4334 IEEE TRANSACTIONS ON INDUSTRY APPLICATIONS, VOL. 59, NO.

4, JULY/AUGUST 2023

Multivariable Grid-Forming Converters

With Direct States Control
Meng Chen , Member, IEEE, Dao Zhou , Senior Member, IEEE, and Frede Blaabjerg , Fellow, IEEE

Abstract—This paper presents an improved multi-input multi- Up to now, several basic grid-forming controls have been
output based grid-forming (MIMO-GFM) converter control widely researched, e.g., droop control [7], [8], virtual syn-
method using multivariable feedback control. Although the origi- chronous generator (VSG) control [9], [10], [11], power syn-
nal MIMO-GFM control is proven to be superior and robust using
only low-order controllers, it is easily affected by high-frequency chronization control [12], matching control [13], etc. All of them
components, particularly for the converter without inner cascaded are proposed based on various assumptions of loops decoupling
voltage and current loops and when it is connected into a strong such as AC power and DC voltage loops, active and reactive
grid. The proposed method selects frequency and internal voltage power loops. In other words, they try to treat the grid-forming
as state variables to be directly controlled while effectively eliminat- converter as several decoupled single-input single-output (SISO)
ing the impact of high-frequency components without increasing
system complexity. The H∞ synthesis is used to tune the parameters systems [9], [10]. These assumptions simplify the analysis but,
to obtain an optimized performance. Experimental results and at the same time, may sacrifice the performance. Therefore,
system-level simulations verify the effectiveness of the proposed although the aforementioned grid-forming controls can basically
method. achieve the frequency and voltage regulation, they may not
Index Terms—Multi-input multi-output grid-forming (MIMO- be superior and robust to different operation conditions. To
GFM), direct states control, H∞ synthesis, power converter, loops improve the performance of the basic grid-forming controls,
coupling. several kinds of their improved forms have been proposed, which
are usually with higher-order controllers and more complicated
control structures to deal with each loops [14], [15].
I. INTRODUCTION Recently, a new perspective from the multi-input multi-output
(MIMO) system to construct and design the grid-forming control
HE integration of inverter-interfaced generators (IIGs) has
T become a vital aspect of modern power systems due to
growing concerns about environmental issues and fossil fuel
has been proposed [16], [17]. In this way, the coupling informa-
tion among different loops can be used to improve the perfor-
mance with simple control structures. In [17], the fundamental
consumption [1], [2]. Grid-following control is commonly used theory has been studied using a multivariable feedback control
in systems with low penetration of IIGs, where the stability is in detail, where a control transfer matrix is used to deal with
maintained by synchronizing with the conventional synchronous all the AC power and DC voltage loops as a MIMO integrity
generator-dominated system, e.g., through a phase-locked loop and unify different kinds of grid-forming controllers. There-
(PLL) [3]. With the rapid increase in the penetration of IIGs, the after, a MIMO based grid-forming (MIMO-GFM) controller is
ability of self-establishment of the frequency and voltage with- proposed, which can provide a superior and robust performance
out relying on external power sources is supposed as essential without increasing the order of the system. Nevertheless, due to
for some of the IIGs in the future, particularly in systems with the coupling between the AC power and DC voltage loops, the
100% IIGs [4]. In this regard, grid-forming control presents a frequency and internal voltage of the MIMO-GFM converter is
promising solution [5], [6]. sensitive to the high-frequency components of the error signals,
which is inevitable. This is because the frequency and internal
voltage are not direct state variables. The influence will be
Manuscript received 20 December 2022; revised 27 March 2023; accepted more obvious when the grid-forming converter is connected to
11 April 2023. Date of publication 21 April 2023; date of current version 19
July 2023. Paper 2022-IPCC-1654.R1, presented at the 2022 IEEE Energy Con- a strong grid without cascaded voltage and current loops. In this
version Congress and Exposition, Detroit, MI, USA, Oct. 9–13, and approved case, in practice, pre-filters for decreasing the high-frequency
for publication in the IEEE TRANSACTIONS ON INDUSTRY APPLICATIONS by disturbances are necessary.
the Industrial Power Converter Committee of the IEEE Industry Applications
Society [DOI: 10.1109/ECCE50734.2022.9947432]. This work was supported It should be mentioned that such pre-filters are very common
by the Reliable Power Electronic-Based Power System (REPEPS) Project at in other control strategies, e.g., the power filters in the droop con-
the Department of Energy Technology, Aalborg University as part of the Villum trol [18] and the DC voltage filter in the matching control [19].
Investigator Program funded by the Villum Foundation. (Corresponding author:
Meng Chen.) As the pre-filters are not related to the principle of the control
The authors are with the AAU Energy, Aalborg University, 9220 Aalborg, strategies, they are sometimes neglected during the analysis in
Denmark (e-mail: [email protected]; [email protected]; [email protected]). some literature, which is not preferable. Take the droop control
Color versions of one or more figures in this article are available at
https://doi.org/10.1109/TIA.2023.3269021. as an example, it has been revealed that the system is hard to
Digital Object Identifier 10.1109/TIA.2023.3269021 be stable for some systems without the power filters, which is

