A Simplified D-Q Small-Signal Modelling method of

2020 IEEE 9th International Power Electronics and Motion Control Conference (IPEMC2020-ECCE Asia) | 978-1-7281-5301-8/20/$31.00 ©2020 IEEE | DOI: 10.1109/IPEMC-ECCEAsia48364.2020.9367750

Grid-Tied Inverters for Stability Analysis

Wei Wang Zhijie Zeng Zheng Wei Huafeng Xiao
NARI Group Corporation School of Electrical Engineering NARI Group Corporation School of Electrical Engineering
State Grid Southeast University State Grid Southeast University
Nanjing, China Nanjing, China Nanjing, China Nanjing, China
[email protected] [email protected] [email protected] [email protected]

Abstract—Harmonic resonance caused by interactions The dq impedance model of three-phase GTIs can be
between grid-tied inverters (GTIs) and weak grid are becoming established as control strategy is usually conducted in dq frame
increasingly prominent. Small-signal modeling of GTIs is the to regulate active and reactive power [3]. Based on the small-
most fundamental work in to analyze the harmonic stability of signal model of dynamic error in phase-locked loop (PLL) [4],
GTI-weak grid coupling system. Different from generic small- Wen at al. established a complete dq impedance model of GTIs
signal modeling of GTIs in dq coordinate, a simplified impedance in [5], and the effects of PLL and control loops on impedance
modeling method has been proposed in this paper after focusing were analyzed by comparing impedance curves. This modeling
interesting frequency band onto low-medium frequency range. In method was further developed to discuss the influence of grid
the simplified modeling method, a partial admittance is
impedance and filter parameters on stability [6]. As the three-
introduced, and coupling terms between d and q channels are
phase inverter is presented as a coupled dual-input and dual-
identified as negligible parts and non-negligible parts,
respectively. As a result, a simplified small-signal model of GTI output system in dq coordinate, matrix calculations weakened
has been obtained, and connections among terms of GTI’s connections between impedance terms and physical parts of
impedance and its physical parts are highlighted to strengthen GTIs, and thus it is difficult to find out the interaction
physical interpretations of dq impedances. In addition, concepts mechanism among GTIs’ parameters and coupling system
of active power admittance and reactive power admittance are stability. In order to avoid multiplex matrix operations, a
introduced in this paper to demonstrate the influence mechanism simplest dq impedance model was proposed by ignoring all
of power-level’s variations on GTI’s impedance. Finally, the coupling terms of impedance matrix [7], but its application
theoretical analysis in frequency domain and simulation results range is limited as it can only be used for stability analysis of
in time domain are provided to prove that the proposed unity power factor condition. Furthermore, a small-signal
simplified modelling method can be used to analyze the model of the GTI with divided d-channel and q-channel was
resonance mechanism and predict the stability of GTI-weak grid proposed to improve stability analysis without matrix
coupling system. calculation in [8], but the connections among impedance terms
and the physical parts of GTIs had not been built yet.
Keywords—dq impedance, admittance, harmonic resonance,
simplified modeling, coupling effects； In order to build the connections discussed above, a
simplified small-signal model with enhanced physical
I. INTRODUCTION interpretation has been proposed in this paper by reducing
matrix calculations and ignoring part of coupling terms during
With the increase of the number and capacity proportion of
the modeling process. Subsequently, by combining the
distributed renewable energy generations, the distribution
simplified inverter impedance with the grid impedance, the
network behaves with more and more obvious characteristics
harmonic resonance mechanism and stability of GTI-weak grid
of weak grid [1]. As interface converters between new energy
coupling system with full power factor range can be effectively
generation devices and the power system, grid-connected
analyzed.
inverters (GTIs) inevitably interacts with the weak grid, which
may excite broadband resonance or even lead to harmonic I. COMPLTED SAMLL-SIGNAL MODEL
stability issues [2].
The power circuit and typical control strategy of three-phase
Generally, the modeling methods for harmonic stability of GTIs for distributed generation are shown in Fig. 1 while Tab.
GTI-weak grid coupling system include state-space approach 1 presents associated parameters. As most widely adopted
and impedance-based approach. By the state-space approach, scheme, the current controller (CC) is implemented in the dq-
the stability is analyzed through the eigenvalue of state matrix frame. Divided by the point of common coupling (PCC), Zi is
with detailed internal parameters obtained [3], which is the impedance of the inverter-side while Zg is that of the grid-
commonly used in bulk power systems without numerous side, where the impedances of transmission cable and
generators and variant parameters. The impedance-based transformer are lumped together as Lg and Rg.
approach is more practical to perform the stability assessment
by extracting terminal characteristics of power electronic The impedance or admittance of three-phase GTIs is
converters in frequency domain [4]. basically a breed of small-signal linearization models for
nonlinear system. Given that Δupdq denotes the small-signal

