Model Predictive Control for Master-Slave Inverters

in Microgrids
IECON 2022 – 48th Annual Conference of the IEEE Industrial Electronics Society | 978-1-6654-8025-3/22/$31.00 ©2022 IEEE | DOI: 10.1109/IECON49645.2022.9968694

Fernanda Carnielutti1 , Mokhtar Aly2 , Margarita Norambuena3 , José Rodriguez2

1
Federal University of Santa Maria (UFSM), Santa Maria, Brazil
2 Facultad
de Ingenierı́a y Tecnologı́ıa, Universidad San Sebastian, Bellavista 7, Santiago, Chile
3 Electrical Engineering department, Universidad Tecnica Federico Santa Maria, Valparaiso 2390123, Chile

[email protected], [email protected], [email protected], [email protected]

Abstract—This paper proposes a Master-Slave Model Predic- disadvantages, such as inherently slow dynamic responses [2].
tive Control for parallel grid-tied inverters in a microgrid. In Also, with these controllers, it is more difficult to deal with
this configuration, the Master is a grid-forming inverter with multi-objective and non-linear control problems.
a Battery Energy Storage System as dc input, while the Slave
is a grid-following PV inverter that provides the power to the Recently, Model Predictive Control, MPC, is being consid-
load. First, the inverter models are derived and then their MPC ered as an interesting and efficient alternative for controlling
controllers are presented in details. Finally, Hardware-in-the- inverters in microgrids [2]–[4], [8]. One of the most common
Loop results are presented for different operational conditions MPC approaches is the Finite Control Set MPC, FCS-MPC,
for the microgrid, including grid connection, islanded mode and that uses the discrete-time system model to predict its future
load variations. The results demonstrate the advantages of the
proposed MPC such as faster dynamic response, multi variable states [9]–[11]. The predicted state that minimizes a cost
control, adequate load sharing between the inverters, robustness function is selected and implemented by the system. The
to parametric uncertainties and variations and the possibility cost function can be designed to encompass multiple control
to directly include system constrains and non-linearities in the objectives and non-linearities. This allows the MPC to consider
controller implementation. in its formulation the operational constrains of the system,
Index Terms—Predictive Control, Microgrids, Grid-Forming
Inverters, Grid-Following Inverters. yielding a faster dynamic response when compared to classical
linear controllers, among other advantages.
MPC for microgrids can be divided into converter-level and
I. I NTRODUCTION
grid-level control. The first one deals with the low-level control
Microgrids are systems that can operate connected to the of the inverters in the microgrid, associated to the primary
main grid or islanded [1]. They are comprised of distributed control i.e. current and power control, regulation of voltages of
generation units (DGs), which are usually renewable energy the inner capacitors in multilevel inverters, virtual impedance
sources (such as wind and solar), energy storage systems, emulation, islanded operation, etc. Many works have been
diesel gensets, etc. with their associated power electronics presented in the literature dealing with the grid connection
converters, distribution lines, loads, relays and communication of single- and three-phase inverters using MPC, considering
systems [2]. The control of a microgrid is usually divided in both grid-tied and microgrid applications [12]–[21].
three levels [2], [3], i.e. hierarchical control. The primary con- On the other hand, grid-level control works at system level,
trol acts at inverter-level in a time scale of milliseconds to sec- performing the duties of secondary and tertiary controls [3],
onds, providing current and voltage control, virtual impedance [4]. A survey of various control architectures for ac microgrids
emulation, islanding detection, etc [1]. The secondary control was presented in [22]. A centralized MPC was presented in
works in a time scale of seconds to minutes, regulating [23] for load frequency control in multi-area interconnected
voltage and frequency deviations of the microgrid, controlling power systems, and a distributed MPC was presented in [24]
active and reactive power, performing grid synchronization, for automatic generation control of four area interconnected
etc. [4]. The tertiary control is responsible for the interaction microgrids. In [25], an MPC was proposed to restore the
of the microgrid and the main grid at larger time scales, from frequency in microgrids considering communication delays.
minutes to hours (including power flow regulation, market- In [26], a distributed MPC was proposed for controlling the
driven optimal dispatch of the energy sources, etc.). voltage and frequency of autonomous microgrids, while also
In order to implement the above-mentioned hierarchical achieving power sharing.
control, common approaches include Centralized or Dis- Concerning tertiary control, [27] proposed a two-layer MPC
tributed Control, Master-Slave Control, and Droop Control for optimal operation of a microgrid composed of photovoltaic
[1], [5], [6], which are usually performed by means of linear systems, batteries and a diesel genset. The first optimization
controllers and modulators [7]. Although these linear multi- layer calculates the optimal energy dispatch, while the second
loop cascaded controllers are well-known and widely applied one is responsible for adjusting the power of the diesel
in the literature and in practical applications, they have some genset. In [28], an MPC was presented to efficiently optimize

