converter system is shown in Fig. 6. An 8 Hz between AC and DC electrical states can be expressed as:
drop in PLL reference frequency is added to the system so
as to get the trajectory in the phase plane. Compared with the dvoa (t ) = Sa0 dvdc + vdc0 dSa
PLECS simulation shown in Fig. 14, the large-signal model
provides an envelope curve of the PLL transient response with dvob (t ) = Sb0 dvdc + vdc0 dSb
10.5% larger predicted amplitude within the second oscilla- dvoc (t ) = Sc0 dvdc + vdc0 dSc
tion. Due to the absence of the inner current loop dynamic,
the model fails to predict the high-frequency oscillation which didc = Sa0 dica + ica0 dSa
can be seen in the characteristic loci shown in Fig. 13 and + Sbodicb + icb0 dSb
the switching component that will be discussed in V-C. In
contrast, a more dedicated model considering the inner loop + Scodicc + icc0 dSc . (36)

VOLUME 3, 2022 461

PENG ET AL.: MODELING TECHNIQUES AND STABILITY ANALYSIS TOOLS FOR GRID-CONNECTED CONVERTERS

⎡ ⎤
Where d denotes small variation of signals. This principle Dc1 Dc2 − 23 k pc I m 1
3 k pc I m
1
3 k pc I m
also applies for HSS models. Thus, the linearized ⎢ ⎥
Dc = ⎣ Dc3 Dc4 1
3 k pc I m − 23 k pc I m 1
3 k pc I m ⎦
(42)
close-loop HSS model can be built by considering the
small-signal instant of the topology electrical states and Dc5 Dc6 1
3 k pc I m
1
3 k pc I m − 23 k pc I m
the controller. Set the topology state variables as X p =
 T Where:    
dI ga dI gb dI gc dV Ca dV Cb dV Cc dI ca dI cb dI cc dV dc , inputs Bc1 = −  23 cos θ C = k pc  cos(θ − 23 π )
 T  c1  
U p = dV pcca dV pccb dV pccc dSa dSb dSc . Moreover, the Bc2 = − 3 cos(θ − 3 π ) C c2
2 2
= kic  cos(θ − 23 π )
 T    
controller state variables are set as X c = dX 1 d dX 2 d dX q , Bc3 = − 23 cos(θ + 23 π ) C c3 = −kic  sin(θ − 23 π )
 T  
inputs U c = dV ∗dc dV dc dI ca dI cb dI cc and outputs  
Bc4 =  23 sin θ C c4 = k pc  cos(θ + 23 π )
Y c = [dSa dSb dSc ]T , where d denotes the small-signal 2   
disturbance and the physical meaning of the variables Bc5 =  3 sin(θ − 3 π )2
C c5 = kic  cos(θ + 23 π )
2   
are shown in Fig. 9. Assuming that the PLL can track Bc6 =  3 sin(θ + 3 π )2
C c6 = −kic  sin(θ + 23 π )
the grid frequency perfectly, ideal Park and Inverse Park Dc1 = k pv k pc [cos θ ]
transformation are used in the controller model to link the
static and synchronized reference frames. Consequently, the Dc2 = −k pv k pc  [cos(θ )]
 
HSS model of the converter topology and the controller can Dc3 = k pv k pc  cos(θ − 23 π )
be derived as: (37) shown at the bottom of this page.  
Dc4 = −k pv k pc  cos(θ − 23 π )
 
⎡ ⎤ Dc5 = k pv k pc  cos(θ + 23 π )
1  
L2 I m ZmZm Zm Zm Zm
Dc6 = −k pv k pc  cos(θ + 23 π ) .
⎢ Zm 1
Zm ⎥
⎢ L2 I m
Zm Zm Zm ⎥ I m and Zm denote the m = (2n + 1)-order identity matrix
⎢ Z 1 ⎥
⎢ m Zm I m Z m Z m Z m ⎥ and zero matrix as the Toeplitz matrix of these two kinds
⎢ L2
⎥
⎢ Zm ZmZm Zm Zm Zm ⎥ are themselves. n denotes the number of harmonics being
⎢ ⎥
⎢ Zm ZmZm Zm Zm Zm ⎥ considered in the model.
Bp = ⎢⎢ Zm ZmZm Zm Zm Zm ⎥
⎥ (38) For symmetrical Sinusoidal PWM modulation (SPWM),
⎢ −[V ⎥
⎢ Zm ZmZm dc0 ]
Zm Zm ⎥ the Fourier decomposition of the switching signal in upper
⎢ L1 ⎥
⎢ Z Zm −[V dc0 ]
Zm ⎥ arm of phase A can be expressed as [94]: (43) shown at the
⎢ m ZmZm ⎥
⎢ L1
⎥ bottom of the next page. Where Mr is the modulation rate, ωsw
⎣ Zm ZmZm Zm Zm −[V L1
dc0 ]
⎦
and ω0 are the switching and fundamental angular frequency,
Zm ZmZm [ICcao] [IC cbo] [I cco ]
Cdc m and n denote the order of the carrier wave and the adja-
dc dc
⎡ ⎤ cent harmonic wave, respectively. The expression besselj(·)
−N Zm Zm
denotes the Bessel function of the first kind, which is:
Ac = ⎣ kiv I m −N Zm ⎦ (39)
(−1)k  x (2k+n)
Zm Zm −N ∞