0093-9994 © 2023 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission.
See https://www.ieee.org/publications/rights/index.html for more information.

Authorized licensed use limited to: Seoul National Univ of Science & Tech (SNUT). Downloaded on December 13,2024 at 03:12:26 UTC from IEEE Xplore. Restrictions apply.
 CHEN et al.: MULTIVARIABLE GRID-FORMING CONVERTERS WITH DIRECT STATES CONTROL 4335

Then the state-space model of the topology can be built in the

d-q frame and it is represented in a compact form in Fig. 1(b) as
ẋ = f (x, u, d) (1)
y = g(x, u) (2)
and the vectors are defined as
 T
x = id iq vd vq iod ioq δ vdc (3)
 T
u = iu ω u Eu (4)
 T
y = vdc p ωu q V (5)
 T
d = ωg V g , (6)
Fig. 1. Grid-forming converter. (a) Topology. (b) Equivalent closed-loop
multivariable feedback control configuration. where idq are the currents of the filter inductor, vdq are voltages
of the filter capacitor, iodq are the output currents, δ is the angle
difference between the grid-forming converter internal voltage
due to the lack of enough inertia of the pure droop control [20]. and the grid voltage, vdc is the DC voltage, ωu and Eu are
Moreover, it is also proven that the droop control with power the frequency and internal voltage to be obtained by the grid-
filters is equivalent to the VSG control to some extent, which forming controller, p and q are the output active and reactive
will not be found if neglecting the power filters [21]. In summary, powers, respectively, V is the magnitude of the capacitor voltage,
the pre-filters may, on the one hand, lead to mismatch between ωg and Vg are the frequency and magnitude of the grid voltage.
the theoretical analysis and the practical results, and on the other Moreover, the detailed mathematical model of the system can
hand, increase the order and complexity of the system [22]. be given as
Motivated from the aforementioned analysis, this paper ex- ωb ωb
tends the work in [17]. First, the reason why the original MIMO- i̇d = Eu − v d + ω b ω u iq (7)
Lf Lf
GFM control is high-frequency sensitive is analyzed from the ωb
point view of the state-space. Thereafter, a new form of control i̇q = − vq − ωb ωu id (8)
Lf
transfer matrix is proposed. The coupling information can still be
used to provide a superior and robust performance. Meanwhile, ωb ωb
v̇d = id − iod + ωb ωu vq (9)
the frequency and internal voltage are chosen as state variables Cf Cf
to be controlled [1]. As a result, the impact of the high-frequency ωb ωb
v̇q = iq − ioq − ωb ωu vd (10)
components on the frequency and internal voltage is suppressed Cf Cf
without relying on the pre-filters and thus without increasing the
ωb ωb ωb R g
complexity of the control system. i̇od = vd − Vg cos δ − iod + ωb ωu ioq (11)
The rest of this paper is organized as follows. In Section II, Lg Lg Lg
the preliminaries on MIMO-GFM control is introduced, where ωb ωb ωb R g
its problem is also analyzed in this section. The proposed control i̇oq = vq + Vg sin δ − ioq − ωb ωu iod (12)
Lg Lg Lg
and design are given in Section III. In Section IV, experimental
results and system-level simulation are presented, and finally, δ̇ = ωb ωu − ωb ωg (13)
conclusions are drawn in Section V. ωb ω b Eu i d
v̇dc = iu − (14)
Cdc Cdc vdc
II. PRELIMINARY OF MIMO-GFM CONVERTERS AND p = vd iod + vq ioq (15)
PROBLEM DEFINITION q = −vd ioq + vq iod (16)
A. Basic Principle of MIMO-GFM Converters 
V = vd2 + vq2 (17)
Fig. 1 shows the studied topology of the grid-forming con-
verter and its equivalent MIMO closed-loop feedback control As observed, the DC dynamics, active power dynamics, and
configuration. A typical three-phase voltage source inverter is reactive power dynamics are coupled with each other. Although
connected to the power grid via an LC filter and a grid line, most of the researches assume that those dynamics are de-
where Lf and Cf are the inductance and capacitance of the LC coupled during the closed-loop design, they are not decoupled
filter, Lg and Rg are the equivalent inductance and resistance of exactly. The coupling information could provide extra freedom
the grid line. If an LCL filter is used, the grid-side inductance to improve the performance of the system. Motivated from this
can be considered into Lg . The DC source is represented by point, in Fig. 1(b), Φ(s) = (φij )3×5 is a control transfer matrix,
a controlled-current source iu paralleled with a DC capacitor, which copes with all the loops as a MIMO integrity rather than
where the capacitance is Cdc . decoupled SISO systems. In [17], Φ is designed as follows for