XXX-X-XXXX-XXXX-X/XX/$XX.00 ©20XX IEEE 2893

Authorized licensed use limited to: North China Electric Power University. Downloaded on May 05,2023 at 01:17:50 UTC from IEEE Xplore. Restrictions apply.
voltage at point of common coupling (PCC), and Δigdq is the Accordingly, Δupdq and Δigdq are excitation and response to
small-signal grid-in current, the completed schematic diagram each other. Supposed that voltage is excitation signal and
of the GTI is illustrated in Fig. 2. Since the space vector current is response signal, the inverter admittance Yi can be
modulation is implemented with constant DC voltage, KPWM is obtained as
a second-order identity matrix. Transfer function matrices
determined by the filter circuit are GL and YL. The current Yi (s) = ( E2 − GL KPWMGdelGcGf )−1
controller matrix is Gc while Gdel is the transfer function matrix (1)
of control delay [7]. Gf is the transfer function matrix of the ⋅ (YL + GL KPWMGdelGf (GPLL
d i
− GcGPLL ))
low-pass filter in the sampling. GiPLL and GdPLL models the where E2 is a second-order identity matrix. On the contrary,
small-signal perturbation path from PCC voltage to controller when the current is considered as excitation signal and the
voltage and current in the dq frame, respectively [7]. voltage is response signal, the inverter impedance Zi is derived
as
Lf rf Zi Zg Lg rg ug
Vdc PCC Zi (s) = Yi −1 (s)
vabc iabc = (YL + GL KPWMGdelGf (GPLL
d i
− GcGPLL ))-1 (2)
SVPWM Sampling Sampling ⋅ ( E2 − GL KPWMGdelGcGf )
αβ dq SRF-PLL abc dq It is revealed that the establishment of the complete
θ PLL iq id impedance model requires lots of matrix operations including
- idref coupling terms and matrix inversion, which leads to fuzzy
PI + connection among the impedance terms and actual physical
-+ iqref parts. As a result, the influence of system parameters and
PI
power level on inverter impedance can only be summarize after
Fig.1 Three-phase inverter-grid connected system
comparison of series of impedance curves. To eliminate the
deficiency discussed, a simplified small-signal modeling
TABLE I. MAIN PARAMETERS OF SYSTEM method is proposed in this paper as the comparison between
modeling processes of the simplified model and the complete
Symbol Description Value model is shown in Fig. 3.
Vdc Inverter dc voltage 700V
Completed model
Pr Rated power 500kW Yi Zi
Inverse matrix
Lf Filter inductance 155.4μH
rf Parasitic resistance 0.01 Ω Y1 Z1 Z1dd Zidd
Grid line-line peak voltage 443V
sim.
Vg
f0 Grid voltage frequency 50 Hz
YPLL-d Zidq
fs Switching frequency 3.2 kHz
Kp1 Proportional gain of CC 0.64 Y2 Y2s sim.
sim.
KI1 Integrator gain of CC 100 YPLL-q Ziqq
Simplified model
Kp2 Proportional gain of PLL 1.7