Authorized licensed use limited to: Universidad Tecnica Federico Santa Maria. Downloaded on June 29,2023 at 20:11:00 UTC from IEEE Xplore. Restrictions apply.
the operation of a microgrid, while satisfying time-varying The output phase currents imx of the Master NPC are:
requests and operation constraints. In [29], a sampling-based
Lf
MPC uses the real-time price of energy and forecasts of PV imx (k + 1) = imx (k)+
and load power to properly dispatch the power and minimize Rf Ts + Lf
overall costs in a Photovoltaic-Battery Energy Storage System Ts Ts
(vmx (k) − vmo (k)) − vmcx (k)
(PV-BESS) microgrid. Also considering a PV-BESS system, Rf Ts + Lf Rf Ts + Lf
[30] presents an MPC for the inner loop of the bidirectional where Ts is the sampling period, Lf is the inductance of
buck-boost BESS converter, while the parallel converters are the LC filter and Rf its parasitic resistance, and vmcx is the
controlled by another MPC with a droop control, ensuring a capacitor voltage of the LC filter.
stable ac voltage for the microgrid and proper load sharing. Finally, the predicted control variable at (k+1), i.e. the
In this context, this paper proposes a Master-Slave MPC capacitor voltage of the LC filter vmcx , is given by:
for parallel inverters in a microgrid. Here, the Master is a
BESS grid-forming inverter. It is controlled by a MPC voltage Ts
vmcx (k + 1) = imx (k + 1) + vmcx (k) (3)
control, that synthesizes high-quality ac voltages and performs Cf
power sharing. The Slave is a grid-following PV inverter with where Cf is the filter capacitance.
an MPC current control that provides adequate power to the The voltages of the dc bus capacitors must also be controlled
loads. Advantages of the proposed MPC over traditional linear to their desired values. In order to do so, let us write the current
controllers are faster dynamic response, multi variable control that flows through the upper and lower capacitors as:
and the possibility to directly include system constrains and
non-linearities in the controller implementation. Moreover, the idcy = iBESS −S1ym (k)ima −S2ym (k)imb −S3ym (k)imc
Master-Slave MPC is also robust to parametric uncertainties
where iBESS is the dc current that comes from the BESS,
and variations, while also providing adequate load sharing
y = 1, 2 refers to, respectively, the upper and lower capacitor
between the inverters. In order to validate the proposed ap-
of the dc bus, S1ym , S2ym and S3ym are the upper switches
proach, Hardware-in-the-Loop (HIL) results are presented for
of each inverter phase and ima , imb and imc are the phase
different operational conditions for the microgrid, including
currents. The capacitor voltage is then predicted as:
grid connection, islanded mode and load variations.
Ts
II. P ROPOSED M ODEL P REDICTIVE C ONTROL FOR vdcy (k + 1) = vdcy (k) + idcy (k) (4)
C
M ASTER -S LAVE I NVERTERS
where vdcy is the voltage of the dc bus capacitor and C is its
For the description of the proposed MPC, let us consider capacitance.
a three-phase microgrid consisting of a generic load and two Considering that vmcx must be controlled in order to provide
Neutral Point Clamped, NPC, inverters with output LC filter, phase voltages with the desired amplitude and frequency, and
as seen in Fig.1. The lower NPC is the grid-forming Master that vdcy is controlled in order to have the desired value of
unit, while the upper one is the grid-following Slave unit, that the dc bus voltage, the cost function for the Master MPC can
provides the power to the load and/or grid. As shown in Fig. be defined as:
1, the microgrid can operate connected to the main grid, or in
islanded mode if switch S1 is open. gm = λ1 [(vmca (k + 1) − va∗ (k + 1))2 + (vmcb (k + 1)
− vb∗ (k + 1))2 + (vmcc (k + 1) − vc∗ (k + 1))2 ]
Master Inverter ∗
+ λ2 [(vdc1 (k + 1) − vdc1 (k + 1))2 + (vdc2 (k + 1)
As previously stated, the Master unit is a grid-forming NPC ∗
inverter with a BESS as input. As a result, it must control the − vdc2 (k + 1))2 ] (5)
voltages of the capacitors of the LC filter, in order to provide to where λ1 and λ2 are the weighting factors for the cost
the microgrid an adequate and stable voltage with the desired function, vx∗ (k + 1) are the references for the voltages of the
amplitude and frequency, while also regulating the voltages of ∗
capacitors of the LC filter and vdc1 (k + 1) are the references
the dc bus capacitors. To develop the MPC controller for the for the voltages of the dc bus capacitors. At each Ts , the
Master unit, let us first write the inverter PWM phase voltages cost function gm is evaluated for all inverter voltage vectors,
in the discrete-time domain as: considering their redundancies (27 voltage vectors for the
vmx (k) = (S1xm (k) + S2xm (k) − 1)vdc (1) three-phase NPC), and the vector with the smallest value of
gm is chosen and implemented by the Master unit.
where m is the Master unit, x = a, b, c represents each inverter
phase, vmx (k) are the inverter PWM phase voltages, S1xm and Slave Inverter
S2xm are, respectively, the two upper switches of each inverter The MPC for the Slave unit, that is, the grid-following NPC
phase, and vdc is the voltage of the dc bus capacitors. inverter, is very similar to the one presented for the Master unit
The inverter common-mode voltage vmo is given by: in the previous Subsection. Here, the dc input of the Slave unit
1 is composed of PV panels. The control variables are now the
vmo (k) = (vma (k) + vmb (k) + vmc (k)) (2) output phase currents of the inverter, that is, the currents in
3