⎡ ⎤ besselj(n, x) = . (44)
Im −Im Zm Zm Zm k!(k + n)! 2
k=0
Bc = ⎣ k pv I m −k pv I m Bc1 Bc2 Bc3 ⎦ (40)
Zm Zm Bc4 Bc5 Bc6 Other signals can be derived by changing the initial phases of
⎡  ⎤ the mathematical expression. The resultant time-series wave-
k pc  [cos θ] kic  [cos θ ] −kic  23 sin θ form is shown in Fig. 16, the modulation rate is 15 and the
Cc = ⎣ C c1 C c2 C c3 ⎦ (41) harmonics calculated is 255th . The switching function pro-
C c4 C c5 C c6 vides a link between DC and AC components and reproduces

Ap =
⎡ −R2 −1 ⎤
L2 Im − N Zm Zm L2 I m Zm Zm Zm Zm Zm Zm
⎢ −R2 −1 ⎥
⎢ Zm L2 I m − N Zm Z m L2 I m Zm Zm Zm Zm Zm ⎥
⎢ −R2 −1 ⎥
⎢
⎢
Zm Zm L2 I m − N Zm Zm L2 I m Zm Zm Zm Zm ⎥
⎥
⎢ 1
−N Zm −1 ⎥
⎢ C Im Zm Zm Zm C Im Zm Zm Zm ⎥
⎢ Zm 1
Zm −N −1 ⎥
⎢ C Im Zm Zm Zm C Im Zm Zm ⎥
⎢ Zm Zm 1
−N −1 ⎥ (37)
⎢ C Im Zm Zm Zm Zm C Im Zm ⎥
⎢ −R1
− [S a0 ] ⎥
L1 I m − N
1
⎢ Zm Zm Zm L1 I m Z m Zm Zm Zm L1 ⎥
⎢  S ⎥
⎢ Zm Zm Zm Zm 1 −R1
−N − [L1b0 ] ⎥
⎢ L1 I m Zm Zm L1 I m Zm ⎥
⎢ ⎥
⎣ Zm Zm Zm Zm Zm 1
L1 I m Zm Zm −R1
L1 m − N
I − [S L1
c0 ]
⎦
[Sa0 ] [Sb0 ] [Sc0 ] −1
Zm Zm Zm Zm Zm Zm Cdc Cdc Cdc RLoad Cdc I m − N
462 VOLUME 3, 2022
 based on the relation in Fig. 17, the close-loop HSS state
iteration can be derived as:
.
dX p = A p dX p + B p1 dU p1 + B p2 dY c
= A p dX p + B p1 dU p1
 
+ B p2 C c dX c + Dc1 dU c1 + Dc2 dX p1
= A p dX p + B p2 Dc2 dX p1 + B p2C c dX c
FIGURE 16. Fourier Decomposition of PWM switching signal. + B p1 dU p1 + B p2 Dc1 dU c1 , (45)
dẊ c = Ac dX c + Bc1 dU c1 + Bc2 dU c2
= Ac dX c + Bc1 dU c1 + Bc2 dX p1 . (46)

Subsequently, fill the empty elements with zero matrices, the

close-loop state matrix can be derived as:
 .    
dX p A p +[B p2 Dc2 |Z ] B p2C c dX p
. =
dX c [Bc2 |Z ] Ac dX c
  
B p1 B p2 Dc1 dU p1
FIGURE 17. Schematic diagram of the close-loop HSS model. + . (47)
Z Bc1 dU c1

the harmonics introduces by the switching behavior. Thus, the Further, to obtain the I/O relation between different electri-
more harmonics being calculated, the more it appears like a cal states, set the combined states as dXa , voltage disturbance
square wave. at the point of common connection (PCC) and DC-link volt-
Fig. 17 shows the schematic diagram of the combined age reference as input dUa , the state variables of the topology
topology and controller model. As illustrated by the dashed as the outputs dYa , and the resultant state matrices as Aa , Ba ,
blue arrow, the state variable of the topology model is the C a and Da . The HTF can be established as [95]:
input of the control model, and the outputs of the control
model act as inputs of the topology model. This coupling rela- Ha (s) = C(sI − Aa )−1 Ba + Da . (48)
tionship between the two systems requires further regulation
of the state matrix to form the compacted linearized system, By setting s as zero and considering up to the 35th order
but the deviation process is not shown in [93]. One way to harmonic components, a steady-state linear mapping between
combine the two parts and eliminate the duplicate states is the Fourier coefficients of the converter electrical states and
shown below. voltage disturbance can be built as shown in Fig. C. The
In order to derive the compacted model, first consider the figure can be interpreted as HD coupling admittance since
variation of the DC-link voltage reference signal as controller the converter input current is seen as the output of the HTF.
input set U c1 , the DC-link voltage and converter output cur- It can be seen that voltage disturbance at a fundamental fre-
rent disturbance as controller input set U c2 , the disturbance in quency can trigger a current disturbance at both fundamental
the PCC voltage as topology input set U p1 and the switch- frequency and around the switching frequency and vice versa.
ing signal as topology input set U p2 and grid side current Besides, the resonant zone can also be predicted as encircled
disturbance. Separate the input matrix of the two systems in red, which is consistent with the major components of the
accordingly and merge the state matrices of the two systems THD shown in Fig. 14(b).

1 ma
sa (t ) = + sin (ω0t + θ0 )
2 2
⎡  
2⎣
+∞

∞

besselj n, mM2r π  mπ 
+ sin cos (mωswt + nω0t )
π π 2
m=1,3··· n=0,±2···
  ⎤
+∞

∞

besselj n, mM2r π  mπ 
+ cos sin (mωswt + nω0t )⎦ . (43)
π 2
m=2,4··· n=±1,±3...