KI2 Integrator gain of PLL 490 Fig.3 Comparison of modeling processes

Δigdq II. SIMPLIFIED SAMLL-SIGNAL MODEL

Gdel(s) KPWM GL(s)
+
Filter A. Modeling of Partial Admittance
-1 As illustrated in Fig. 3, the admittance model of inverter Yi is
Gf (s) YL(s)
Δupdq divided to partial admittance Y1 and Y2 firstly during the
Gf (s) modeling process of simplified small-signal model. Supposed
that the influence of PLL dynamic error is ignored
GdPLL(s) PLL GiPLL(s) + temporarily, the small-signal schematic diagram associated
with partial admittance Y1 is shown in Fig. 2(a). Accordingly,
the matrix expression of Y1 is
+ - Δidq-ref
Gc(s)
Y1 = ( E2 − GL K PWM GdelGcGf )−1 ⋅ YL (3)
Fig.2 Completed small-signal model of GTI

2894

Authorized licensed use limited to: North China Electric Power University. Downloaded on May 05,2023 at 01:17:50 UTC from IEEE Xplore. Restrictions apply.
 Due to the existence of PLL dynamic error, dq coordinates The expression depicting the relationship among the small-
of actual control-system is not in accordance with that of ideal signal voltage, small-signal current and partial admittances is
control-system [6]. Subsequently, PCC voltage harmonics will obtained as
generate grid-in current harmonics through PLL and control
loops, which is illustrated by the small-signal schematic Δigdq = Y1 (s) ⋅ Δupdq + Y2 (s) ⋅ Δupdq (9)
diagram in Fig. 2(b). Relevant matrixes are listed as formula
(4) to (6). Since the first columns of the matrices in (4) and (5) are all
zero, the first column of the admittance matrix Y2 is also zero,
Δigdq then (9) is rewritten as
KPWM GL(s)
+
 Δigd  Y1dd Y1dq   Δupd  0 YPLL − d   Δupd 
-1  =  +   (10)
Δupdq
YL(s)  Δigq  Y1qd Y1qq   Δupq  0 YPLL − q   Δupq 
Gf (s)
Consequently, Y1 and Y2 can be conceived as virtual parallel
- Δidq-ref admittances of GTIs. In the admittance Y2 introduced by PLL
Gdel(s) Gc(s)
dynamic error, YPLL-d and YPLL-q represent the influence of PCC
(a) harmonic voltage on d-channel component and q-channel
Δigdq component of grid-in current, respectively. When the GTI
KPWM GL(s) operates in different power levels, the bode plots of YPLL-d and
YPLL-q in frequency domain are shown in Fig. 5. It is worth
Δupdq PLL -1 noting that the admittance in (10) is calculated by matrixes of
Gf (s)
the complete small-signal model with associated main
Gf (s) parameters shown in Tab. 1.

