Zhou et al.

Protection and Control of

Protection and Control of Modern Power Systems (2023) 8:64
https://doi.org/10.1186/s41601-023-00337-3 Modern Power Systems

ORIGINAL RESEARCH Open Access

Discrete space vector modulation

and optimized switching sequence model
predictive control for three‑level voltage source
inverters
Sheng Zhou1,3, Minlong Zhu1, Jiaqi Lin1,2, Paul Gistain Ipoum‑Ngome1, Daniel Legrand Mon‑Nzongo1 and
Tao Jin1,2*   

Abstract
This paper proposes a discrete space vector modulation and optimized switching sequence model predictive control‑
ler for three-level neutral-point-clamped inverters in grid-connected applications. The proposed strategy is based
on cascaded model predictive control (MPC) for controlling the grid current while maintaining the capacitor voltage
balanced without weighting factor. To enhance the closed-loop performance, the external MPC evaluates 19 basic
and 138 virtual vectors (VV) of the proposed space vector method. The optimal control voltage is then selected using
an extended deadbeat method to reduce the execution time of the proposed control algorithm. By using the dis‑
crete space vector modulation principle, the VV are synthesized based on switching sequence (SS) and are divided
into negative and positive SSs considering their impact on the neutral point (NP) potential. The inner MPC evaluates
both types of SSs and selects the one that keeps the capacitor voltage balanced. Various controllers are evaluated
and compared against the proposed control strategy. The results show that the proposed strategy improves perfor‑
mance without weighting factor, while maintaining a total harmonic distortion of current to be less than 2%. Com‑
pared to the modulated MPC which provides the same fixed switching frequency, the proposed controller reduces
the computational burden by over 50% while also providing better NP voltage balance accuracy.
Keywords Three-level inverter, Fixed switching frequency, Model predictive control (MPC), Optimal switching
sequence (OSS), Discrete space vector modulation (DSVM)

1 Introduction As numerous new energy generation systems and flex-

Energy shortage and environmental pollution have ible AC transmission devices are integrated into the
become critical concerns. This has led to an increased grid, inverters have become an indispensable part of
focus on the development of renewable energy sources. energy conversion systems [1]. In comparison to two-
level inverters, the 3L-NPC voltage source inverters boast
advantages such as lower output harmonics and reduced
*Correspondence: semiconductor voltage stress [2, 3]. Consequently, they
Tao Jin have been widely adopted in many product lines in
[email protected]
1
College of Electrical Engineering and Automation, Fuzhou University, renewable energy systems. In addition, the advance of
Fuzhou 350116, China microprocessor technology has enabled the implemen-
2
Fujian Key Laboratory of New Energy Generation and Power Conversion, tation of novel and computationally intensive control
Fuzhou 350116, China
3
State Grid Fujian Electric Power Co., LTD, Fuzhou 350001, China algorithms for power electronic topologies and electri-
cal drives, such as predictive control [4]. Such control

© The Author(s) 2023. Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which
permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the
original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or
other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line
to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory
regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this
licence, visit http://creativecommons.org/licenses/by/4.0/.
Zhou et al. Protection and Control of Modern Power Systems (2023) 8:64 Page 2 of 16

algorithms often have higher computational requirement applicability. An effective method to resolve the issue of
than the traditional PI-type controllers [5, 6]. Among variable switching frequency in MPC strategies is to seek
these control strategies, the most prominent ones include an optimal switching sequence (OSS) instead of a sin-
deadbeat, lag-based, trajectory-based and model predic- gle switching state per control period. In [27], six local
tive control. OSSs are considered and evaluated in the power control
For classic finite control-set model predictive control objectives to determine the global OSS for the next sam-
(FCS-MPC) [7], optimal control actions are obtained by ple, whereas in [28], OSS-MPC based on two cost func-
predicting system behavior and evaluating cost functions tions is proposed to independently control current and
over all possible states of the converter [8, 9]. FCS-MPC capacitor voltages without WF. However, this method
offers advantages such as fast dynamic response, simplic- needs to compute the solution of the relaxation problem
ity in handling nonlinearity and constraints, and a multi- first, followed by the use of non-negative constraints to
variable control approach [10, 11]. Nevertheless, its main solve the OSS, resulting in a high computational bur-
drawbacks include the short sampling time requirement, den. Li et al. [29] introduces the use of a cost function to
variable switching frequency and high computation define the region with the OSS candidates for evaluation,
time. These limit its application in multi-level convert- thereby reducing the computational burden. However,
ers [12, 13]. To apply to three-level inverters, the classic the execution time is still quite high because of the need
FCS-MPC needs to evaluate 27 virtual vectors (VV) in to calculate the duty cycle corresponding to the OSS. To
each control cycle, resulting in poor THD performance further reduce the execution time, the deadbeat control
and significant current ripple [14]. It also entails a high technique is proposed to select the required control volt-
computational workload, posing a challenge for its imple- age without evaluation of the voltage control objective
mentation on DSP/FPGA control hardware. In the case [30, 31]. In [32], deadbeat-predictive torque control with
of 3L-NPC inverters, MPC employs current and capaci- discrete space-vector modulation is proposed to reduce
tor voltage control objectives for closed-loop control, the torque ripple and the computational burden of the
whose performance is influenced by the weighting factors conventional predictive torque control method. Never-
(WFs) [15, 16]. The selected WFs establish the trade-off theless, a control method that can achieve high control
between current accuracy and capacitor voltage control, precision while simultaneously addressing the challenges
with higher WF values reducing current accuracy and related to the absence of WF, fixed switching frequency,
lower values increasing neutral point (NP) voltage imbal- and low computational burden still requires further
ance. WF selection often relies on empirical methods, research and development.
incurring a significant amount of time. Also, classic FCS- In this paper, a discrete space vector modulation and
MPC results in variable switching frequency, which com- optimized switching sequence model predictive control-
plicates the design of filters [17, 18]. ler (DSVM-OSS-MPC) strategy is proposed. The main
Many researchers have made improvements to the contributions of this paper include the following three
FCS-MPC strategy to address the above issues. Regard- aspects:
ing the tuning of WFs, references [19] and [20] use
fuzzy methods and neural networks to obtain the best (1) Controlling grid current based on the cascade MPC
WF solution for each operating condition, respectively. while maintaining capacitor voltage balance with-
In [21], a fast finite switching state MPC without WF is out WF, thus eliminating the cost of WF selection.
proposed, where the selected voltage vectors are used for (2) Achieving superior current tracking accuracy and
tracking the current reference and the redundant vectors NP voltage balance. To improve the closed-loop
for balancing the DC capacitance voltage. However, the performance, DSVM is used to achieve a new space
method is less effective because of the limited number of vector with 157 voltage vectors instead of the 19
states used to control NP voltage. basic vectors used in classic MPC. The additional
To address the issue of variable switching frequency 138 vectors are virtual and synthesized using the
in the MPC strategies, a digital filter is employed in OSS which considers the impact of each vector on
[22–24] to narrow the switching frequency to a specific NP voltage and the reduction of switching commu-
range. In [25], an artificial intelligence method is pro- tation.
posed for online tuning of the WFs and regulating the (3) Significantly reduced computational burden is
average switching frequency. Reference [26] proposes achieved. To avoid the exhaustive search for the
a modulated MPC ­(M2PC) strategy to realize capacitor optimal control solution among the 157 SS candi-
voltage balance by controlling the duty cycle of redun- dates, the extended deadbeat method is used in the
dant vectors. However, this method has the drawback of outer MPC to reduce the closed-loop control to a
imposing a high computational burden and is of limited sub-optimal problem. The inner MPC then focuses
Zhou et al. Protection and Control of Modern Power Systems (2023) 8:64 Page 3 of 16