Authorized licensed use limited to: Universidad Tecnica Federico Santa Maria. Downloaded on June 29,2023 at 20:11:00 UTC from IEEE Xplore. Restrictions apply.
 Slave Unit
s1as s1bs s1cs
vc1

vPV s2as s2bs s2cs isa

uas
isb
ubs
ucs isc
vc2 s3as s3bs s3cs
3-Phase Switch

s4as s4bs s4cs

S1 Grid

s1am s1bm s1cm

vmca vmcb vmcc
vc1
ila ilb ilc
s2am s2bm s2cm ima
vBESS
uam imb Load
ubm
ucm imc
vc2 s3am s3bm s3cm

s4am s4bm s4cm

Master Unit

Fig. 1. Schematic diagram of the implemented microgrid.

the inductor of the LC filter, as the Slave unit is responsible The control of the voltages of the dc bus capacitors is carried
for providing the desired power to the load and/or grid. As for out in the same way a for the Master unit. Now, the currents
the Master unit, the voltages of the capacitors of the dc bus that flow through the upper and lower capacitors are:
must also be controlled.
Let us start by writing the inverter PWM phase voltages in idcy = iP V − S1ys (k)isa − S2ys (k)isb − S3ys (k)isc
the discrete-time domain as:
where iP V is the dc current that comes from the PV panels,
vsx (k) = (S1xs (k) + S2xs (k) − 1)Vdc (6) y = 1, 2 refers to, respectively, the upper and lower capacitor
of the dc bus, S1ys , S2ys and S3ys are the upper switches of
where s is the Slave unit, x = a, b, c represents each inverter each inverter phase and isa , isb and isc are the phase currents.
phase, vmx (k) are the inverter PWM phase voltages, S1xs and The capacitor voltage is then predicted as:
S2xs are, respectively, the two upper switches of each inverter
Ts
phase, and Vdc is the voltage of the dc bus capacitors. vdcy (k + 1) = vdcy (k) + idcy (k) (8)
C
The inverter common-mode voltage vso is given by:
where vdcy is the voltage of the dc bus capacitor and C is the
1 capacitance.
vso (k) = (vsa (k) + vsb (k) + vsc (k)) (7)
3 For the Slave unit, the controlled variables are isx (k + 1),
The output phase currents isx of the Slave NPC are: that must be controlled in order to provide adequate phase
currents for the grid/load, and the voltages of the dc bus
Lf capacitors. As a result, the cost function for the Slave MPC
isx (k + 1) = isx (k)+
Rf Ts + Lf can be written as:
Ts Ts
(vsx (k) − vso (k)) − vmcx (k) gs = (isa (k + 1) − i∗a (k + 1))2 + isb (k + 1)
Rf Ts + Lf Rf Ts + Lf
− i∗b (k + 1))2 + isc (k + 1) − i∗c (k + 1))2
where Ts is the sampling period, Lf is the inductance of ∗
+ λ2 [(vdc1 (k + 1) − vdc1 (k + 1))2 + (vdc2 (k + 1)
the LC filter and Rf its parasitic resistance, and vmcx is the ∗
voltage of the capacitor of the LC filter. − vdc2 (k + 1))2 ] (9)

Authorized licensed use limited to: Universidad Tecnica Federico Santa Maria. Downloaded on June 29,2023 at 20:11:00 UTC from IEEE Xplore. Restrictions apply.
where λ1 and λ2 are the weighting factors for the cost
function, i∗x (k + 1) are the references for the phase currents
∗
and vdc1 (k + 1) are the references for the voltages of the dc
bus capacitors. At each Ts , the cost function gs is evaluated
for the 27 voltage vectors of the NPC inverter, and the one
with the smallest value of gs is chosen and implemented by
the Slave unit.
For the Slave unit, it is also important to point out that a
”Perturb and Observe” MPPT was implemented for providing Fig. 3. Transient HIL result for the transition of grid-connected to islanded
the total dc bus voltage reference, and the load demand mode of operation for the microgrid. From top to bottom: currents isa and
isb from the Slave NPC, and voltages vmca and vmcb from the Master NPC.
is used in order to obtain the current references. Also, a Vertical scale of 200 A/div and 500 V/div; Horizontal scale of 20 ms.
Synchronous Reference Frame Phase Locked Loop, SRF-PLL,
was implemented to synchronize the inverter to the grid [31].
Figure 2 shows the block diagram of the control strategy,
where it is clear that the main reference for the Master unit
is the voltage in the output LC-filter and the main reference
for the Slave unit is the power of the load-grid. The DC bus
voltages for each unit is controlled via the cost function (5)
and (9) for the Master and Slave units, respectively.