+ The admittance curve of GTIs in unit power factor case is

GdPLL(s) GiPLL(s) shown in Fig. 5(a). The amplitude of YPLL-d is lower than 0dB
at most frequencies, so the PLL dynamic error will not cause
+ - Δidq-ref the harmonic of d-channel current increasing significantly. On
Gdel(s) Gc(s)
the contrary, the amplitude of the YPLL-q is much higher than
(b) 0dB in the low-medium frequency range. With the increase of
Fig.4 Small-signal schematic diagram of inverter
(a) partial admittance Y1; (b) partial admittance Y2
power level, the amplitude of YPLL-q increases, which indicates
that harmonic resonance will be excited in q-channel. In
contrast, the admittance curve of GTIs in non-unity power
i 0 I q GPLL  factor case is shown in Fig. 5(b). As the reactive power lever
GPLL =  (4)
0 − I d GPLL  increases, YPLL-q is invariant while the amplitude of YPLL-d
increases, which indicates that harmonic resonance will also be
d  0 −Viq GPLL  excited in d-channel.
GPLL =  (5)
 0 Vid GPLL  Since low-pass filtering and control delay only affect the
tf PLL high-frequency characteristics of GTIs [12], the matrix
GPLL = (6) expression of Y2 can be simplified as (11) when investigating
s + Vpd ⋅ tfPLL the connection between low-medium frequency characteristics
and internal physical parts.
Idq is the fundamental grid-in current’s amplitude in the
given power level, Vidq is the fundamental component’s Y2 s = ( E2 − GLGc )−1 ⋅ GL ⋅ (GPLL
d i
− GcGPLL ) (11)
amplitude of modulation wave in SVPWM, Vpd is the
fundamental voltage amplitude of PCC; tfPLL is the transfer When the coupling terms of the filter inductor admittance
function of PI controller in the PLL. According to Fig. 4(b), the GL is ignored, an approximate expression of simplified partial
transfer function matrix of the second partial admittance Y2 is admittance Y2s can be obtained as
obtained as
1
YPLL − d ≈ (Viq + I q ⋅ tfCC )GPLL (12)
Y2 = ( E2 − GL K PWMGdelGcGf )−1 tfCC + Ls
(7)
⋅ (GL K PWMGdelGf (GPLL
d i
− GcGPLL ))
−1
Denote the transfer functions of the current controller as tfCC, YPLL− q ≈ (Vid + I d ⋅ tfCC )GPLL (13)
tfCC + Ls
then the current controller matrix GC is
As revealed by (12) and (13), YPLL-d and YPLL-q introduced
tf 0  by the dynamic error of PLL are not only associated with the
GC =  CC (8)
 0 tfCC  parameters of PLL, but also affected by current controller and
the power level. Given that the inverter is in steady operation
and the controller parameters remain unchanged, there is a

2895

Authorized licensed use limited to: North China Electric Power University. Downloaded on May 05,2023 at 01:17:50 UTC from IEEE Xplore. Restrictions apply.
positive correlation between the amplitude of YPLL-d and the B. Modeling of Simplified Impedance
current value of reactive power. Simultaneously, there exists a Based on the partial admittance Y1, the active power
positive correlation between the amplitude of YPLL-q and the admittance YPLL-q and the reactive power admittance YPLL-d
current value of active power. Accordingly, YPLL-d and YPLL-q derived previously, expressions of independent terms in the
can be defined as reactive power admittance and active inverter impedance matrix will be derived to clarify the
power admittance, respectively. It is worth mentioning that connection among them and physical parts. Formula (10) can
the phase of active power admittance is 180° during low be rewritten as
frequency range, which behaves as a negative resistance [6].
Δigd − YPLL−d Δupq  Y1dd Y1dq  Δupd 
0
YPLL-d YPLL-q  =   (14)
 Δigq − YPLL−q Δupq  Y1qd Y1qq  Δupq 
Mag(dB)

-40 Id=100; Iq=0

Id=400; Iq=0 Define the impedance matrix Z1 as the inverse matrix of Y1,
-80
Id=800; Iq=0 (14) is converted into
360
 Δupd   Z1dd Z1dq   Δigd − YPLL − d Δupq 
Pha(deg)

 =   (15)
 Δupq   Z1qd Z1qq   Δigq − YPLL − q Δupq 
180

0 Generally, the mutual-impedance (Z1dq, Z1qd) is neglectable

100 101 102 103 100 10 1 102 103
Frequency(Hz) compared with self-impedances (Z1dd, Z1qq), and thus (15) is
simplified as follows:
(a)
20  Δupd   Z1dd 0   Δigd − YPLL − d Δupq 
YPLL-d YPLL-q  ≈ (16)
 Δupq   0 Z1qq   Δigq − YPLL − q Δupq 
Mag(dB)

-20
The impedance-based voltage-current relationship can be
-40 rewritten as
360
 − Z1dd Z1qqYPLL − d 
Pha(deg)

Id=800 ; Iq=100  Z1dd 

180  Δupd   1 + Z1qqYPLL − q   Δigd 
Id=800 ; Iq=300 (17)
 =   Δigq 
0
Id=800 ; Iq=500
 Δupq   0 Z1qq
 