on the redundant SSs to select the optimal solution. From the modeling approach described in [33] and
Therefore, the proposed control strategy offers the considering the variables given in Table 1, the grid cur-
benefit of delivering accurate current response and rents, and the sum and difference in the dynamics of
maintaining capacitor voltage balance, while simul- the DC-link voltage in the αβγ reference frame can be
taneously exhibiting characteristics of no WF influ- expressed as:
ence and reduced computational burden.
1 √1 Sα Sγ
d 1 √ Sβ2 − Sα2 −
L iαβ = x1 Sαβ + 2 6 3 x2 − eαβ − Riαβ
The rest of this paper is organized as follows: the clas- dt 2 Sα Sβ
√
6
−
Sβ Sγ
√
3
sic MPC method is presented in Sect. 2, while Sect. 3 (1)
introduces the basic principles of switching sequence
T
MPC. Section 4 provides a detailed description of the C ẋ1 = −Sαβ iαβ + 2idc (2)
proposed DSVM-OSS-MPC, including the extension
of the space vector based on the DSVM and OSS, the 2   1 T
selection method of the optimum voltage vector, and the C ẋ2 = − √ Sα2 − Sβ2 , −Sα Sβ iαβ − √ Sαβ iαβ Sγ
6 6
optimization of the capacitor voltage balance. In Sect. 5, (3)
experiments and simulations are carried out to verify the
where x1 = vC1 + vC2 and x2 = vC1—vC2 are the sum and
effectiveness and superiority of DSVM-OSS-MPC. The
the difference of the upper and lower DC-link voltages,
conclusions are given in Sect. 6.
respectively.
To ensure correct converter operation, x2 must be near
2 Classic FCS‑MPC strategy for 3L‑NPC
zero or at least one order of magnitude lower than x1
Figure 1 presents a 3L-NPC inverter connected to the
[33]. In addition, the averaged duty cycle Sαβ is defined
grid. The DC-side of the converter consists of two capaci-
within [ − 1, 1], and therefore the third term located
tors supplied by a DC voltage vdc. At the AC-side, the
on the right-hand side of (1) can be assumed to be two
converter is connected to the three-phase sources (ea,
orders of magnitude lower than the second term. With
eb, ec) through the RL filter and injects three-phase cur-
this consideration, the inductor current dynamics can be
rents (ia, ib, ic) to the grid. Each phase of the three-level
approximated by:
NPC consists of four switches Sxi with x ∈ {a, b, c} and
1 ≤ i ≤ 4 , and generates up to three switching levels d
Sx ∈ {−1, 0, 1}. Considering the three legs, the inverter L iαβ = −eαβ + u − Riαβ (4)
dt
can generate 27 switching states Sjabc.
with 1 ≤ j ≤ 27. To obtain the optimal state for the where u = 12 x1Sαβ is the output voltage vector of the
next sample with the MPC, the dynamic response of the inverter, iαβ is the output current vector, and eαβ is the
system due to each switching state candidate Sjabc is pre- grid voltage vector. From [34], the current predictions
dicted and evaluated. at k + 1 in the α− axis and β− axis which are noted as
p p
iα (k + 1) and iβ (k + 1), are given by:
 p      
iα (k + 1) RTs iα (k) Ts uα (k) − eα (k)
p = 1 − +
iβ (k + 1) L iβ (k) L uβ (k) − eβ (k)
(5)
where i(k) is the measured current at the k sample, while
e(k) and u(k) are the measured grid voltage and inverter
output voltage at the k sample, respectively.