Master
vx* NPC converter
vmcx LC filter
FCS-MPC Sjxm Fig. 4. Transient HIL result for the transition of grid-connected to islanded
imx cost function
gm mode of operation for the microgrid. From top to bottom: currents isa and
vc1,vc2 12
ima from the Slave and Master NPCs, load current ila and voltage vmca
vc1,vc2
imx
Grid from the Master NPC. Vertical scale of 500 A/div and 500 V/div; Horizontal
vmcx
n
scale of 10 ms.
Slave
vmcx NPC converter
ix*
FCS-MPC Sjxs
isx cost function
vc1,vc2 gs 12
Load
resulting in λ1 = 200 and λ2 = 1 for the Master NPC, and
LC filter
vc1,vc2 λ1 = 50 and λ2 = 1 for the Slave NPC.
isx ilx
First, a simulation was performed where the microgrid is
Fig. 2. Block diagram of the Master-Slave control strategy.
connected to the grid via switch S1 , and with a three-phase
RL load where R = 1.03Ω and L = 3.3 mH. The illumination
of the PV panel is 1000 W/m2 for a temperature of 25o C. The
III. H ARDWARE - IN - THE -L OOP R ESULTS power is supplied to the load by the grid, while the current
In order to verify the performance of the proposed Master- of the Slave unit is zero, as can be seen in the left-side of
Slave MPC, the microgrid of Fig.1 was implemented in a Fig.3 that shows, from top to bottom, the currents isa and isb
Typhoon HIL 404 Hardware-in-the-Loop platform. The pa- from the Slave NPC, and the voltages vmca and vmcb from the
rameters of the simulation are: grid voltage amplitude of 220 Master NPC. Then, the switch S1 is opened, the microgrid is
VRMS and frequency of 50 Hz, total dc bus voltage vdc = disconnected from the main grid, and the Slave NPC provides
800 V (400 V for each capacitor), Lf = 3 mH, Cf = 300µF, the load currents in islanded mode. This can be seen in the
Rf = 0.012Ω and fs = 10 kHz. The BESS is composed of right-side of Fig.3, where we can notice that very fast transient
lithium-ion batteries with total nominal voltage of 800 V and response of the MPC controller of the Slave unit. Also, Fig.4
total capacity of 300 Ah. The rated load to be supplied in shows, from top to bottom and for phase a, the current of the
islanded mode is S = 100 kVA; therefore, both inverters are Slave and Master units, the load current and the voltage of
rated for this power, as the Slave unit is responsible for feeding the filter capacitor of the Master unit. Again we can see the
the load, while the Master unit is the grid-forming inverter. In fast transient response, and that the load current is properly
the case where the PV input is not able to fully feed the load, provided by the Slave NPC.
i.e. due to adverse climatic conditions, the Master unit with its In islanded mode, the illumination was changed from 1000
BESS performs load sharing or feeds the load entirely. Switch W/m2 to 500 W/m2 for a temperature of 25o C. Figs. 5 and 6
S1 is modeled here as an ideal switch. show the setady-state response for both conditions, where we
The MPC controllers of both NPCs, as well as other con- can see, from top to bottom and for phase a, the current of
trollers such as the SRF-PLL, MPPT, etc, were implemented in the Slave and Master units, the load current and the voltage of
the C Block of Typhoon HIL, that includes the implementation the filter capacitor of the Master unit. It can be seen that, even
delay of the processor. As a result, the cost functions were for a variation in the illumination, the Slave NPC is still able
calculated for the predicted variables at k+2. The weighting to properly provide the power demanded by the load. Finally,
factors for both MPCs were obtained with a heuristic approach, Fig.7 shows a condition where the illumination is not enough

Authorized licensed use limited to: Universidad Tecnica Federico Santa Maria. Downloaded on June 29,2023 at 20:11:00 UTC from IEEE Xplore. Restrictions apply.
 Fig. 8. Transient HIL result for islanded mode of operation with a step change
in the load. From top to bottom: currents isa and isb from the Slave NPC,
and voltages vmca and vmcb from the Master NPC. Vertical scale of 500
A/div and 500 V/div; Horizontal scale of 10 ms.
Fig. 5. Steady-state HIL result for islanded mode of operation with illumina-
tion of 1000 W/m2 for a temperature of 25o C. From top to bottom: currents
isa and ima from the Slave and Master NPCs, load current ila and voltage
vmca from the Master NPC. Vertical scale of 500 A/div and 500 V/div;
Horizontal scale of 10 ms.