100 101 102 103 100 101 102 103
 1 + Z1qqYPLL − q 
Frequency(Hz)
(b) where Ziqd is approximated as zero while Z1dd is equal to Z1qq.
Fig. 5 Bode plot of partial admittance Y2 The simplified expressions can be obtained from (16) and (17).
(a) unity power factor; (b) non-unity power factor
Z1 = YL−1 ⋅ ( E2 − GL K PWMGdelGcGf )
To verify the feasibility of the simplified admittance (18)
modeling method, bode curves of the complete admittance and ≈ Gc + YL−1 = Gc + Z L
the simplified admittance are compared in the condition of
non-unit power factor (Id=800A, Iq=500A) as shown in Fig. 6. Z1dd = Z1qq ≈ tfCC + ZL (19)
It is clarified obviously by this example that the deriving of
simplified partial admittance YPLL-d and YPLL-q is reasonable and Since the variation trend of amplitude-frequency
effective. characteristic of PI controller is opposite to that of inductance
impedance, the low-frequency characteristic of Z1dd is mainly
20 YPLL-d YPLL-q determined by PI controller, while the high-frequency
Mag(dB)

0 characteristic is mainly determined by the inductance. The final

-20 simplified inverter impedance Zi is derived in the following
completed after substituting (19) into (17).
-40 simplified
360  −Z1dd 2YPLL − d 
Z 
Zidq   1dd
Pha(deg)

 Zidd 1 + Z1dd YPLL − q 

180 Zi =  = (20)
 Ziqd Ziqq   Z1dd 
 0 
0
 1 + Z1dd YPLL − q 
100 101 102 103 100 101 102 103
Frequency(Hz)
Based on formula (2) of completed model [7], only the
Fig. 6 Comparison of partial admittance final result of the inverter impedance matrix can be obtained.
Nevertheless, equation (12), (13) and (20) clearly reveals the

2896

Authorized licensed use limited to: North China Electric Power University. Downloaded on May 05,2023 at 01:17:50 UTC from IEEE Xplore. Restrictions apply.
connections among inverter impedance and physical parts GTIs, the resonance mechanism and stability assessment of the
including CC, PLL and filter. In the modeling process of GTI-weak grid coupling system is discussed in this section.
simplified impedance, coupling terms in the partial impedance
or partial admittance can be divided into two parts. It must be A. Analysis of Resonance Mechanism
retained if coupling terms are introduced by the PLL, while the In order to investigate the low-medium frequency harmonic
coupling terms introduced by other segments can be ignored. resonance mechanism of GTI-weak grid connected system, (20)
is substituted into (14) and further divided into expressions of
According to (12), (13) and Fig. 3, the influence of the d-channel and q- channel.
power level on inverter impedance can be directly summarized.
When the power level changes, the variations of active power Δupd = Z1dd (Δigd − YPLL − d Δupq ) (22)
admittance and reactive power admittance yield changes of the
inverter impedances Zidq and Ziqq while Zidd and Ziqd are Δupq = Z1qq (Δigq − Δupq / ZPLL − q ) (23)
independent with the power level. Specifically, only YPLL-d and
Zidq will be changed if the active power level is remained while Meanwhile, voltage-current expressions at PCC depicted by
the reactive power level is modified, which agrees with the the grid-side impedance are written as follows:
conclusion drawn from impedance comparison in [7].
In low-medium frequency range, Z1dd YPLL-q>>1, Zidq and
Δupd = −(Z gdd Δigd + Z gdq Δigq ) (24)
Ziqq can be further simplified into formula (21). YPLL-q is
negative admittance, so it can be concluded directly that the Δupq = −( Z gqd Δigd + Z gqq Δigq ) (25)
negative resistance characteristic of Ziqq is introduced by the
active power admittance. As illustrated in Fig. 8(a), the d-channel small-signal
harmonic equivalent circuit is obtained by combining (22) and
− Z1dd YPLL − d 1 (24), while the q-channel small signal harmonic equivalent
Zidq ≈ ; Ziqq ≈ = ZPLL − q (21) circuit is obtained by combining (23) and (25) and shown in
YPLL − q YPLL − q
Fig. 8(b). The voltage source us is the harmonic voltage of the
Taking the case of non-unity power factor (Id=800A, inverter-side, and the voltage source ug is the harmonic voltage
Iq=500A) as an example, the comparison between curves of the of the grid-side.
complete impedance model and the simplified impedance
model is shown in Fig. 7. It is revealed that only in high- Zi Zg
frequency range, there are differences between two groups of igd + -
impedance curves, which indicates that the simplified inverter
impedance model can well depict the terminal characteristics Zgdq (s)igq
of GTIs overall. Z1dd (s) Zgdd(s)
upd +
20 YPLL-d (s)upq
Pha(deg) Mag(dB)