VOLUME 3, 2022 463

PENG ET AL.: MODELING TECHNIQUES AND STABILITY ANALYSIS TOOLS FOR GRID-CONNECTED CONVERTERS

FIGURE 20. Converter output voltage under phase jump.

FIGURE 18. Close-loop input admittance magnitude map of VSC HSS

model.

FIGURE 21. PLL frequency error signal under frequency drop.

VI. CONCLUSION
FIGURE 19. HiL test setup. In this paper, three mainstream modeling methods and rel-
evant analysis tools are discussed for the study of grid-
connected converters in the following aspects: derivation pro-
D. HARDWARE-IN-THE-LOOP VALIDATION cess, key feature, and application scenarios. The small-signal
In order to validate previous analysis, HiL tests with the aid impedance model deals with low-frequency and synchronize
of PLECS RT-Box have been done. The setup of HiL test stability issues. The TD nonlinear large-signal model can
is shown in Fig. 19. The system parameters are the same as also be used to aid the controller design with the visualized
shown in Table 4. phase plane portrait. Besides, HD modeling methods could
Similar to the simulation, an 8 Hz frequency drop is added extend the range of interest to the switching frequency and the
to the PLL reference frequency is triggered at 0.45 s. The wide-band controller dynamic. More specifically, the DPM
resultant converter output voltage and Fast Fourier transform could be used for both large-signal TD simulation and stability
results are shown in Fig. 20 and the PLL error signal is analysis with the system order properly selected, while HSS is
shown in Fig. 21. The medium frequency at around 700 Hz more suitable for steady-state harmonic stability assessment
can be observed in Fig. 20(b), which agrees with the result due to its LTP nature and complex-valued state matrices. A
of the GNC analysis. The PLL frequency error signal con- case study with a three-phase VSC further validates these
forms to the phase plane trajectory, the dynamic is better methods and gives a paradigm on how they are utilized with
damped than simulation which is caused by sampling delay in PWM converters. This critical evaluation of the modeling and
real-time system. Also, the magnitudes harmonic components analysis tools gives the engineers theoretical guidance for the
are consistent with the major components shown in the HSS design and evaluation of power electronics-dominated power
admittance map. systems.