Fig. 9. Transient HIL result for islanded mode of operation with a step change
from a linear to a nonlinear load. From top to bottom: currents isa and isb
from the Slave NPC, and voltages vmca and vmcb from the Master NPC.
Vertical scale of 500 A/div and 500 V/div; Horizontal scale of 10 ms.

and L = 7 mH was connected in parallel to the original

RL load. The transient response of the system is shown in
Fig. 6. Steady-state HIL result for islanded mode of operation with illumina- Fig.8, where we can see, from top to bottom, the currents
tion of 500 W/m2 for a temperature of 25o C. From top to bottom: currents isa isa and isb from the Slave NPC, and the voltages vmca and
and ima from the Slave and Master NPCs, load current ila and voltage vmca vmcb from the Master NPC. It is possible to note the very
from the Master NPC. Vertical scale of 500 A/div and 500 V/div; Horizontal
scale of 10 ms. fast transient response of both NPCs, as expected for MPC
controllers. The Slave unit is able to properly feed the load
for the two conditions, while, at the same time, the Master unit
maintains the voltage of the islanded microgrid at the desired
values of amplitude and frequency.
Another transient result in islanded mode is depicted in
Fig.8, presenting, from top to bottom, currents isa and isb
from the Slave NPC and voltages vmca and vmcb from the
Master NPC. The original RL load is changed abruptly to
a nonlinear load, composed of a three-phase rectifier with an
input L filter (L = 1 mH) and an output RL load with R = 5Ω
and L = 5 mH. Even though this is not a condition that would
be seen on a real application, it demonstrates again the very
fast transient response of both NPC inverters, as well as the
Fig. 7. Steady-state HIL result for islanded mode of operation with very low correct synthesis of the currents and voltages of, respectively,
illumination for a temperature of 25o C. From top to bottom: currents isa and the Slave and Master units. It is important to point our that,
ima from the Slave and Master NPCs, load current ila and voltage vmca
from the Master NPC. Vertical scale of 500 A/div and 500 V/div; Horizontal as can be seen in the models described in Section II for the
scale of 10 ms. Master and Slave units, the load is an unknown parameter for
the MPC controllers; therefore, these results show the robust
performance of the proposed Master-Slave MPC with respect
to make the Slave inverter operate. As a result, it is turned to uncertainties and variations on the system.
off (zero current at the output), and the Master NPC, besides
acting as a grid-forming inverter, also feeds the load. IV. C ONCLUSIONS
Also on islanded mode, a step change on the load was This paper proposed a Master-Slave MPC for microgrids.
applied, where a new three-phase RL load with R = 2Ω The Master is a grid-forming inverter, controlling the voltages