usd ugd
0 +
-20
180 (a)
completed
90 simplified Zi Zg
0 igq
Zidd Zidq + -
-90
0 Ziqd Ziqq Zgqd (s)igd
Pha(deg) Mag(dB)

Z1qq(s) Zgqq(s)
-20
upq
ZPLL-q (s)
+
-40
usq ugq
90 +
0
-90
(b)
-180 0 Fig.8 Small-signal harmonic equivalent circuit of inverter-weak grid coupling
10 101 102 103 100 101 102 103
Frequency(Hz) system: (a) d channel; (b) q channel

Fig.7 Inverter impedance of completed model and simplified model As previously discussed, parallel impedance introduced by
PLL in q-channel behaves as a negative resistor in low-medium
II. APPLICATION OF SIMPLIFIED MODEL frequency range, leading the whole system to operate with
Impedance of weak grid is non-neglectable, and its negative damping characteristics during low-medium
interaction with inverter impedance will easily yield harmonic frequency resonance. Due to the existence of grid impedance,
stability issues in inverter-grid connected system [1]. Based on there is coupling between PCC voltage and the grid-in current.
the grid impedance and the simplified impedance model of Meanwhile, there is also coupling between d-channel and q-
channel in the system due to the equivalently controlled current

2897

Authorized licensed use limited to: North China Electric Power University. Downloaded on May 05,2023 at 01:17:50 UTC from IEEE Xplore. Restrictions apply.
source as depicted in Fig. 8. Once the low-medium frequency plane is determined by the low-medium frequency impedance
harmonic current is generated in q-channel, the harmonic of inverters and the weak grid. Whether the inverter operates
current in the d-channel will also oscillate due to the coupling with unit power factor or non-unit power factor, the simplified
effect discovered. Consequently, the harmonic resonance of the inverter impedance can be used for GNC to accurately predict
GTI-weak grid coupling system is obviously excited in both the stability of the GTI-weak grid coupling system.
channels.
3
B. Analysis of System Stability λ1 λ2
2
Nyquist stability criterion has been widely used in