Authorized licensed use limited to: Seoul National Univ of Science & Tech (SNUT). Downloaded on December 13,2024 at 03:12:26 UTC from IEEE Xplore. Restrictions apply.
 4336 IEEE TRANSACTIONS ON INDUSTRY APPLICATIONS, VOL. 59, NO. 4, JULY/AUGUST 2023

the MIMO-GFM converter

ΦM IM O
⎡ ⎤
kpdc + kidc
s k12 0 k14 k15
⎢ Dp k22 /(s + k22 ) 0 k24 ⎥
=⎣ k21 k24 Dq ⎦
k34 /Dq
k31 k32 0 k34 /s s
(18)
where Dp and Dq are the droop coefficients for active and re-
active power controls, respectively. As shown, the MIMO-GFM
controller keeps all the favorable features of the basic controllers,
i.e., inertia and droop characteristics. Meanwhile, it uses several
simple proportional gains to deal with the coupling terms, which
has been proved as an effective way to provide a superior and
robust performance without increasing the order of the system.

B. Impact of High-Frequency Components

To better illustrate the problem of the MIMO-GFM controller, Fig. 2. Practical control block diagram of MIMO-GFM converter with pre-
we will first consider a basic VSG control. The control transfer filters.
matrix Φ can be derived by setting all the coupling gains of (18)
as zero
3) Comparing (23) with (27), the state variables between the
ΦV SG VSG and MIMO-GFM controls are quite differnt.
⎡ ⎤ Therefore, it is clear that the principle of the MIMO-GFM
kpdc + kidc /s 0 0 0 0
=⎣ 0 Dp k22 /(s + k22 ) 0 0 0 ⎦ control is to change the state variables by adding the coupling
k34 /Dq terms but not to change the form of the state differential equa-
0 0 0 k34 /s s tions compared with the basic VSG control. These coupling
(19)
terms are expected to improve the performance because they
based on which the state differential equations of ΦV SG can be provide a multi-degree-of-freedom control and the useful cou-
expressed as pling information among various loops can be considered as
ẋφV SG1 = kidc e1 (20) well.
Nevertheless, due to the changes of the state variables, the
ẋφV SG2 = −k22 xφV SG2 + Dp k22 e2 (21) frequency ω̃u and internal voltage Ẽu are not the state variables
k34 to be directly controlled anymore for the MIMO-GFM converter.
ẋφV SG3 = k34 e4 + e5 (22) Instead, they will be directly influenced by the errors e as
Dq
and the state variables are defined as shown in (27), especially the high-frequency components, which
 T are inevitable. On one hand, there are always high-frequency
xφV SG = ĩu − kpdc e1 ω̃u Ẽu (23) components in the steady-state powers and DC voltage. On the
Similarly, the state differential equations of ΦM IM O in (18) other hand, steps of the references Y ref will also introduce
can be expressed as the high-frequency components into the errors. Therefore, in
ẋφM IM O1 = kidc e1 (24) practice, pre-filters are usually used before the coupling terms
in order to suppress the influence of the high-frequency compo-
ẋφM IM O2 = −k22 xφM IM O2 + Dp k22 e2 (25) nents as shown in Fig. 2. However, these pre-filters increase the
k34 order and complexity of the system and may highly influence
ẋφM IM O3 = k34 e4 + e5 (26) the stability.
Dq
and the state variables are defined as
⎡ ⎤ III. PROPOSED DIRECT STATES CONTROL
ĩu − kpdc e1 − k12 e2 − k14 e4 − k15 e5
xφM IM O = ⎣ ω̃u − k21 e1 − k24 e4 − Dk24
q
e5 ⎦ (27) A. Principle of Proposed Direct States Control
Ẽu − k31 e1 − k32 e2 According to the aforementioned analysis, the problem of the
The following conclusions can be summarized. MIMO-GFM control is due to the fact that the added coupling
1) From (20)–(23), the control of the basic VSG is decoupled, terms change the state variables compared with the basic VSG
i.e., ĩu is controlled by e1 , ω̃u is controlled by e2 , Ẽu is control. Motivated from this point, an improved MIMO-GFM
controlled by e4 and e5 . control is proposed, where the state differential equations of the
2) Comparing (20)–(22) with (24)–(26), it is observed that control transfer matrix is designed as
the state differential equations of the VSG and MIMO- k14
ẋφ1 = −k12 xφ2 + kidc e1 + Dp k12 e2 + k14 e4 + e5 (28)
GFM controls have the same structure. Dq