Table 1 System variables and parameters

Variable Description

eαβ = {eα, eβ}T Grid voltage vector in αβγ reference frame

iαβ = {iα, iβ}T Inductors currents in αβγ reference frame
Sαβγ = {Sα, Sβ, Sγ}T Averaged duty cycles in αβγ reference frame
vC1, vC2 DC-link capacitors voltages
Fig. 1 Grid-connected 3L-NPC converter and flow diagram L, C, R Smoothing inductor, DC-link capacitor
of the classic FCS-MPC strategy and Line resistance
Zhou et al. Protection and Control of Modern Power Systems (2023) 8:64 Page 4 of 16

For NP voltage balance, the capacitor voltages vc1(k + 1)

and vc2(k + 1) related to the capacitors C1 and C2 at the
k + 1 sample are predicted as:
� p � � �  Ts 
vc1 (k + 1) vc1 (k) C1 i c1 (k)
p
= +
Ts
 (6)
vc2 (k + 1) v c2 (k) i c2 (k)
C2

where vc1(k) and vc2(k) are the measured capacitor volt-

ages at the k sample. ic1(k) and ic2(k) are the currents
flowing through the capacitors at the k sample and are
given by:

ic1 (k) = idc (k) − H 1x ix (k)
x=a,b,c
(7)
Fig. 2 Grid-connected 3L-NPC converter and flow diagram

ic2 (k) = idc (k) + H 2x ix (k)
of the SS-MPC control strategy
x=a,b,c
 
1, if Sx = 1 1, if Sx = −1
where H1x = , and H2x = .
0, otherwise 0, otherwise Maintaining the capacitor voltages to be balanced
The standard cost function for tracking the current requires the control of the NP voltage O, which is defined
reference and regulating the capacitor voltage balance is as vo = vc2-vc1. Considering that the system is balanced
defined in [35], given by: and the DC-link voltage is constant, the dynamic of vo is
2  p
2 given by:
g = iα∗ (k + 1) − iαp (k + 1) + iβ∗ (k + 1) − iβ (k + 1)

dvo (t) 1
 p p
+ dc vc1 (k + 1) − vc2 (k + 1)
2 = io (t) (9)
dt C
(8)
The NP current io is obtained by:
where iα (k + 1) and iβ (k + 1) are the current reference
∗ ∗

components at instant k + 1, and λdc is the WF. For time  T

io = Sjabc  iabc , (10)
delay compensation [36], the evaluation of (8), (6), and
(5) is considered at k + 2 rather than k + 1.
      T
where Sjabc = Sja  Sjb  Sjc  , and iabc = [ia , ib , ic ]T.

To obtain good performance when using (8), an appro- From Fig. 3, the analysis of the 27 switching state can-
priate trade-off needs to be achieved between tracking didates of the three-level NPC is divided into four cate-
current and balancing capacitor voltage. Since λdc is a gories of vectors: zero using black dots; small using red
function of the operating point and a parameter of the dots; medium using blue dots; and larger using green
system, the design of λdc is not trivial [15]. In addition,
the unified cost function which provides a single optimal
solution does not guarantee that both individual control
objectives are optimized [37]. An alternative method to
simultaneously track the current and control the capaci-
tor voltages without the need for WF is to use a cascaded
MPC approach [15, 37].

3 Basic principle of the SS‑MPC

The cascaded SS-MPC approach is proposed in [28] for
controlling the grid current and the capacitor voltages of
the three-level inverter without the use of WF, as shown
in Fig. 2. The outer MPC determines the suboptimal SS
candidates which satisfy the current objective, whereas
the inner MPC selects between the suboptimal candi-
dates, with the SS ensuring capacitor voltage balance.
Fig. 3 Grid-connected 3L-NPC converter space vector diagram
Zhou et al. Protection and Control of Modern Power Systems (2023) 8:64 Page 5 of 16

Table 2 Classification of small vectors total number of possible dispositions of switching states
Type Switching states (Sabc)
in an SS per region increases with the number of redun-
dant vectors. To regroup the SS in each region into two
Sabc.P [POO], [PPO], [OPO], [OPP], [OOP], [POP] types, the nature of an SS is exclusively defined by the
Sabc.N [ONN], [OON], [NON], [NOO], [NNO], [ONO] type of its small vectors. For example, in region 1-IV, if
a P-type small vector [PPO] is used, the number of pos-
sible dispositions of P-type SS are [PON-PPN-PPO],
dots. Except for the zero-voltage vectors, only the small [PON-PPO-PPN] [PPO-PPN-PON], [PPO-PON-PPN],
vectors can connect the inverter terminals to the same [PPN-PPO-PON] and [PPN-PON-PPO]. In this work, to
type of DC-link potential. A switching state Sjabc is called reduce the switching losses within a control period, the
P-type and noted as Sabc.P when only the positive terminal switching effort is restricted. In the first restriction, the
of the DC link is connected to the grid, and N-type Sabc.N switching state per phase (Sx) cannot change between P
when connected to the negative terminal. The classifica- and N and vice-versa. In the second restriction, only one
tion of small vectors according to the type of switching phase of an applied three-phase switching state (Sabc) can
states Sabc.P and Sabc.N is given in Table 2. Using (10), the change. In such a case, the candidate SSs in subsector
direction of NP current can be defined. 1-IV are PON-PPN-PPO for the P-type and PPN-PON-
For example, by applying the P-type [O,P,O]T and its OON for the N-type. The same principle is used for all
associated N-type redundancy [N,O,N]T, the associ- the SSs of sector 1 as given in Table 3.
ated NP current according to (10) are io = ib and io = -ib, However, the type of the optimal SS, which is applied
respectively. With an appropriate distribution of several between two consecutive samples and selected by the
switching combinations within the sampling period, it inner MPC, can change between the P-type and N-type.
is possible to control the current as well as NP voltage For example, considering the previously applied SS of
potential. 1-II-P, the next optimal candidate SS selected by the
The order of the converter applying several Sjabc within inner MPC is either 1-II-P or 1-II-N if the required con-
the control period is known as SS and the controller is trol voltage is in sector 1-II. Between two consecutive
called SS-MPC. In PWM and SVM modulation meth- samples, two phases of the inverter can change. In this
ods, various SS dispositions are reported in the literature. scenario, the second restriction is not respected, which
Since SVM is synthesized based on the space vector, it is will lead to an extra switching effort.
easier to implement SS-MPC based on SVM than PWM, To synthesize the applicable OSS during each sam-
and it provides a simpler identification of small voltage ple, the conduction time associated with each switch-
vectors to control the NP voltage. ing state within a control period has to be calculated.
To synthesize SVM, the space vector presented in Fig. 3 Various online methods are proposed for obtaining
is divided into six sectors, and each sector is further the OSS and the related duty cycle as a function of the
divided into four regions or sub-sectors. For a 3L-NPC, resulting current and power errors of the primary term
the total number of regions is 24 and each region has of the cost function [26–29]. Even though they provide
three switching states with one or two small vectors. For an optimal duration candidate for tracking the control
instance, region II within sector 1, noted as 1-II, has one objectives, these methods result in a higher computa-
medium and two small vectors. Considering SVM based tional burden.
on a symmetric pulse pattern, an SS applied over a con-
trol period is given by:
    