Imaginary Axis
1
impedance-based stability analysis of single-input and single-
output system. However, The three-phase AC system is 0
featured as coupled dual-input dual-output, then the
-1
generalized Nyquist criterion (GNC) was introduced [2]. The
return ratio matrix is defined as: -2 completed
simplified
-3
L = Z g ⋅ Zi−1 (27) -1 0 1 2 3 -1 0 1 2 3
Real Axis
The assessment of system stability can be performed
Fig.10 Characteristic loci of GNC (non-unity power factor)
according to characteristic loci of the return ratio matrix [7].
This section combines the simplified dq impedance model with III. VERIFICATION IN SIMULATIONS
GNC to prove that the simplified modeling method can be used
for stability analysis. Different power levels and grid In order to verify theoretical analysis above, a simulation
impedances are designed, then the characteristic loci based on model was built in Simulink according to the structure and
the complete inverter impedance model and the simplified parameters shown in Fig. 1 and Tab.1. Furthermore, the
inverter impedance model are compared. When the grid feasibility of the simplified inverter impedance model for
impedance changes, the equivalent resistance component Rg stability analysis is verified by changing the grid-in power
will be remained as constant (0.02Ω), only the equivalent level and the grid impedance.
inductance component Lg is changed because the existence of When the GTI operates at unity power factor (Id=800A),
Lg is the main reason of harmonic stability issues. curves of the characteristic loci with different grid impedances
When the GTI is operating at unit power factor (Id=800A) are shown in Fig. 11. It is revealed that the curves with
and the equivalent inductance of the grid impedance is 350uH, Lg=350μH do not encircle the critical point (-1,0). After the
the characteristic loci curves obtained according to the equivalent inductor of the grid impedance increases to 500μH,
complete impedance model and the simplified impedance one of the characteristic loci curves encircles the critical point,
model are shown in Fig. 9. In the left half plane, when the which indicates that the system is unstable. The simulation
characteristic loci curves are close to the critical stable point (- waveforms in time domain are in agreement with the
1,0), two groups of curves coincide with each other. theoretical assessment as shown in Fig. 12.
Consequently, the stability assessment results obtained by When the GTI operates at non-unity power factor
completed impedance and simplified impedance are consistent. (Id=800A, Iq=500A), the results presented by frequency domain
Although there are some differences in the right half plane, it is and time domain are shown in Fig. 13 and Fig. 14,
independent with the result of stability assessment. respectively, the assessment from theoretical characteristic loci
3 curves in frequency domain also match the simulation results
λ1 λ2 in time domain, which further proves that the simplified
2
impedance model can be used to predict the stability of the
Imaginary Axis

1 GTI-weak grid coupling system.

0
3 λ1 λ2
-1 2
Imaginary Axis

-2 completed 1
simplified 0
-3
-1 0 1 2 3 -1 0 1 2 3
-1
Real Axis
-2
Fig.9 Characteristic loci of GNC (unity power factor) Lg=350μH(stable)
-3 Lg=500μH(unstable)
When the GTI operates at non-unity power factor (Id=800A, -1 0 1 2 3 4 -1 0 1 2 3 4
Real Axis
Iq=500A) and the equivalent inductance of the grid impedance
is still 350uH, characteristic loci in two groups are also well Fig.11 Characteristic loci of GNC (unity power factor)
coincident in the left half plane as shown in Fig. 10, which
indicates the results of stability assessment are consistent. As a
conclusion, the partial characteristic loci located in the left half

2898

Authorized licensed use limited to: North China Electric Power University. Downloaded on May 05,2023 at 01:17:50 UTC from IEEE Xplore. Restrictions apply.
 id Stable Unstable frequency range. By the simplified modeling method, the non-
iq negligibility of coupling terms introduced by PLL is
1000
Current (A)
emphasized while coupling terms introduced by other
segments is neglected. According to the simplified modeling
0
results, two equivalent parallel admittances are proposed to
-1000
reflect the influence of PLL’s dynamic errors on inverters’
performance. The parallel admittance YPLL-d in d-channel is
Lg : 350μH→500μH positively correlated with reactive power current, while the
iabc parallel admittance YPLL-q of q-channel is positively correlated
1000 with active power current. Furthermore, the characteristic of
Current (A)

negative resistance in YPLL-q increases the risk of harmonic

0 resonances. Finally, derived impedance model can reveals the
connections among impedance terms, CC, PLL, the filter and
-1000
the power level to strengthen physical interpretation of dq
impedance.
0 0.05 0.1 0.15 0.2 0.25
Time(s) REFERENCE
[1] J. H. R. Enslin, and P. J. M. Heskes, “Harmonic interaction between a
Fig.12 Waveforms of inverter output current(unity power factor) large number of distributed power inverters and the distribution
network,” IEEE Trans. Power Electron., vol. 19, no. 6, pp. 1586-1593,
Nov. 2004.
3 λ1 λ2 [2] X. Wang, and F. Blaabjerg, “Harmonic Stability in Power Electronic-
2 Based Power Systems: Concept, Modeling, and Analysis,” IEEE Trans.
Imaginary Axis

Smart Grid, vol. 10, no. 3, pp. 2858-2870, May 2019.