Authorized licensed use limited to: Seoul National Univ of Science & Tech (SNUT). Downloaded on December 13,2024 at 03:12:26 UTC from IEEE Xplore. Restrictions apply.
 CHEN et al.: MULTIVARIABLE GRID-FORMING CONVERTERS WITH DIRECT STATES CONTROL 4337

k24
ẋφ2 = −k22 xφ2 + k21 e1 + Dp k22 e2 + k24 e4 + e5 (29)
Dq
k34
ẋφ3 = −k32 xφ2 + k31 e1 + Dp k32 e2 + k34 e4 + e5 (30)
Dq
and the state variables are defined as
 T
xφ = ĩu − kpdc e1 ω̃u Ẽu (31)
The following conclusions about the proposed control can be
summarized.
1) Comparing (31) with (23), the proposed control chooses
the same state variables as the basic VSG control. Espe-
cially, the frequency and internal voltage are still directly
controlled state variables.
2) Comparing (28)–(30) with (20)–(22) and (24)–(26), it
is observed that the state differential equations of the
proposed control has a different structure, which is taking
the coupling terms into consideration.
In this way, the proposed control can not only suppress the Fig. 3. Block diagram of proposed control transfer matrix Φ.
influence of the high-frequency components but also improve
TABLE I
the performance using the coupling terms. It should also be PARAMETERS OF TEST SYSTEM
mentioned, from (28)–(30), that the proposed method still hold
the steady-state droop characteristics. Therefore, the steady-
state operation points will be automatically determined through
the defined droop characteristics when multiple inverters are
paralleled. The corresponding control transfer matrix of the
proposed method can be derived as
Φ(s) = (φij )3×5 (32)
where
kpdc s2 + (kpdc k22 + kidc )s + kidc k22 − k12 k21
φ11 = (33)
s2 + k22 s
Dp k12 k14 s + k14 k22 − k12 k24
φ12 = , φ14 = (34)
s + k22 s2 + k22 s
(k14 /Dq )s + k14 k22 /Dq − k12 k24 /Dq
φ15 = (35)
s2 + k22 s
k21 Dp k22
φ21 = , φ22 = (36)
s + k22 s + k22
k24 k24 /Dq
φ24 = , φ25 = (37)
s + k22 s + k22
k31 s + k22 k31 − k21 k32 Dp k32
φ31 = , φ32 = (38)
s2 + k22 s s + k22 only uses simple proportional controllers to include the coupling
k34 s + k22 k34 − k24 k32 terms. After obtaining ω̃u , the angle of the power converter
φ34 = (39)
s2 + k22 s voltage can be derived as
t
(k34 /Dq )s + k22 k34 /Dq − k24 k32 /Dq
φ35 = (40) θ= (1 + ω̃u )ωb dτ (42)
s2 + k22 s 0

φ13 = φ23 = φ33 = 0 (41) where θ is used for all the transformations between d-q frame
and abc frame.
According to the aforementioned analysis, the block diagram
of the proposed control transfer matrix is as shown in Fig. 3.
B. Parameters Design Based on H∞ Optimization
It is worth mentioning that although the above elements of φij
seem to have complicated forms, according to (28)–(31), the To make a fair comparison with the original MIMO-GFM
order of the system is not increased and the control structure control, this paper also uses the H∞ synthesis to tune the
is straightforward as well, which is because φij actually share parameters to obtain an optimal performance [17]. Therefore, the
many common parts as shown in Fig. 3. As observed, it still block diagram of the proposed control transfer matrix in Fig. 3

Authorized licensed use limited to: Seoul National Univ of Science & Tech (SNUT). Downloaded on December 13,2024 at 03:12:26 UTC from IEEE Xplore. Restrictions apply.
 4338 IEEE TRANSACTIONS ON INDUSTRY APPLICATIONS, VOL. 59, NO. 4, JULY/AUGUST 2023

TABLE II
PARAMETERS OF CONTROL TRANSFER MATRIX FOR ORIGINAL MIMO-GFM
AND PROPOSED CONTROLLERS

Fig. 4. Equivalent block diagram of proposed control transfer matrix for the
formulation of H∞ synthesis.