Authorized licensed use limited to: Universidad Tecnica Federico Santa Maria. Downloaded on June 29,2023 at 20:11:00 UTC from IEEE Xplore. Restrictions apply.
of the capacitors of the LC filter, while the Slave provides the [15] D. Xiao, K. S. Alam, M. Norambuena, M. F. Rahman, and J. Ro-
load currents. Both inverters must also control the voltages driguez, “Modified modulated model predictive control strategy for a
grid-connected converter,” IEEE Transactions on Industrial Electronics,
of their dc bus capacitors. Hardware-in-the-Loop results were vol. 68, no. 1, pp. 575–585, 2021.
provided for both on-grid and islanded operation, with RL and [16] X. Chen, W. Wu, N. Gao, H. S. H. Chung, M. Liserre, and F. Blaab-
non-linear loads, demonstrating the advantages of the proposed jerg, “Finite control set model predictive control for lcl-filtered grid-
tied inverter with minimum sensors,” IEEE Transactions on Industrial
MPC such as faster dynamic response, multi variable control, Electronics, vol. 67, no. 12, pp. 9980–9990, 2020.
adequate load sharing between the inverters, robustness to [17] J. Scoltock, T. Geyer, and U. K. Madawala, “Model predictive direct
parametric uncertainties and variations and the possibility to power control for grid-connected npc converters,” IEEE Transactions
on Industrial Electronics, vol. 62, no. 9, pp. 5319–5328, 2015.
directly include system constrains and non-linearities in the [18] M. G. Judewicz, S. A. González, J. R. Fischer, J. F. Martı́nez, and D. O.
controller implementation. Carrica, “Inverter-side current control of grid-connected voltage source
inverters with lcl filter based on generalized predictive control,” IEEE
ACKNOWLEDGMENT Journal of Emerging and Selected Topics in Power Electronics, vol. 6,
no. 4, pp. 1732–1743, 2018.
The authors would like to acknowledge the support of ANID [19] M. Aly, F. Carnielutti, A. Shawky, E. M. Ahmed, M. Norambuena,
S. Kouro, and J. Rodriguez, “Weighting factorless sequential model
through projects FB0008 and 1210208. predictive control method with fixed switching frequency for five-level
t-type photovoltaic inverters,” in IECON 2021 – 47th Annual Conference
R EFERENCES of the IEEE Industrial Electronics Society, 2021, pp. 1–6.
[20] Y. Shan, J. Hu, Z. Li, and J. M. Guerrero, “A model predictive control
[1] A. Alfergani, K. A. Alfaitori, A. Khalil, and N. Buaossa, “Control for renewable energy based ac microgrids without any pid regulators,”
strategies in ac microgrid: A brief review,” in 2018 9th International IEEE Transactions on Power Electronics, vol. 33, no. 11, pp. 9122–9126,
Renewable Energy Congress (IREC), 2018, pp. 1–6. 2018.
[2] Z. Zhang, O. Babayomi, T. Dragicevic, R. Heydari, C. Garcia, J. Ro- [21] T. Dragičević, “Model predictive control of power converters for robust