1
[3] Y. Wang, X. Wang, F. Blaabjerg, and Z. Chen, “Harmonic Instability
0 Assessment Using State-Space Modeling and Participation Analysis in
-1 Inverter-Fed Power Systems,” IEEE Trans. Ind. Electron., vol. 64, no. 1,
pp. 806-816, Jan. 2017.
-2
Lg=350μH(stable) [4] J. Sun, “Impedance-Based Stability Criterion for Grid-Connected
-3 Lg=500μH(unstable) Inverters,” IEEE Trans. Power Electron., vol. 26, no. 11, pp. 3075-3078,
Nov. 2011.
-1 0 1 2 3 4 -1 0 1 2 3 4
Real Axis [5] S. Hiti, V. Vlatkovic, D. Borojevic, and F. C. Y. Lee, “A new control
algorithm for three-phase PWM buck rectifier with input displacement
factor compensation,” IEEE Trans. Power Electron., vol. 9, no. 2, pp.
Fig.13 Characteristic loci of GNC (non-unity power factor) 173-180, Mar. 1994.
[6] L. Harnefors, M. Bongiorno, and S. Lundberg, “Input-Admittance
id Stable Unstable Calculation and Shaping for Controlled Voltage-Source Converters,”
iq IEEE Trans. Ind. Electron., vol. 54, no. 6, pp. 3323-3334, Aug. 2007.
1000
Current (A)

[7] B. Wen, D. Boroyevich, R. Burgos, P. Mattavelli, and Z. Shen,

“Analysis of D-Q Small-Signal Impedance of Grid-Tied Inverters,”
0 IEEE Trans. Power Electron., vol. 31, no. 1, pp. 675-687, Jan. 2016.
[8] L. Yang, Y. Chen, A. Luo, Z. Chen, L. Zhou, X. Zhou, W. Wu, W. Tan,
-1000
and J. M. Guerrero, “Effect of phase-locked loop on small-signal
Lg : 350μH→500μH perturbation modelling and stability analysis for three-phase LCL-type
inverter connected to weak grid,” IET Renew. Power Gener., vol. 13, no.
iabc
1000 1, pp. 86-93, Jan. 2019.
Current (A)

[9] T. Messo, J. Jokipii, A. Mäkinen, and T. Suntio, "Modeling the grid

synchronization induced negative-resistor-like behavior in the output
0 impedance of a three-phase photovoltaic inverter." in Proc. 4th IEEE
Int.Symp. Power Electron. Distrib. Generation, Jul. 2013, pp. 1–7.
-1000 [10] Z. Shuai, Y. Li, W. Wu, C. Tu, A. Luo, and Z. J. Shen, “Divided DQ
Small-Signal Model: A New Perspective for the Stability Analysis of
0 0.05 0.1 0.15 0.2 0.25
Three-Phase Grid-Tied Inverters,” IEEE Trans. Ind. Electron., vol. 66,
no. 8, pp. 6493-6504, Aug. 2019.
Time(s)
[11] B. Wen, D. Boroyevich, P. Mattavelli, Z. Shen, and R. Burgos,
"Influence of phase-locked loop on input admittance of three-phase
Fig.14 Waveforms of inverter output current(non-unity power factor)
voltage-source converters." presented at the CPES Power Electronics
Conf., Blacksburg, VA, USA, Apr. 6–8, 2013.
IV. CONCLUSION
[12] L. Harnefors, X. Wang, A. G. Yepes, and F. Blaabjerg, “Passivity based
A simplified small-signal modeling method for GTIs has stability assessment of grid-connected VSCs—An overview,” IEEE J.
been proposed in this paper. Furthermore, it can be Emerging Sel. Topics Power Electron., vol. 4, no. 1, pp. 116–125, Mar.
2016.
successfully applied to the analysis of resonance mechanism
and stability assessment of the GTI-weak grid coupling system
when harmonic resonance frequency is located in low-medium

2899

Authorized licensed use limited to: North China Electric Power University. Downloaded on May 05,2023 at 01:17:50 UTC from IEEE Xplore. Restrictions apply.