Fig. 5. Block diagram of grid-forming converter in linear fractional

transformation.

is equivalently changed to Fig. 4 by defining two intermediate

vectors û and ŷ, which have the following relationship
û = diag(kpdc , kidc , k21 , k31 , k12 I 2 , k22 I 2 , k32 I 2 , k14 ,
Fig. 6. Log magnitude curves of designed transfer function T21 (s) and its
k24 , k34 )ŷ weighting function W21 (s).

= K ŷ (43)
s + 60
where K is a gain vector only containing all the parameters W41 = (51)
to be tuned. Thereafter, the standard form of linear fractional s + 0.006
transformation for H∞ optimization can be obtained as shown in and the considerations of choosing the weighting functions can
Fig. 5, where the grid-forming converter in Fig. 1(b) is collapsed be found in [17]. Finally, the parameters can be derived by
into G (except for K). The disturbance inputs and evaluation solving the following H∞ optimization problem
outputs for the H∞ synthesis are defined as min ||diag(Wij (s)Tij (s))||∞ (52)
 T K
w = Pref ωg (44) According to the aforementioned method with the parameters
 T listed in Table I, the parameters of the proposed control can be
z = Pref − p p ωu q + V /Dq (45)
derived as shown in Table II. As an example, Fig. 6 presents the
where the transfer functions from wj to zi is denoted as Tij (s). log magnitude curves of the designed transfer function T21 (s)
For example, T21 (s) is the transfer function from w1 = Pref to from Pref to p and its weighting function W21 (s). As observed,
z2 = p. These transfer functions are limited by the following by solving (52), T21 (s) is limited by W21 −1
(s). Therefore, the
chosen weighting functions Wij (s) high-frequency suppression of T21 (s) can be guaranteed by
s+4 properly choosing W21 (s). Other transfer functions as well as
W11 (s) = (46)
s + 0.0004 their corresponding weighting functions are similar. For the
2 following comparison, the parameters of the original MIMO-
1.447 × 10−3 s + 1 GFM control are also presented in Table II. Supposing there
W21 (s) = (47)
1.447 × 10−5 s + 1 is a disturbance in e1 , Fig. 7 compares the bode diagrams of
1 1.447 × 10−3 s + 1 the closed-loop transfer function from this disturbance to ωu .
W22 (s) = × (48) The advantage of considering a disturbance in e1 compared to a
100 1.447 × 10−5 s + 1
disturbance in reference is that both the disturbances in the ref-
1 s erence and the feedback channel can be included. As observed,
W31 (s) = × (49)
0.015 1.447 × 10−5 s + 1 the original MIMO-GFM control without those pre-filters fails
1.447 × 10−3 s + 1 to suppress the components with the frequency over 200 rd/s.
W32 (s) = (50) In comparison, the proposed control make the log-magnitude
1.447 × 10−5 s + 1

Authorized licensed use limited to: Seoul National Univ of Science & Tech (SNUT). Downloaded on December 13,2024 at 03:12:26 UTC from IEEE Xplore. Restrictions apply.
 CHEN et al.: MULTIVARIABLE GRID-FORMING CONVERTERS WITH DIRECT STATES CONTROL 4339

Fig. 7. Comparison of bode diagram between original MIMO-GFM control

and proposed control without any pre-filter.