driguez, and R. Kennel, “Advances and opportunities in the model and fast operation of ac microgrids,” IEEE Transactions on Power
predictive control of microgrids: Part i–primary layer,” International Electronics, vol. 33, no. 7, pp. 6304–6317, 2018.
Journal of Electrical Power Energy Systems, vol. 134, p. 107411, 2022. [22] Y. Khayat, Q. Shafiee, R. Heydari, M. Naderi, T. Dragicevic, J. W.
[3] J. Hu, Y. Shan, J. M. Guerrero, A. Ioinovici, K. W. Chan, and Simpson-Porco, F. Dorfler, M. Fathi, F. Blaabjerg, J. M. Guerrero, and
J. Rodriguez, “Model predictive control of microgrids – an overview,” H. Bevrani, “On the secondary control architectures of AC microgrids:
Renewable and Sustainable Energy Reviews, vol. 136, p. 110422, 2021. An overview,” IEEE Transactions on Power Electronics, vol. 35, no. 6,
[4] O. Babayomi, Z. Zhang, T. Dragicevic, R. Heydari, Y. Li, C. Garcia, pp. 6482–6500, Jun. 2020.
J. Rodriguez, and R. Kennel, “Advances and opportunities in the model [23] A. B. Kunya and M. Argin, “Model predictive load frequency control
predictive control of microgrids: Part ii–secondary and tertiary layers,” of multi-area interconnected power system,” in 2018 IEEE Texas Power
International Journal of Electrical Power Energy Systems, vol. 134, p. and Energy Conference (TPEC), 2018, pp. 1–6.
107339, 2022. [24] A. Venkat, I. Hiskens, J. Rawlings, and S. Wright, “Distributed MPC
[5] Z. Chen, K. Wang, Z. Li, and T. Zheng, “A review on control strate- strategies with application to power system automatic generation con-
gies of ac/dc micro grid,” in 2017 IEEE International Conference on trol,” IEEE Transactions on Control Systems Technology, vol. 16, no. 6,
Environment and Electrical Engineering and 2017 IEEE Industrial and pp. 1192–1206, Nov. 2008.
Commercial Power Systems Europe (EEEIC / I CPS Europe), 2017, pp. [25] C. Ahumada, R. Cárdenas, D. Sáez, and J. M. Guerrero, “Secondary
1–6. control strategies for frequency restoration in islanded microgrids with
[6] M. Prodanovic, T. Green, and H. Mansir, “A survey of control methods consideration of communication delays,” IEEE Transactions on Smart
for three-phase inverters in parallel connection,” in 2000 Eighth Inter- Grid, vol. 7, no. 3, pp. 1430–1441, 2016.
national Conference on Power Electronics and Variable Speed Drives [26] G. Lou, W. Gu, Y. Xu, M. Cheng, and W. Liu, “Distributed mpc-based
(IEE Conf. Publ. No. 475), 2000, pp. 472–477. secondary voltage control scheme for autonomous droop-controlled
[7] M. L. R. Teodorescu and P. Rodriguez, “Grid converters for photovoltaic microgrids,” IEEE Transactions on Sustainable Energy, vol. 8, no. 2,
and wind power systems.” Wiley, 2011. pp. 792–804, 2017.
[8] J. G. J. Hu and S. Islam, “Model predictive control for microgrids - [27] J. Sachs and O. Sawodny, “A two-stage model predictive control strategy