464 VOLUME 3, 2022

REFERENCES [23] P. Krein, J. Bentsman, R. Bass, and B. Lesieutre, “On the use of aver-
[1] G. Buticchi et al., “The role of renewable energy system in reshaping aging for the analysis of power electronic systems,” IEEE Trans. Power
the electrical grid scenario,” IEEE Open J. Ind. Electron. Soc., vol. 2, Electron., vol. 5, no. 2, pp. 182–190, Apr. 1990.
pp. 451–468, 2021, doi: 10.1109/OJIES.2021.3102860. [24] J. Yang, G. Buticchi, C. Gu, S. Gunter, H. Zhang, and P. Wheeler,
[2] E. Mollerstedt and B. Bernhardsson, “Out of control because of “A generalized input impedance model of multiple active bridge con-
harmonics-an analysis of the harmonic response of an inverter loco- verter,” IEEE Trans. Transport. Electrific., vol. 6, no. 4, pp. 1695–1706,
motive,” IEEE Control Syst. Mag., vol. 20, no. 4, pp. 70–81, Aug. Dec. 2020.
2000. [25] J. Sun, “Small-signal methods for AC distributed power systems–A
[3] G. Buticchi, S. Bozhko, M. Liserre, P. Wheeler, and K. Al-Haddad, review,” IEEE Trans. Power Electron., vol. 24, no. 11, pp. 2545–2554,
“On-board microgrids for the more electric aircraftreview,” IEEE Trans. Nov. 2009.
Ind. Electron., vol. 66, no. 7, pp. 5588–5599, Jul. 2019. [26] M. Belkhayat, Stability Criteria for AC Power Syst. With Regulated
[4] X. Wang and F. Blaabjerg, “Harmonic stability in power electronic- Loads. West Lafayette, IN, USA: Purdue Univ., 1997.
based power systems: Concept, modeling, and analysis,” IEEE Trans. [27] B. Wen, D. Boroyevich, R. Burgos, P. Mattavelli, and Z. Shen, “Anal-
Smart Grid, vol. 10, no. 3, pp. 2858–2870, May 2019. ysis of DQ small-signal impedance of grid-tied inverters,” IEEE Trans.
[5] G. Gaba, S. Lefebvre, and D. Mukhedkar, “Comparative analysis and Power Electron., vol. 31, no. 1, pp. 675–687, Jan. 2016.
study of the dynamic stability of AC/DC systems,” IEEE Trans. Power [28] B. Wen, D. Boroyevich, R. Burgos, P. Mattavelli, and Z. Shen, “Small-
Syst., vol. 3, no. 3, pp. 978–985, Aug. 1988. signal stability analysis of three-phase AC systems in the presence of
[6] R. D. Middlebrook and S. Cuk, “A general unified approach to constant power loads based on measured D-Q frame impedances,” IEEE
modelling switching-converter power stages,” in Proc. IEEE Power Trans. Power Electron., vol. 30, no. 10, pp. 5952–5963, Oct. 2015.
Electron. Specialists Conf., 1976, pp. 18–34. [29] Z.-X. Zou, R. Rosso, and M. Liserre, “Modeling of the phase detector
[7] M. Cespedes and J. Sun, “Impedance modeling and analysis of grid- of a synchronous-reference-frame phase-locked loop based on second-
connected voltage-source converters,” IEEE Trans. Power Electron., order approximation,” IEEE Trans. Emerg. Sel. Topics Power Electron.,
vol. 29, no. 3, pp. 1254–1261, Mar. 2014. vol. 8, no. 3, pp. 2534–2545, Sep. 2020.
[8] H. Liu et al., “Subsynchronous interaction between direct-drive PMSG [30] E. Unamuno, A. Rygg, M. Amin, M. Molinas, and J. A. Barrena,
based wind farms and weak AC networks,” IEEE Trans. Power Syst., “Impedance-based stability evaluation of virtual synchronous machine
vol. 32, no. 6, pp. 4708–4720, Nov. 2017. implementations in converter controllers,” in Proc. Int. Power Electron.
[9] J. Kwon, X. Wang, F. Blaabjerg, and C. L. Bak, “Comparison of LTI Conf., 2018, pp. 759–766.
and LTP models for stability analysis of grid converters,” in Proc. IEEE [31] Y. Xu, H. Nian, T. Wang, L. Chen, and T. Zheng, “Frequency coupling
17th Workshop Control Model. Power Electron., 2016, pp. 1–8. characteristic modeling and stability analysis of doubly fed induction
[10] T. Messo, A. Aapro, and T. Suntio, “Generalized multivariable small- generator,” IEEE Trans. Energy Convers., vol. 33, no. 3, pp. 1475–1486,
signal model of three-phase grid-connected inverter in DQ-domain,” Sep. 2018.
in Proc. IEEE 16th Workshop Control Model. Power Electron., 2015, [32] A. Rygg, M. Molinas, C. Zhang, and X. Cai, “A modified sequence-
pp. 1–8. domain impedance definition and its equivalence to the DQ-domain
[11] J. J. Chavez and A. Ramirez, “Dynamic harmonic domain modeling of impedance definition for the stability analysis of AC power electronic
transients in three-phase transmission lines,” IEEE Trans. Power Del., systems,” IEEE Trans. Emerg. Sel. Top. Power Electron., vol. 4, no. 4,
vol. 23, no. 4, pp. 2294–2301, Oct. 2008. pp. 1383–1396, Dec. 2016.
[12] S. Sanders, J. Noworolski, X. Liu, and G. Verghese, “Generalized [33] Y. Liao and X. Wang, “Stationary-frame complex-valued frequency-