Fig. 9. Experimental comparison when Pref steps from 0.5 p.u. to 1 p.u.
(a) Original MIMO-GFM control. (b) Proposed MIMO-GFM control using
direct state control.

from 0.5 p.u. to 1 p.u. It should be mentioned that no any

pre-filter is included in the control system. As shown, although
the original MIMO-GFM control can guarantee the stability
Fig. 8. Experimental configuration of MIMO-GFM converter.
and good low-frequency dynamics, the focused variables have
large high-frequency components especially in the frequency
curve to decrease with a slope of −40 dB/decade even though ωu . Therefore, as in the aforementioned discussion, pre-filters
we have neither defined an explicit weighting function to limit are necessary, where the corresponding waveforms can be found
the corresponding high-frequency components nor added any in [17]. In comparison, the proposed control can highly damp
pre-filter, which verify the advantages of the proposed control the high-frequency components without additional means.
method. A further comparison when ωg decreases from 50 Hz to
49.9 Hz is shown in Fig. 10. Similar to Fig. 9, the proposed
IV. EXPERIMENTAL RESULTS AND SYSTEM-LEVEL control can well damp the high-components, which is obviously
SIMULATION VALIDATION observed with the original MIMO-GFM control. The results of
Figs. 9 and 10 are in accordance with the theoretical analysis of
A. Experimental Results Fig. 7.
The performance of the proposed control is tested by using
the experimental setup shown in Fig. 8. The parameters are the
B. System-Level Simulation Validation
same as those in Tables I and II, which represents a system of a
grid-forming converter connected to a strong grid. The structure This section study the system-level performance of the pro-
of the control system is same as Fig. 1 without cascaded voltage posed method with disturbances from both inverter-side and
and current loops. Meanwhile, due to the DC control is fixed in system-side using the IEEE 9-Bus Test System shown in Fig. 11,
the DC source, only the power loops can be tested. Nevertheless, where G3 is replaced by a VSC. The parameters are same as
it does not influence the conclusion. those in Tables I and II in p.u. values. The average model of the
Fig. 9 compares the experimental results with the original inverter is used to improve the speed of the simulation. There-
MIMO-GFM control and the proposed control when Pref steps fore, the steady-state high-frequency components, which has

Authorized licensed use limited to: Seoul National Univ of Science & Tech (SNUT). Downloaded on December 13,2024 at 03:12:26 UTC from IEEE Xplore. Restrictions apply.
 4340 IEEE TRANSACTIONS ON INDUSTRY APPLICATIONS, VOL. 59, NO. 4, JULY/AUGUST 2023

Fig. 10. Experimental comparison when ωg decreases from 50 Hz to 49.9 Hz.

(a) Original MIMO-GFM control. (b) Proposed MIMO-GFM control using
direct state control.

Fig. 12. Simulation results using IEEE 9-Bus Test System.

controls can guarantee the stable operation of the system re-

sponding to different disturbances. However, when a DC-side
disturbance occurs, large high-frequency oscillations will be
excited in the frequency and output powers of the inverter with
original MIMO-GFM control. It has been explained from Fig. 2
that an error of e1 = Vdcref − vdc will directly influence the
frequency and internal voltage if there is not a pre-filter. Such
high-frequency oscillations can be suppressed if pre-filters are
Fig. 11. Single-line diagram of IEEE 9-Bus Test System. added. However, as the time constant of pre-filters increases,
oscillations with low frequencies will occur as the pre-filters are
not considered during the design of the original MIMO-GFM
been well shown in the experiments are neglected. The following control. In comparison, the proposed MIMO-GFM control can
disturbances are exerted to the system from the steady-state: well dampen such oscillations.
1) t = 5 s, a grid-side distubance is applied, where the output
of G1 increases 0.1 p.u.;
2) t = 10 s, a inverter-side disturbance is applied, where Pref V. CONCLUSION
of the VSC increases from 0.5 p.u. to 1 p.u.; This paper proposes a novel control transfer matrix for
3) t = 15 s, a DC-side disturbance is applied, where Vdcref the MIMO-GFM control. Instead of changing the states, the
of the VSC increases from 1 p.u. to 1.01 p.u.; coupling terms of the proposed method only change the structure
Fig. 12 shows the simulation results using the IEEE 9-Bus of the state differential equations and, at the same time, keep the
Test System. It is observed that, in general, both the original frequency and voltage as the controlled states. By designing with
MIMO-GFM (without and with pre-filters) and the proposed the same H∞ synthesis as the original MIMO-GFM control, the