from power electronics converters to energy management.” IET, 2021. for economic diesel-pv-battery island microgrid operation in rural areas,”
[9] J. Rodriguez, M. P. Kazmierkowski, J. R. Espinoza, P. Zanchetta, IEEE Transactions on Sustainable Energy, vol. 7, no. 3, pp. 903–913,
H. Abu-Rub, H. A. Young, and C. A. Rojas, “State of the Art of Finite 2016.
Control Set Model Predictive Control in Power Electronics,” IEEE Trans. [28] A. Parisio, E. Rikos, and L. Glielmo, “A model predictive control
Ind. Informat., vol. 9, no. 2, pp. 1003–1016, May. 2013. approach to microgrid operation optimization,” IEEE Transactions on
[10] S. Vazquez, J. Rodriguez, M. Rivera, L. G. Franquelo, and M. No- Control Systems Technology, vol. 22, no. 5, pp. 1813–1827, 2014.
rambuena, “Model predictive control for power converters and drives: [29] J. Ospina, N. Gupta, A. Newaz, M. Harper, M. O. Faruque, E. G. Collins,
Advances and trends,” IEEE Trans. Ind. Electron., vol. 64, no. 2, pp. R. Meeker, and G. Lofman, “Sampling-based model predictive control
935–947, Feb. 2017. of pv-integrated energy storage system considering power generation
[11] P. Karamanakos, E. Liegmann, T. Geyer, and R. Kennel, “Model forecast and real-time price,” IEEE Power and Energy Technology
predictive control of power electronic systems: Methods, results, and Systems Journal, vol. 6, no. 4, pp. 195–207, 2019.
challenges,” IEEE Open J. Ind. Appl., vol. 1, pp. 95–114, Aug. 2020. [30] Y. Shan, J. Hu, K. Wai Cheng, and M. Liu, “A universal model predictive
[12] M. Aly, F. Carnielutti, J. Rodriguez, M. Norambuena, S. Kouro, and control for practical ac microgrids with pvs and battery energy storage
A. K. Rathore, “Finite control set model predictive control without systems,” in 2018 IEEE Energy Conversion Congress and Exposition
weighting factors for common grounded five-level pv inverter,” in 2021 (ECCE), 2018, pp. 6257–6262.
IEEE 30th International Symposium on Industrial Electronics (ISIE), [31] R. Teodorescu, M. Liserre, and P. Rodriguez, Grid Converters for
2021, pp. 01–06. Photovoltaic and Wind Power Systems. Wiley-IEEE Press, 2007.
[13] C. R. D. Osório, D. A. Schuetz, G. G. Koch, F. Carnielutti, D. M.
Lima, L. A. M. Jr, V. F. Montagner, and H. Pinheiro, “Modulated model
predictive control applied to lcl-filtered grid-tied inverters: A convex
optimization approach,” IEEE Open Journal of Industry Applications,
vol. 2, pp. 366–377, 2021.
[14] S. R. Mohapatra and V. Agarwal, “Model predictive controller with re-
duced complexity for grid-tied multilevel inverters,” IEEE Transactions
on Industrial Electronics, vol. 66, no. 11, pp. 8851–8855, 2019.

Authorized licensed use limited to: Universidad Tecnica Federico Santa Maria. Downloaded on June 29,2023 at 20:11:00 UTC from IEEE Xplore. Restrictions apply.