averaging method for power conversion circuits,” IEEE Trans. Power domain modeling of three-phase power converters,” IEEE Trans.
Electron., vol. 6, no. 2, pp. 251–259, Apr. 1991. Emerg. Sel. Topics Power Electron., vol. 8, no. 2, pp. 1922–1933,
[13] P. Mattavelli and A. Stankovic, “Dynamical phasors in modeling and Jun. 2020.
control of active filters,” in Proc. IEEE Int. Symp. Circuits Syst., 1999, [34] C. Zhang, X. Cai, A. Rygg, and M. Molinas, “Sequence domain SISO
vol. 5, pp. 278–282. equivalent models of a grid-tied voltage source converter system for
[14] N. M. Wereley, “Analysis and control of linear periodically time varying small-signal stability analysis,” IEEE Trans. Energy Convers., vol. 33,
systems,” Ph.D. dissertation, Dept. Aeronaut. Astronaut., Massachusetts no. 2, pp. 741–749, Jun. 2018.
Inst. Technol., Cambridge, MA, USA, 1990. [35] Y. Zhang, X. Chen, and J. Sun, “Sequence impedance modeling and
[15] R. Ortega, “Passivity–based control of Euler–Lagrange systems: Appli- analysis of MMC in single-star configuration,” IEEE Trans. Power
cations to robots, AC motors and power converters,” in Proc. Model. Electron., vol. 35, no. 1, pp. 334–346, Jan. 2020.
Control Mech. Syst. World Sci., 1997, pp. 75–92. [36] I. Vieto and J. Sun, “Sequence impedance modeling and analysis of
[16] H. Yu, H. Tu, and S. Lukic, “A passivity-based decentralized control type-III wind turbines,” IEEE Trans. Energy Convers., vol. 33, no. 2,
strategy for current-controlled inverters in AC microgrids,” in Proc. pp. 537–545, Jun. 2018.
IEEE Appl. Power Electron. Conf. Expo., 2018, pp. 1399–1406. [37] J. Sun, “Two-port characterization and transfer immittances of AC-
[17] N. Blin, P. Riedinger, J. Daafouz, L. Grimaud-Salmon, and P. Feyel, DC converters I: Modeling,” IEEE Open J. Power Electron., vol. 2,
“Necessary and sufficient conditions for harmonic control in contin- pp. 440–462, 2021, doi: 10.1109/OJPEL.2021.3104502.
uous time,” IEEE Trans. Autom. Control, early access, Oct. 4, 2021, [38] L. Xu, H. Xin, L. Huang, H. Yuan, P. Ju, and D. Wu, “Symmetric
doi: 10.1109/TAC.2021.3117540. admittance modeling for stability analysis of grid-connected con-
[18] D. Sun, S. Wang, L. Du, and X. Lu, “Harmonic compensation con- verters,” IEEE Trans. Energy Convers., vol. 35, no. 1, pp. 434–444,
trol of grid interactive inverters based on data-driven harmonic state Mar. 2020.
space modeling,” in Proc. IEEE Energy Convers. Congr. Expo., 2021, [39] S. Shah and L. Parsa, “Impedance modeling of three-phase voltage
pp. 3134–3139. source converters in DQ, sequence, and phasor domains,” IEEE Trans.
[19] Z. Dai, G. Li, M. Fan, J. Huang, Y. Yang, and W. Hang, “Global Energy Convers., vol. 32, no. 3, pp. 1139–1150, Sep. 2017.
stability analysis for synchronous reference frame phase-locked loops,” [40] Z. Xu, B. Li, S. Wang, S. Zhang, and D. Xu, “Generalized single-phase
IEEE Trans. Ind. Electron., vol. 69, no. 10, pp. 10182–10191, Oct. harmonic state space modeling of the modular multilevel converter
2022. with zero-sequence voltage compensation,” IEEE Trans. Ind. Electron.,
[20] H. Wu and X. Wang, “Design-oriented transient stability analysis vol. 66, no. 8, pp. 6416–6426, Aug. 2019.
of PLL-synchronized voltage-source converters,” IEEE Trans. Power [41] D. Maksimovic, A. Stankovic, V. Thottuvelil, and G. Verghese, “Model-
Electron., vol. 35, no. 4, pp. 3573–3589, Apr. 2020. ing and simulation of power electronic converters,” Proc. IEEE, vol. 89,
[21] M. Dokus and A. Mertens, “On the coupling of power-related and inner no. 6, pp. 898–912, Jun. 2001.
inverter control loops of grid-forming converter systems,” IEEE Access, [42] H. Konara, U. D. Annakkage, and C. Karawita, “Stability criterion for
vol. 9, pp. 16173–16192, 2021. interfacing a transient stability model to a dynamic phasor model,” in
[22] L. Jessen, S. Gunter, F. W. Fuchs, M. Gottschalk, and H.-J. Hinrichs, Proc. IEEE/PES Transmiss. Distrib. Conf. Expo., 2018, pp. 1–5.
“Measurement results and performance analysis of the grid impedance [43] Y. Liao and X. Wang, “Small-signal modeling of AC power
in different low voltage grids for a wide frequency band to support electronic systems: Critical review and unified modeling,”
grid integration of renewables,” in Proc. IEEE Energy Convers. Congr. IEEE Open J. Power Electron., vol. 2, pp. 424–439, 2021,
Expo., 2015, pp. 1960–1967. doi: 10.1109/OJPEL.2021.3104522.