Authorized licensed use limited to: Seoul National Univ of Science & Tech (SNUT). Downloaded on December 13,2024 at 03:12:26 UTC from IEEE Xplore. Restrictions apply.
 CHEN et al.: MULTIVARIABLE GRID-FORMING CONVERTERS WITH DIRECT STATES CONTROL 4341

proposed method will have improved ability to decrease the im- [21] J. Liu, Y. Miura, and T. Ise, “Comparison of dynamic characteristics
pact of the high-frequency components on the system dynamics between virtual synchronous generator and droop control in inverter-based
distributed generators,” IEEE Trans. Power Electron., vol. 31, no. 5,
without increasing the complexity of the control system. pp. 3600–3611, May 2016.
[22] S. Dong and Y. C. Chen, “A method to directly compute synchronverter
REFERENCES parameters for desired dynamic response,” IEEE Trans. Energy Convers.,
vol. 33, no. 2, pp. 814–825, Jun. 2018.
[1] M. Chen, D. Zhou, and F. Blaabjerg, “Multivariable grid-forming con-
verters with direct states control,” in Proc. IEEE Energy Convers. Congr.
Expo., 2022, pp. 1–7.
[2] B. K. Bose, “Power electronics in smart grid and renewable energy Meng Chen (Member, IEEE) received the B.Eng.
systems,” Proc. IEEE, vol. 105, no. 11, pp. 2011–2018, Nov. 2017. degree in electrical engineering and automation from
[3] M. Eskandari and A. V. Savkin, “Robust PLL synchronization unit for the Qingdao University of Science and Technology,
grid-feeding converters in micro/weak grids,” IEEE Trans. Ind. Inform., Qingdao, China, in 2013, the first Ph.D. degree in
vol. 19, no. 4, pp. 5400–5411, Apr. 2022. power system and automation from North China Elec-
[4] F. Deng, Y. Li, X. Li, W. Yao, X. Zhang, and P. Mattavelli, “A decentralized tric Power University, Beijing, China, in 2018, and the
impedance reshaping strategy for balanced, unbalanced and harmonic second Ph.D. degree in energy technology from Aal-
power sharing in islanded resistive microgrids,” IEEE Trans. Sustain. borg University, Aalborg, Denmark, in 2022. He was
Energy, vol. 13, no. 2, pp. 743–754, Apr. 2022. an academic Guest with the Swiss Federal Institute
[5] J. Rocabert, A. Luna, F. Blaabjerg, and P. Rodríguez, “Control of power of Technology (ETH) Zurich, Zurich, Switzerland, in
converters in AC microgrids,” IEEE Trans. Power Electron., vol. 27, no. 11, 2021. Since 2022, he was been with Aalborg Uni-
pp. 4734–4749, Nov. 2012. versity, where he is currently a Postdoctoral Researcher. His research interest
[6] R. Rosso, X. Wang, M. Liserre, X. Lu, and S. Engelken, “Grid-forming include the control, stability, and reliability of power systems with power
converters: Control approaches, grid-synchronization, and future trends–A electronics. He was the recipient of the 2020 MPCE Best Paper Award.
review,” IEEE Open J. Ind. Appl., vol. 2, pp. 93–109, 2021.
[7] M. Chen and X. Xiao, “Hierarchical frequency control strategy of hy-
brid droop/VSG-based islanded microgrids,” Electric Power Syst. Res.,
vol. 155, pp. 131–143, Feb. 2018. Dao Zhou (Senior Member, IEEE) received the B.S.
[8] N. Mohammed and M. Ciobotaru, “Adaptive power control strategy for degree in electrical engineering from Beijing Jiaotong
smart droop-based grid-connected inverters,” IEEE Trans. Smart Grid, University, Beijing, China, in 2007, the M. S. degree
vol. 13, no. 3, pp. 2075–2085, May 2022. in electrical engineering from Zhejiang University,
[9] H. Wuet al., “Small-signal modeling and parameters design for virtual Hangzhou, China, in 2010, and the Ph.D. degree in
synchronous generators,” IEEE Trans. Ind. Electron., vol. 63, no. 7, electrical engineering from Aalborg University, Aal-
pp. 4292–4303, Jul. 2016. borg, Denmark, in 2014. Since 2014, he has been with
[10] J. Chen and T. O’Donnell, “Parameter constraints for virtual synchronous the Department of Energy Technology, Aalborg Uni-
generator considering stability,” IEEE Trans. Power Syst., vol. 34, no. 3, versity, where currently he is an Associate Professor.
pp. 2479–2481, May 2019. His research interests include modeling, control, and
[11] M. Chen, D. Zhou, C. Wu, and F. Blaabjerg, “Characteristics of parallel reliability of power electronics in renewable energy