t1 t2
   
t2 t1 Table 3 The switching sequence of sector 1
S  u1 , u2 , u3 [t3 ], u2 , u1 (11) Sector Switching sequence type Switching sequence
2 2 2 2
1-I P OOO-POO-PPO
where ui (i = 1, 2, 3) is the vector related to the ith switch-
N ONN-OON-OOO
ing state of a subset. ti (i = 1, 2, 3) is the duration of the ith
1-II P PON-POO-PPO
switching state in a subsection and satisfies the following
N ONN-OON-PON
formula:
1-III P PNN-PON-POO
t1 + t2 + t3 = Ts (12) N ONN-PNN-PON
1-IV P PON-PPN-PPO
Focusing on half-sampling time, several dispositions
N OON-PON-PPN
of SS are possible using the vectors u1, u2, and u3. The
Zhou et al. Protection and Control of Modern Power Systems (2023) 8:64 Page 6 of 16

4 Proposed DSVM‑OSS‑MPC strategy configuration, the current THD, the average switching
An offline approach provides an alternative solution frequency (ASF), and the complexity of implementing
to tune the SS candidates without increasing the com- MPC when nv increases are considered. To select the
putational burden of the MPC algorithm. In this paper, suitable expanded region, the MPC with 6, 10, 19, 28
DSVM based on virtual vectors is used to synthesize the vectors per region are compared with MPC-PWM, as
OSS without the need for the online evaluation of the shown in Fig. 5. As seen, when nv = 6, the resulting ASF
related switching durations. is lower than that of MPC-PWM, and for a closed-loop
control with the lowest ASF and complexity, the suit-
4.1 DSVM based on virtual vectors able configuration is nv = 6. However, the resulting cur-
To improve controller performance, the number of vir- rent THD is higher than that under MPC-PWM for the
tual vectors is selected so that the closed-loop perfor- different operating currents.
mance is similar to the one achieved under MPC-PWM To achieve a similar current THD as under MPC-
[38]. The virtual vector noted uv is synthesized by its cor- PWM, the switching frequency needs to be increased.
responding SSs as given by: However, this can be challenging to implement in a
 low-cost digital processor because of the high com-
uv = di ui
(13) putational requirement. For nv = 10, 19, and 28, the
i=1,2,3 resulting current THD and ASF are similar to the val-
ues under MPC-PWM. Thus, to obtain performances
d1 + d2 + d3 = 1 (14) that are equivalent to the ones achieved under MPC-
PWM, the possible candidates are nv = 10, 19, and 28.
where ui is the basic vector and uv is the virtual vec- Considering that for nv ∈ {10, 19, 28}, the improvement
tor. di is the duty cycle of vector ui calculated as di = ti / of the current THD and the reduction of the ASF are
Ts. By substituting the coordinates of the three basic and negligible, MPC-DSVM with nv = 10 represents the sce-
virtual vectors in (13), Eqs. (13) and (14) are developed nario for achieving a similar closed response to that of
into a system of three equations. The resulting system is MPC-PWM, without imposing excessive implementa-
solved offline for obtaining the di associated with each ui. tion complexity.
To expand the number of voltage vectors of sector Considering the case with nv = 10, the expanded space
1-II from 3 vectors to 6, 10, 19, and 28, the region is fur- vector is presented in Fig. 6a. Focusing the analysis on
ther divided as presented in Fig. 4. In [36], the expanded the subsector 1-II for instance, the number of virtual vec-
region consists of three actual and three virtual vectors, tors synthesized based on three basic vectors is equal to
as shown in Fig. 4a. To select the appropriate extended seven as presented in Fig. 6b, and each virtual vector is
synthesized according to Table 4. Knowing that a subsec-
tor has two types of SS (as seen in Table 3), each virtual
vector uv can be decomposed into P-type and N-type SSs
as illustrated in Fig. 7. Hence, the expanded space vec-
tor presented in Fig. 6a has a total of 157 vectors divided
into 19 basic vectors and 138 virtual vectors. Since the
total number of candidate vectors is 6 times the number
of states generated by 3L-NPC, it is anticipated that the

Fig. 4 Subsector 1-II virtual vector arrangement. a nv = 6. b nv = 10. c

nv = 19. d nv = 28 Fig. 5 Current THD and switching frequency change with nv
Zhou et al. Protection and Control of Modern Power Systems (2023) 8:64 Page 7 of 16

Fig. 7 Type of switching sequences: a P-type switching sequence

of sub-sector 1- II. b N-type switching sequence of sub-sector 1- II

2  2 2
g = u∗α − uvα + u∗β − uvβ + dc vop (k + 1)
 

(15)
u∗α (k) iα∗ (k
       
L + 1) − iα (k) iα (k) eα (k)
=
u∗β (k) Ts + 1) − iβ (k) + R iβ (k) + eβ (k)
iβ∗ (k
(16)
where uαβ (k) is the reference voltage components, and vo
∗

is the NP voltage potential given by:

Fig. 6 Proposed extended space vectors. a The distribution map 1 
3
of virtual vectors in the expanded space vector. b candidate basis vop (k + 1) = ti (|Sia |ia + |Sib |ib + |Sic |ic ) + vo (k)
and virtual vectors in a region 1-II C
i=1
(17)
Using the cascaded MPC method, the first primary
Table 4 The synthesis method of virtual vector term of (15) is used by the outer MPC as given by:
Virtual Synthesis method Virtual Synthesis method 2  2
vector vector gouter = u∗α − uvα + u∗β − uvβ (18)


uv d1us1 + d2us2 + d3um1 uv4 1 1 1

3 us1 + 3 us2 + 3 um1
The evaluation of (18) requires 157 cycles of calcula-
uv1 2 1 1 uv5 1 2 1
3 us1 + 6 us2 + 6 um1 6 us1 + 3 us2 + 6 um1 tion which is difficult to achieve, especially with stand-
uv2 1 1 uv6 1 1 2
2 us1 + 2 us2 6 us1 + 6 us2 + 3 um1 ard digital control processors. To obtain the optimal
uv3 1 1 uv7 1 1
vector with the lowest computational requirement,
2 us1 + 2 um1 2 us2 + 2 um1
an extended deadbeat method is developed. The main
idea is to define the boundaries associated with each
computation time will be excessive when using the clas- candidate voltage vector in the expanded space vec-
sic optimization approach. To implement a control algo- tor, and use the coordinates of the reference to identify
rithm with a reduced computational burden, it is crucial the region which is associated with the optimal control
to reduce the optimization problem to evaluate only the action.
SS candidates which satisfy the optimal control voltage. Considering sector 1 in Fig. 6a, the vectors are pro-
jected in the reference formed by the three axes L1,
4.2 Outer and inner MPCs L2, and L3 as illustrated in Fig. 8. As seen, L1 is paral-
To obtain the OSS control action for the next sample, the lel to the axis noted as [ONN-PPN], L2 is parallel to
cost function given in (8) is developed as a function of the axis noted as [OOO-PON], and L3 is parallel to
voltage control objectives. Using the unified optimization the axis noted as [PNN-PPO]. It can also be seen from
method, the cost function is given by: Fig. 8 that each virtual vector is the center of a smaller
Zhou et al. Protection and Control of Modern Power Systems (2023) 8:64 Page 8 of 16

Fig. 9 Type of subsectors or regions: a A type. b B type

the second is to find the corresponding smaller hexa-

gon. For the first step, each small sector is defined by
Fig. 8 Schematic diagram of optimized voltage vector selection the boundary conditions in three axes. For example,
the voltage reference is within the small sector 1-II, if
L1 ≤ 6, L2 ≥ 6, and L3 ≥ -6. The second step is to deter-
hexagon, i.e., the region defined by each small hexa- mine the small hexagon with the small vector adjacent
gon corresponds to a given optimal voltage vector. to the vector reference. It should be noted that two dis-
Assuming that the reference voltage vector is u* as rep- positions of small sectors are possible in a sector, and
resented by the light blue vector in Fig. 8, the virtual the distribution of virtual vectors depends on the type
vector uv which is the center of the light blue hexagon of disposition as presented in Fig. 9. The error, L ,
is the optimal control voltage. between the reference voltage and the center of a given
In general, the coordinates of the reference vector are small sector is defined by:
given by:      
√ �L1 L1 (u∗ ) L1 (ucenter )
12 3u∗β (k)  �L2  =  L2 (u∗ )  −  L2 (ucenter )  (23)
L1 = (19) �L3 L3 (u∗ ) L3 (ucenter )
Vdc

√ ∗ where L1(u*), L2(u*), and L3 (u*) are the coordinates of u*

18u∗α (k) + 6 3uβ (k) on L1, L2, and L3, respectively. L1(ucenter), L2(ucenter), and
L2 = (20) L3(ucenter) are the coordinates of ucenter on L1, L2, and L3
Vdc
respectively. When the reference voltage is adjacent to an
√ actual vector, ux, uy or uz, the new center used to evalu-
6 3u∗β (k) − 18u∗α (k)
L3 = (21) ate (23) is associated with a small sector and is given by
Vdc ucenter = (ux + uy + uz)/3. In summary, the rules for select-
ing the optimal control voltage are given in Table 5.
In the case where the reference vector u* exceeds the
Taking u* represented in Fig. 8 as an example, with the
maximum modulation range, u* is scaled back into the
coordinates L1 = 3.6, L2 = 7.3, and L3 = − 4.5, u* belongs
converter operating region as illustrated by the green
to sector 1− II which is an A-type disposition. The adja-
vector in Fig. 8 and is given by:
cent center to the reference control voltage is obtained
by rounding up u* and the resulting coordinates are (4,

u∗ (k) ||umax |
 ∗ 
∗ u (k)|
∗ , u (k) > |umax |
u (k) = ∗ (22) 8, -4). With the coordinates of u* and ucenter, L is (-0.4,
u (k) otherwise
-0.7, -0.5). From Table 5, the optimal voltage vector is uv4,
which is synthesized by applying ux, uy, and uz during the
where |umax | is the module of the maximum voltage
duty cycles 1/3, 1/3, and 1/3, respectively.
vector.
The optimal voltage control provided by the outer MPC
With the coordinates of the reference voltage, two
is used by the inner controller for the NP voltage balance.
steps are needed to determine the optimal control volt-
The cost function of the inner loop is defined as:
age. The first step is to localize the smaller sector and
Zhou et al. Protection and Control of Modern Power Systems (2023) 8:64 Page 9 of 16