VOLUME 3, 2022 465

PENG ET AL.: MODELING TECHNIQUES AND STABILITY ANALYSIS TOOLS FOR GRID-CONNECTED CONVERTERS

[44] Q. Zhong, Q. Shi, G. Wang, and H. Li, “Short circuit analysis of PV [64] J. Lyu, X. Zhang, J. Huang, J. Zhang, and X. Cai, “Comparison of
inverter under unbalanced conditions with dynamic phasor sequence harmonic linearization and harmonic state space methods for impedance
components,” in Proc. IEEE Power Energy Soc. Gen. Meeting, 2016, modeling of modular multilevel converter,” in Proc. Int. Power Electron.
pp. 1–5. Conf., 2018, pp. 1004–1009.
[45] A. Nazari, Y. Xue, J. K. Motwani, I. Cvetkovic, D. Dong, and D. [65] C. Wildrick, F. Lee, B. Cho, and B. Choi, “A method of defining the load
Boroyevich, “Dynamic phasor modeling of multi-converter systems,” impedance specification for a stable distributed power system,” IEEE
in Proc. IEEE Energy Convers. Congr. Expo., 2021, pp. 2716–2721. Trans. Power Electron., vol. 10, no. 3, pp. 280–285, May 1995.
[46] Z. Shuai, Y. Peng, J. M. Guerrero, Y. Li, and Z. J. Shen, “Transient [66] X. Feng, J. Liu, and F. Lee, “Impedance specifications for stable DC
response analysis of inverter-based microgrids under unbalanced con- distributed power systems,” IEEE Trans. Power Electron., vol. 17, no. 2,
ditions using a dynamic phasor model,” IEEE Trans. Ind. Electron., pp. 157–162, Mar. 2002.
vol. 66, no. 4, pp. 2868–2879, Apr. 2019. [67] S. Sudhoff, S. Glover, P. Lamm, D. Schmucker, and D. Delisle, “Admit-
[47] D. Shu, X. Xie, V. Dinavahi, C. Zhang, X. Ye, and Q. Jiang, “Dynamic tance space stability analysis of power electronic systems,” IEEE Trans.
phasor based interface model for EMT and transient stability hybrid Aerosp. Electron. Syst., vol. 36, no. 3, pp. 965–973, Jul. 2000.
simulations,” IEEE Trans. Power Syst., vol. 33, no. 4, pp. 3930–3939, [68] J. Liu, X. Feng, F. Lee, and D. Borojevich, “Stability margin monitoring
Jul. 2018. for DC distributed power systems via perturbation approaches,” IEEE
[48] H. Wang, K. Jiang, M. Shahidehpour, and B. He, “Reduced-order Trans. Power Electron., vol. 18, no. 6, pp. 1254–1261, Nov. 2003.
state space model for dynamic phasors in active distribution net- [69] L. Fan and Z. Miao, “Admittance-based stability analysis: Bode plots,
works,” IEEE Trans. Smart Grid, vol. 11, no. 3, pp. 1928–1941, nyquist diagrams or eigenvalue analysis?,” IEEE Trans. Power Syst.,
May 2020. vol. 35, no. 4, pp. 3312–3315, Jul. 2020.
[49] D. Shu, Y. Wei, V. Dinavahi, K. Wang, Z. Yan, and X. Li, “Cosimu- [70] Y. Liao and X. Wang, “Impedance-based stability analysis for inter-
lation of shifted-frequency/dynamic phasor and electromagnetic tran- connected converter systems with open-loop RHP poles,” IEEE Trans.
sient models of hybrid LCC-MMC DC grids on integrated CPU– Power Electron., vol. 35, no. 4, pp. 4388–4397, Apr. 2020.
GPUs,” IEEE Trans. Ind. Electron., vol. 67, no. 8, pp. 6517–6530, [71] D. Lumbreras, E. L. Barrios, A. Urtasun, A. Ursúa, L. Marroyo, and
Aug. 2020. P. Sanchis, “On the stability of advanced power electronic converters:
[50] D. Venkatramanan and V. John, “Dynamic phasor modeling and sta- The generalized bode criterion,” IEEE Trans. Power Electron., vol. 34,
bility analysis of SRF-PLL-based grid-tie inverter under islanded no. 9, pp. 9247–9262, Sep. 2019.
conditions,” IEEE Trans. Ind. Appl., vol. 56, no. 2, pp. 1953–1965, [72] Y. Li et al., “Stability analysis and location optimization method for
Mar./Apr. 2020. multiconverter power systems based on nodal admittance matrix,” IEEE
[51] J. Vega-Herrera, C. Rahmann, F. Valencia, and K. Strunz, “Analysis Trans. Emerg. Sel. Topics Power Electron., vol. 9, no. 1, pp. 529–538,
and application of quasi-static and dynamic phasor calculus for stability Feb. 2021.
assessment of integrated power electric and electronic systems,” IEEE [73] Y. Zhu, Y. Gu, Y. Li, and T. C. Green, “Participation analysis in
Trans. Power Syst., vol. 36, no. 3, pp. 1750–1760, May 2021. impedance models: The grey-box approach for power system stability,”
[52] R. Dimitrovski and M. Luther, “Generalized dynamic phasor model of IEEE Trans. Power Syst., vol. 37, no. 1, pp. 343–353, Jan. 2022.
an MMC using novel mathematical framework,” in 2020 IEEE Power [74] Y. Liao, X. Wang, and X. Wang, “Frequency-domain participation
Energy Soc. Gen. Meeting (PESGM), 2020, pp. 1–5. analysis for electronic power systems,” IEEE Trans. Power Electron.,
[53] S. Almer and U. Jönsson, “Harmonic analysis of pulse-width modulated vol. 37, no. 3, pp. 2531–2537, Mar. 2022.