inverters applying virtual synchronous generator control,” IEEE Trans. applications. He is an Associate Editor for IET Renewable Power Generation
Smart Grid, vol. 12, no. 6, pp. 4690–4701, Nov. 2021. and IET Power Electronics. He was the recipient of a few IEEE prized paper
[12] D. Pan, X. Wang, F. Liu, and R. Shi, “Transient stability of voltage-source awards.
converters with grid-forming control: A design-oriented study,” IEEE J.
Emerg. Sel. Top. Power Electron., vol. 8, no. 2, pp. 1019–1033, Jun. 2020.
[13] C. Arghir, T. Jouini, and F. Dörfler, “Grid-forming control for power Frede Blaabjerg (Fellow, IEEE) From 1988 to 1992,
converters based on matching of synchronous machines,” Automatica, he received the Ph.D. degree in electrical engineer-
vol. 95, pp. 273–282, Sep. 2018. ing from Aalborg University, Aalborg, Denmark, in
[14] J. Liu, Y. Miura, H. Bevrani, and T. Ise, “A unified modeling method 1995. From 1987 to 1988, he was with ABB-Scandia,
of virtual synchronous generator for multi-operation-mode analyses,” Randers, Denmark. He became an Assistant Professor
IEEE J. Emerg. Sel. Top. Power Electron., vol. 9, no. 2, pp. 2394–2409, in 1992, an Associate Professor in 1996, and a Full
Apr. 2021. Professor of power electronics and drives in 1998
[15] M. Chen, D. Zhou, and F. Blaabjerg, “Active power oscillation damping at AAU Energy. From 2017, he has been a Villum
based on acceleration control in paralleled virtual synchronous generators Investigator. He is honoris causa with University Po-
system,” IEEE Trans. Power Electron., vol. 36, no. 8, pp. 9501–9510, litehnica Timisoara (UPT), Timisoara, Romania, in
Aug. 2021. 2017 and Tallinn Technical University (TTU), Es-
[16] L. Huang, H. Xin, and F. Dörfler, “H∞ -control of grid-connected con- tonia, in 2018. He has authored or coauthored more than 600 journal papers
verters: Design, objectives and decentralized stability certificates,” IEEE in the fields of power electronics and its applications. His research interests
Trans. Smart Grid, vol. 11, no. 5, pp. 3805–3816, Sep. 2020. include power electronics and its applications such as, in wind turbines, PV
[17] M. Chen, D. Zhou, A. Tayyebi, E. Prieto-Araujo, F. Dörfler, and F. systems, reliability, Power-2-X, power quality and adjustable speed drives. He
Blaabjerg, “Generalized multivariable grid-forming control design for is the co-author of eight monographs and Editor of fourteen books in power
power converters,” IEEE Trans. Smart Grid, vol. 13, no. 4, pp. 2873–2885, electronics and its applications. He was the recipient of the 38 IEEE Prize Paper
Jul. 2022. Awards, IEEE PELS Distinguished Service Award in 2009, EPE-PEMC Council
[18] H. Zhang, W. Xiang, W. Lin, and J. Wen, “Grid forming converters Award in 2010, IEEE William E. Newell Power Electronics Award 2014, Villum
in renewable energy sources dominated power grid: Control strategy, Kann Rasmussen Research Award 2014, Global Energy Prize in 2019 and 2020
stability, application, and challenges,” J. Mod. Power Syst. Clean Energy, IEEE Edison Medal. From 2006 to 2012, he was the Editor-in-Chief of IEEE
vol. 9, no. 6, pp. 1239–1256, Nov. 2021. TRANSACTIONS ON POWER ELECTRONICS. He was Distinguished Lecturer of the
[19] A. Tayyebi, D. Gross, A. Anta, F. Kupzog, and F. Dörfler, “Frequency IEEE Power Electronics Society from 2005 to 2007 and for the IEEE Industry
stability of synchronous machines and grid-forming power converters,” Applications Society from 2010 to 2011 and 2017 to 2018. During 2019–2020,
IEEE J. Emerg. Sel. Top. Power Electron., vol. 8, no. 2, pp. 1004–1018, he was a President of IEEE Power Electronics Society. He is a Vice-President of
Jun. 2020. the Danish Academy of Technical Sciences. He is nominated in 2014–2021 by
[20] Arani and El-Saadany, “Implementing virtual inertia in DFIG-based wind Thomson Reuters to be between the most 250 cited researchers in Engineering
power generation,” IEEE Trans. Power Syst., vol. 28, no. 2, pp. 1373–1384, in the world.
May 2013.

Authorized licensed use limited to: Seoul National Univ of Science & Tech (SNUT). Downloaded on December 13,2024 at 03:12:26 UTC from IEEE Xplore. Restrictions apply.