Table 5 Rules for selecting the optimal control voltage (1) Measure iαβ(k), eαβ(k), vc1(k) and vc2(k).
The output boundaries Duty cycle
(2) Apply the optimal switching sequence.
voltage vector (3) Predict iαβ(k + 1), eαβ(k + 1), iαβ
∗ (k + 2), and u∗ (k + 2).
αβ
A type B type dux duy duz (4) Calculate L1, L2, and L3 associated to u∗αβ(k + 2).
ux ΔL1 ≤ − 3 ΔL2 ≤ − 3 1 0 0 (5) Use the coordinates of u∗αβ(k + 2) to select uαβ(k + 2).
uy ΔL3 ≥ 3 ΔL3 ≤ − 3 0 1 0 (6) Select N- or P-type SS which minimizes ­ginner.
uz ΔL2 ≥ 3 ΔL1 ≥ 3 0 0 1
uv1 − 3 < ΔL1 ≤ − 1 ΔL1 ≤ 0 2/3 1/6 1/6 5 Simulation and experimental results
ΔL2 ≤ 0 − 3 < ΔL2 ≤ − 1 For validation purposes, the effectiveness of the proposed
ΔL3 ≤ 0 0 < ΔL3 controller is compared with the classic MPC [33], the
uv2 ΔL1 ≤ 0 ΔL1 ≤ − 1 1/2 1/2 0 FS-MPC without WF (WMPC) [21], and the M ­ 2PC [26].
ΔL2 ≤ − 1 ΔL2 < 0
0 < ΔL3 ΔL3 ≤ 0 The parameters of the system used for the evaluation are
uv3 ΔL1 ≤ 0 0 < ΔL1 1/2 0 1/2 given in Table 6 and the controller operating sampling
0 < ΔL2 ΔL2 ≤ 0 frequency is 10 kHz.
ΔL3 ≤ − 1 1 < ΔL3
uv4 − 1 < ΔL1 < 1 − 1 < ΔL1 < 1 1/3 1/3 1/3
− 1 < ΔL2 < 1 − 1 < ΔL2 < 1
− 1 < ΔL3 < 1 − 1 < ΔL3 ≤ 1 Table 6 System parameters
uv5 0 < ΔL1 ΔL1 ≤ 0 1/6 2/3 1/6 Parameter Grid-Connected
ΔL2 ≤ 0 0 ≤ ΔL2
1 ≤ ΔL3 < 3 − 3 ≤ ΔL3 ≤ − 1 Simulation Experimental
uv6 0 < ΔL1 1 ≤ ΔL1 < 3 1/6 1/6 2/3
1 ≤ ΔL2 < 3 0 < ΔL2 Vdc DC-link voltage 800 V 110 V
ΔL3 ≤ 0 0 < ΔL3 Em Line-line voltage (RMS) 380V 50 V
uv7 1 ≤ ΔL1 0 < ΔL1 0 1/2 1/2 C DC link capacitors 500 μF 2200 μF
0 < ΔL2 1 ≤ ΔL2 f Fundamental frequency 50 Hz 50 Hz
0 < ΔL3 ΔL3 ≤ 0
R Resistance 0.1 Ω 0.5 Ω
Iref Reference current 15/30 A 3/6 A
L Filter inductance 5 mH 7 mH
 3
2
1 
ginner = ti (|Sia |ia + |Sib |ib + |Sic |ic ) + vo (k)
C
i=1
(24)
The optimal vector can be either a basic or virtual vec-
tor, and so is selected from the 157 vectors. In the case
where it is a virtual vector, the inner MPC selects the type
of SS that ensures a better NP voltage balance. The pro-
posed DSVM-OSS-MPC uses the whole extended space
vector with 157 compared to 19 for the classic MPC, and
therefore a better current accuracy can be achieved.
The block diagram of the proposed strategy is shown
in Fig. 10 and the algorithm is described by the follow-
ing main steps.

Fig. 11 Comparative evaluation of the current response. a Classic

Fig. 10 DSVM-OSS-MPC control block diagram MPC. b WMPC. c ­M2PC. d DSVM-OSS-MPC
Zhou et al. Protection and Control of Modern Power Systems (2023) 8:64 Page 10 of 16

5.1 Simulation results and discussion

The comparative evaluation of the four control strategies
with a current step change from 15 to 30A is presented
in Figs. 11, 12. As seen from Fig. 11 ­M2PC provides the
fastest response with a time response ts(c) = 0.72 ms, i.e.
faster than ts(d) = 0.83 ms with the proposed controller.
However, compared to the respective response times of
ts(a) = 1.57 ms and ts(b) = 1.64 ms with MPC and WMPC,
the proposed DSVM-OSS-MPC presents a faster
dynamic response.
From Fig. 12, with both 15 A and 30 A operating cur-
rents, the NP voltage with classic MPC and WMPC is
higher with a value equal vo(a) = 9.2 V and vo(b) = 8.2 V,
respectively. While the proposed strategy and ­M2PC
ensure a better balanced capacitor voltage with the
peak NP voltages of vo(c) = 5.3 V and vo(d) = 5.2 V,
respectively.
To provide a fair comparison on the steady-state
performances of different control methods, the four
controllers are operated at the same ASF. The ASF of a
3L-NPC inverter is defined as
4 Fig. 13 Comparative evaluation of the steady state current response.
1   a Classic MPC. b WMPC. c ­M2PC. d DSVM-OSS-MPC
ASF = ASFni (25)
12
n=a,b,c i=1

where ­ASFni denotes the switching times of the ith IGBT

of n-phase in one second.
The comparative evaluation of the four control strat-
egies is made with a similar resulting ASF (2 kHz)
and the results are presented in Figs. 13 and 14,
where the sampling frequencies of the classic MPC,
WMPC, ­M2PC and DSVM-OSS-MPC are 15 kHz,
15 kHz, 6 kHz, and 6 kHz, respectively. It can be
noted that the four control strategies provide simi-
lar output current THD when operating at a similar
average switching frequency. The current THD with
classic MPC, WMPC, M ­ 2PC and DSVM-OSS-MPC Fig. 14 Current spectrum. a Classic MPC. b WMPC. c ­M2PC. d
are ­THD(a) = 2.15%, ­THD(b) = 2.16%, ­THD(c) = 2.32%, DSVM-OSS-MPC

and ­THD(d) = 2.31%, respectively. However, compared

to MPC and WMPC, both the proposed controller and
­M2PC strategies operate at a fixed switching frequency
of 6 kHz.