systems,” Automatica, vol. 45, no. 4, pp. 851–862, Apr. 2009. [75] S. Chandrasekaran, D. Borojevic, and D. Lindner, “Input filter interac-
[54] J. B. Kwon, X. Wang, F. Blaabjerg, C. L. Bak, A. R. Wood, and tion in three phase AC-DC converters,” in Proc. 30th Annu. IEEE Power
N. R. Watson, “Harmonic instability analysis of a single-phase grid- Electron. Specialists Conf. Rec., 1999, pp. 987–992.
connected converter using a harmonic state-space modeling method,” [76] Z. Liu, J. Liu, W. Bao, and Y. Zhao, “Infinity-norm of impedance-based
IEEE Trans. Ind. Appl., vol. 52, no. 5, pp. 4188–4200, Sep. 2016. stability criterion for three-phase AC distributed power systems with
[55] K. Gao, R. Xie, and G. Zhang, “Harmonic analysis of nonlinear periodic constant power loads,” IEEE Trans. Power Electron., vol. 30, no. 6,
network using the dynamic harmonic domain,” in Proc. 7th Int. Forum pp. 3030–3043, Jun. 2015.
Elect. Eng. Automat., 2020, pp. 616–622. [77] R. Rosso, M. Andresen, S. Engelken, and M. Liserre, “Analysis of
[56] D. Sun, X. Lu, and L. Du, “Optimal design of grid interactive inverters the interaction among power converters through their synchronization
based on harmonic state space modeling,” in Proc. IEEE Energy Con- mechanism,” IEEE Trans. Power Electron., vol. 34, no. 12, pp. 12321–
vers. Congr. Expo., 2020, pp. 5040–5045. 12332, Dec. 2019.
[57] J. Lyu, X. Zhang, X. Cai, and M. Molinas, “Harmonic state-space based [78] R. Ortega, A. Loría, P. J. Nicklasson, and H. Sira-Ramirez, Passivity-
small-signal impedance modeling of a modular multilevel converter Based Control of Euler-Lagrange Systems. Berlin, Germany: Springer,
with consideration of internal harmonic dynamics,” IEEE Trans. Power 1998.
Electron., vol. 34, no. 3, pp. 2134–2148, Mar. 2019. [79] X. Wang, F. Blaabjerg, and P. C. Loh, “Passivity-based stability analysis
[58] K. Gao, R. Xie, and G. Zhang, “Harmonic analysis of nonlinear periodic and damping injection for multiparalleled VSCs with LCL filters,” IEEE
network using the dynamic harmonic domain,” in Proc. 7th Int. Forum Trans. Power Electron., vol. 32, no. 11, pp. 8922–8935, Nov. 2017.
Elect. Eng. Automat., 2020, pp. 616–622. [80] Y.-C. Jeung, D.-C. Lee, T. Dragicevic, and F. Blaabjerg, “Design of
[59] J. Kwon, X. Wang, C. L. Bak, and F. Blaabjerg, “The modeling and har- passivity-based damping controller for suppressing power oscillations
monic coupling analysis of multiple-parallel connected inverter using in DC microgrids,” IEEE Trans. Power Electron., vol. 36, no. 4,
harmonic state space (HSS),” in Proc. IEEE Energy Convers. Congr. pp. 4016–4028, Apr. 2021.
Expo., 2015, pp. 6231–6238. [81] A. M. Lyapunov, “The general problem of the stability of motion,” Int.
[60] R. Mittal, L. Fan, and Z. Miao, “Harmonic state-space model of second- J. Control, vol. 55, no. 3, pp. 531–534, 1992. [Online]. Available: https:
order generalized integrator phase-locked loop,” in Proc. IEEE Power //doi.org/10.1080/00207179208934253
Energy Soc. Gen. Meeting, 2021, pp. 1–5. [82] H. D’Angelo, Linear Time-Varying Systems: Analysis and Synthesis,
[61] F. Cecati, R. Zhu, M. Liserre, and X. Wang, “Nonlinear mod- Ser. Allyn and Bacon Series in Electrical Engineering. Boston, MA,
ular state-space modeling of power-electronics-based power sys- USA: Allyn and Bacon, 1970.
tems,” IEEE Trans. Power Electron., vol. 37, no. 5, pp. 6102–6115, [83] P. Montagnier, R. J. Spiteri, and J. Angeles, “The control of linear
May 2022. time-periodic systems using Floquet–Lyapunov theory,” Int. J. Control,
[62] P. D. Rua, Özgür Can Sakinci, and J. Beerten, “Comparative study of vol. 77, no. 5, pp. 472–490, 2004.
dynamic phasor and harmonic state-space modeling for small-signal [84] N. R. Chaudhuri, R. Oliveira, and A. Yazdani, “Stability analysis
stability analysis,” Electric Power Syst. Res., vol. 189, Dec. 2020, of vector-controlled modular multilevel converters in linear time-
Art. no. 106626. periodic framework,” IEEE Trans. Power Electron., vol. 31, no. 7,
[63] E. Karami, G. B. Gharehpetian, and M. Madrigal, “A step forward pp. 5255–5269, Jul. 2016.
in application of dynamic harmonic domain: Phase shifting property [85] Y. Han et al., “Floquet-theory-based small-signal stability analysis of
of harmonics,” IEEE Trans. Power Del., vol. 32, no. 1, pp. 219–225, single-phase asymmetric multilevel inverters with SRF voltage control,”
Feb. 2017. IEEE Trans. Power Electron., vol. 35, no. 3, pp. 3221–3241, Mar. 2020.

466 VOLUME 3, 2022

[86] J. Zhu, J. Hu, S. Wang, and M. Wan, “Small-signal modeling and NADIA MEI LIN TAN (Senior Member, IEEE)
analysis of MMC under unbalanced grid conditions based on linear was born in Kuala Lumpur, Malaysia. She received
time-periodic (LTP) method,” IEEE Trans. Power Del., vol. 36, no. 1, the B.Eng. (Hons.) degree in electrical engineer-
pp. 205–214, Feb. 2021. ing from the University of Sheffield, Sheffield,
[87] J. Zhu, Z. Guo, J. Hu, S. Ma, and J. Guo, “Truncation number selec- U.K., in 2002, the M. Eng. degree in electrical
tion of harmonic state-space model based on the Floquet characteristic engineering from National Energy University, Ka-
exponent,” IEEE Trans. Ind. Electron., early access, May 11, 2022, jang, Malaysia, in 2007, and the Ph.D. degree in
doi: 10.1109/TIE.2022.3172780. electrical engineering from the Tokyo Institute of
[88] H. Wu and X. Wang, “Transient angle stability analysis of grid- Technology, Tokyo, Japan, in 2010, . She is cur-
connected converters with the first-order active power loop,” in Proc. rently an Associate Professor with the Department
IEEE Appl. Power Electron. Conf. Expo., 2018, pp. 3011–3016. of Electrical and Electronics Engineering, Univer-
[89] U. Markovic, O. Stanojev, P. Aristidou, E. Vrettos, D. Callaway, and sity of Nottingham Ningbo China. Since 2017, she has been an Associate
G. Hug, “Understanding small-signal stability of low-inertia systems,” Professor with the Department of Electrical and Electronics Engineering,
IEEE Trans. Power Syst., vol. 36, no. 5, pp. 3997–4017, Sep. 2021. Universiti Tenaga Nasional, Kajang, Malaysia. From October 2016 to March
[90] Z.-X. Zou, G. Buticchi, and M. Liserre, “Analysis and stabilization of a 2017, she was a Visiting Researcher with Power Electronics Laboratory,
smart transformer-fed grid,” IEEE Trans. Ind. Electron., vol. 65, no. 2, Tokyo Institute of Technology. From July to September 2016, she was an
pp. 1325–1335, Feb. 2018. Adjunct Associate Professor with the Tokyo Institute of Technology. She is a
[91] R. Teodorescu, M. Liserre, and P. Rodriguez, Grid Converters for Pho- Chartered Engineer registered with Engineering Council, U.K., a Professional
tovoltaic and Wind Power Systems. Hoboken, NJ, USA: Wiley, 2011. Engineer registered with the Board of Engineers Malaysia, a Member of the
[92] D. Lu, X. Wang, and F. Blaabjerg, “Impedance-based analysis of DC- Institution of Engineering and Technology, and a Member of the Institution of
link voltage dynamics in voltage-source converters,” IEEE Trans. Power Engineers Malaysia. Her research interests include power conversion systems
Electron., vol. 34, no. 4, pp. 3973–3985, Apr. 2019. for renewable energy and battery energy storage applications, and bidirec-
[93] J. Kwon, X. Wang, F. Blaabjerg, C. L. Bak, V.-S. Sularea, and C. Busca, tional isolated DC-DC converters. She was the recipient of the 2018 IET
“Harmonic interaction analysis in a grid-connected converter using Young Engineer Award and the 2018 IPCC IEEE IAS Prize Paper Award.
harmonic state-space (HSS) modeling,” IEEE Trans. Power Electron., From 2018 to 2019 she was the Vice-Chair of the IEEE Power Electronics
vol. 32, no. 9, pp. 6823–6835, Sep. 2017. Society Malaysia Chapter.
[94] D. G. Holmes and T. A. Lipo, Pulse Width Modulation for Power
Converters. IEEE, Hoboken, NJ, USA: Wiley, 2003. SANDRO GUENTER (Member, IEEE) received
[95] X. Yue, X. Wang, and F. Blaabjerg, “Review of small-signal modeling the M.Sc. degree in electrical engineering from the
methods including frequency-coupling dynamics of power converters,” University of Duisburg-Essen, Duisburg, Germany
IEEE Trans. Power Electron., vol. 34, no. 4, pp. 3313–3328, Apr. 2019. in 2011 and the Ph.D. degree from the Univer-
sity of Kiel, Kiel, Germany, in 2017. From 2012
to 2017, he was a Post Graduate Researcher with
QILIN PENG received the B.S. degree in elec- the University of Kiel. After working in the PV
trical engineering from the Harbin Institute of inverter industry, he joined The University of Not-
Technology, Weihai, China, in 2020. He was a tingham Ningbo China, Ningbo, China, as a Senior
Visiting Student with the Center for Power Elec- Research Fellow, in 2021. His research interests in-
tronics Systems, Blacksburg, VA, USA, in 2019. clude grid connected multilevel converters as well
He is currently working toward the Ph.D. degree as isolated DC/DC converters.
with Key Laboratory of More Electric Aircraft
Technology of Zhejiang Province, University of JIAJUN YANG (Member, IEEE) received the
Nottingham Ningbo China, Ningbo, China. He was B.Eng. (Hons.) and the Ph.D. degrees in electronic
with VOLVO Construction Equipment Jinan Tech- and electrical engineering from the University of
nology Center, Jinan, China. His research interests Nottingham Ningbo China, Ningbo, China, in 2017
include grid-connected converters and stability analysis of AC microgrids for and 2022, respectively. From 2017 to 2018, he was
more electric aircraft. a Hardware Engineer in power electronics with
Nottingham Electrification Centre, Ningbo, China,
the University of Nottingham. He is currently a
Research Fellow in power electronics for intelli-
GIAMPAOLO BUTICCHI (Senior Member, gent manufacturing with China Beacons Institute,
IEEE) received the master’s and Ph.D. degrees Ningbo, China. His current research interests in-
in electronic engineering and information tech- clude stabilization, optimization, and protection of power electronics based
nologies, from the University of Parma, Parma, microgrids.
Italy, in 2009 and 2013, respectively. In 2012, he
was a Visiting Researcher with The University PATRICK WHEELER (Fellow, IEEE) received the
of Nottingham, Nottingham, U.K. Between 2014 B.Eng. (Hons) degree and the Ph.D. degree in elec-
and 2017, he was a Postdoctoral Researcher and a trical engineering for his work on matrix converters
Guest Professor with the University of Kiel, Kiel, from the University of Bristol, Bristol, U.K., in
Germany. In 2017, he was an Associate Professor 1990 and 1994, respectively. In 1993, he moved to
of electrical engineering with The University of the University of Nottingham, Ningbo, China, as a
Nottingham Ningbo China, Ningbo, China and the Head of power electronics Research Assistant with the Department of Elec-
of the Nottingham Electrification Center, where he was a Professor in 2020. trical and Electronic Engineering. In 1996, he was
He has authored or coauthored more than 230 scientific papers. His research a Lecturer with Power Electronics, Machines and
interests include power electronics for renewable energy systems, smart trans- Control Group, University of Nottingham, where
former fed micro-grids, and DC grids for the more electric aircraft. He is he has been a Full Professor since January 2008.
one of the best advocates for DC distribution systems and multi-port power From 2015 to 2018, he was the Head of Department of electrical and elec-
electronics onboard the future aircraft. He was an Associate Editor for the tronic engineering with the University of Nottingham. He has authored or
IEEE TRANSACTIONS ON INDUSTRIAL ELECTRONICS, THE IEEE TRANSAC- coauthored more than 750 academic publications in leading international
TIONS ON TRANSPORTATION ELECTRIFICATION, and the IEEE OPEN JOURNAL conferences and journals. He is currently the Head of the Power Electronics,
OF THE INDUSTRIAL ELECTRONICS SOCIETY. He is currently the Chair of the Machines, and Control Research Group, the Global Director of the Institute
IEEE-IES Technical Committee on Renewable Energy SystemsAND THEIES of Aerosapce Technology, University of Nottinghamas and was the Li Dak
Energy Cluster Delegate. Sum Chair Professor of electrical and aerospace engineering. He is a Member
He was the recipient of the Von Humboldt Postdoctoral Fellowship to carry of the PELs AdCom and is currently the Vice-President of the IEEE Power
out research related to fault tolerant topologies of smart transformers. Electronics Society.

VOLUME 3, 2022 467