5.2 Experimental results and validation

The simplified diagram and a picture of the experimental
set-up are presented in Fig. 15. The converter parameters
and the grid voltage are given in Table 6. The different
control algorithms are implemented in a real-time plat-
Fig. 12 Comparative evaluation of the NP voltage. a Classic MPC. b form and further details can be found in [34].
WMPC. c ­M2PC. d DSVM-OSS-MPC
Zhou et al. Protection and Control of Modern Power Systems (2023) 8:64 Page 11 of 16

1.704 kHz, 4.330 kHz, and 3.815 kHz, respectively. The

operating average switching frequencies with ­M2PC and
the proposed method are over twice those with the clas-
sic MPC and WMPC. These results show that the current
THD with MPC, WMPC, ­M2PC and DSVM-OSS-MPC
are ­THD(a) = 4.40%, ­THD(b) = 4.46%, ­THD(c) = 1.53%,
and ­THD(d) = 1.57%, respectively. Compared with classic
MPC and WMPC, the proposed control strategy pro-
vides a lower current THD. With the classic MPC and
WMPC, the converter operates at a variable switching
frequency with the average value lower than 5 kHz while
both ­M2PC and DSVM-OSS-MPC are operating at a
fixed switching frequency of 10 kHz.
Fig. 15 Experimental system structure diagram. a Simplified
The comparative evaluation of the four control strate-
diagram. b. Experimental set-up photograph gies at a similar average switching frequency (ASF = 3
kHz) is shown in Fig. 17, where the operating sampling
frequency of the classic MPC, WMPC, ­ M2PC, and
DSVM-OSS-MPC are 18 kHz, 18 kHz, 7 kHz, and 8
The comparative evaluation of the four control strat- kHz, respectively. The four control strategies show simi-
egies at the same sampling frequency (fs = 10 kHz) is lar current THD, i.e., ­ THD(a) = 2.43%, ­THD(b) = 2.50%,
shown in Fig. 16, where the ASF of the classic MPC, ­THD(c) = 2.76%, and ­THD(d) = 2.71%, respectively. With
WMPC, ­M2PC and DSVM-OSS-MPC are 1.712 kHz, MPC and WMPC, the converter operates at a variable

Fig. 16 Output current, capacitor voltage, current spectrum, average switching frequency, with ASF 1.712 kHz, 1.704 kHz, 4.330 kHz, and 3.815 kHz
respectively. a Classic MPC. b WMPC. c ­M2PC. d DSVM-OSS-MPC
Zhou et al. Protection and Control of Modern Power Systems (2023) 8:64 Page 12 of 16

Fig. 17 Output current, capacitor voltage, current spectrum, average switching frequency with ASF 3 kHZ. a Classic MPC. b WMPC. c ­M2PC. d
DSVM-OSS-MPC

of MPC and WMPC has to be increased, which is chal-

lenging to implement in low-cost digital processors.
The above results are further summarized in Fig. 18.
The proposed controller results in a low current THD
similar to the value under ­M2PC. However, the oper-
ating average switching frequencies with M ­ 2PC and
the proposed method are almost twice those with the
classic MPC and WMPC. When the resulting switch-
ing frequency is approximately the same with the four
Fig. 18 The experimental results at the same operating sampling controllers as shown in Fig. 19, the proposed approach
frequency. a current THD with the four control methods, b ASF operating at 8 kHz sampling frequency results in slightly
with the four control methods
higher current THD than the values with MPC and
WMPC since those are operating at 18 kHz sampling
frequency. However, implementing MPC and WMPC
switching frequency while ­M2PC and DSVM-OSS-MPC at such a high sampling frequency can be challenging
operate at a fixed switching frequency of 7 kHz and 8 because of the computational requirement. Therefore,
kHz, respectively. To achieve a similar current THD as the proposed DSVM-OSS-MPC which has the low-
with ­M2PC or DSVM-OSS-MPC, the sampling frequency est computation time is a suitable solution to improve
Zhou et al. Protection and Control of Modern Power Systems (2023) 8:64 Page 13 of 16

Fig. 19 The experimental results at around the same average

switching frequency. a current THD with the four control methods,
and b ASF with the four control methods

Fig. 21 Evaluation of the Computational Burden. a Classic MPC. b

WMPC. c ­M2PC. d DSVM-OSS-MPC

Table 7 Experimental and simulation comparison of the four

methods
Method Classic MPC WMPC M2PC DSVM-OSS-
MPC

Weight coef‑ With Without Without Without

Fig. 20 Output current, capacitor voltage. (a) Classic MPC. a Classic ficient
MPC. b WMPC. c ­M2PC. d DSVM-OSS-MPC THD 4.40% 4.46% 1.53% 1.57%
ASF 1.71 kHz 1.70 kHz 4.33 kHz 3.82 kHz
Frequency Wide Wide Concentrate Concentrate
spectrum at fs at fs
closed-loop performance in a scenario where a low-cost character‑
digital processor is used. istics
The dynamic responses of the current and capaci- vo < 1%Vdc < 1%Vdc < 1%Vdc < 1%Vdc
tor voltages are shown in Fig. 20. As seen, the DC bus Calculating 43.41 μs 36.28 μs 53.62 μs 23.26 μs
voltages remain balanced when the reference current is time
changed from 3 to 6 A. Compared with the classic MPC
and WMPC, M ­ 2PC and the proposed control strategy
have faster response time. The comparative study of the four control methods is
To extract the computation times of the four control summarized in Table 7. Compared with the classic MPC
methods, each control algorithm is implemented in the and WMPC, the proposed method significantly improves
TMS320F28379 DSP, and the digital output is set to a the accuracy of the current tracking, and the current har-
high voltage level when the algorithm is running and monics are mainly concentrated on the fixed frequency.
reset to a low voltage level when the processing is com- Compared to the ­M2PC method, the proposed control
pleted. The computation time required by each control- strategy achieves similar control accuracy and THD per-
ler is presented in Fig. 21. As can be seen, the proposed formance. It’s worth highlighting that the computation
strategy requires the lowest computational time of 23.26 time of the proposed strategy is significantly shorter than
μs, compared to 53.62 μs for ­ M2PC, which provides the other methods, leading to a substantial reduction in
almost the same closed-loop performance for the same computational burden. In addition, the proposed strategy
operating switching frequency. limits the capacitor voltage imbalance to be less than 1%.
Zhou et al. Protection and Control of Modern Power Systems (2023) 8:64 Page 14 of 16

6 Conclusion Author contributions

All authors contributed to the study conception and commented on previous
In this paper, a DSVM-OSS-MPC strategy operating versions of the manuscript. Sheng Zhou is the main contributor of the revised
with a fixed switching frequency is investigated for the paper and completed the design, simulation, data analysis, and validation
control of a 3L-NPC inverter. The strategy is based on of the proposed method in the revised paper. Minlong Zhu completed the
design and simulation of train working conditions in the origin submitted
a cascaded MPC approach for controlling the grid cur- paper. Jiaqi Lin helped the hardware experiments, data collection, and modify‑
rent and balancing the capacitor voltages without any ing of the revised paper. Paul Gistain Ipoum-Ngome was mainly responsible
WF. To improve control precision, an optimal voltage is for the selection of circuit parameters in the hardware design and put forward
modification suggestions for the language of the paper. Daniel Legrand
selected from the basic and virtual vectors of the pro- Mon-Nzongo guided the direction of the paper and put forward modification
posed extended space vector method. In the proposed suggestions. Prof. Tao Jin proposed the idea and designed the methodology
algorithm, the outer MPC employs an extended deadbeat of this paper, and guided the direction of the paper. All authors read and
approved the final manuscript.
method to output the optimal control voltage, reducing
the execution time of the proposed control algorithm. Funding
The inner MPC evaluates the optimal vector and its This work was supported by the Chinese National Natural Science Foundation
(grant number 51977039) and The central government guiding local science
potential redundancy, and selects the vector that mini- and technology development project under Grant 2021L3005.
mizes the NP voltage. Additionally, each virtual vector is
creatively synthesized as an OSS by using the DSVM and Availability of data and materials
Please contact author for data and material request.
considering its impact on the NP voltage and the inverter
switching commutations. The simulation and experimen-
Declarations
tal results indicate that compared to the classic MPC and
WMPC control algorithms, the closed-loop performance Competing interests
of the proposed algorithm is improved, and the current The authors declare that they have no known competing financial interests
or personal relationships that could have appeared to influence the work
THD is maintained below 2%. The computational burden reported in this paper.
of the proposed strategy is reduced by over 50% com-
pared to M ­ 2PC with similar closed-loop performance,
Received: 25 December 2022 Accepted: 20 November 2023
making it the most efficient option among all the com-
pared methods.
It is evident that the proposed strategy offers the bene-
fits of precise current response, capacitor voltage balance,
References
absence of WF influence, and low computational burden. 1. Hosseinkhani, V., & Sarvi, M. (2022). A new five-level inverter with reduced
These characteristics render it suitable for application in leakage current for photovoltaic system applications. Protection and
Control of Modern Power Systems., 7(2), 287–299.
3L-NPC inverters. In future work, we will further explore
2. Mukherjee, S., Giri, S. K., & Banerjee, S. (2019). A flexible discontinuous
and optimize this strategy in practical engineering applica- modulation scheme with hybrid capacitor voltage balancing strategy for
tions while researching potential extension and enhance- three-level NPC traction inverter. IEEE Transactions on Industrial Electronics,
66(5), 3333–3343.
ment to address the evolving challenges in grid-connected
3. Ouanjli, N. E., Derouich, A., & Ghzizal, A. E. (2019). Direct torque control of
renewable energy generation. doubly fed induction motor using three-level NPC inverter. Protection and
Control of Modern Power Systems., 4(1), 17.
4. Cortes, P., Kazmierkowski, M. P., Kennel, R. M., Quevedo, D. E., & Rodriguez,
Abbreviations J. (2008). Predictive control in power electronics and drives. IEEE Transac-
3L-NPC Three-level neutral-point-clamped inverter tions on Industrial Electronics, 55(12), 4312–4324.
MPC Model predictive control 5. Mohamed, A. (2021). High performance decoupled active and reactive
VV Virtual vectors power control for three-phase grid-tied inverters using model predictive
DSVM Discrete space vector modulation control. Protection and Control of Modern Power Systems., 6(3), 311–329.
SS Switching sequence 6. Guo, L., Xu, Z., Jin, N., Li, Y., & Wang, W. (2021). A weighted voltage model
NP Neutral point predictive control method for a virtual synchronous generator with
THD Total harmonic distortion enhanced parameter robustness. Protection and Control of Modern Power
M2PC Modulated MPC Systems., 6(4), 482–492.
OSS Optimal switching sequence 7. Camacho, E. F. (2007). Model predictive control. Springer.
FCS-MPC Finite control-set model predictive control 8. Zhao, Y. Y., An, A., Xu, Y., Wang, Q., & Wang, M. (2021). Model predictive
WF Weighting factor control of grid-connected PV power generation system considering
DSVM-OSS-MPC Discrete space vector modulation and optimized switch‑ optimal MPPT control of PV modules. Protection and Control of Modern
ing sequence model predictive controller Power Systems., 6(1), 2367–2617.
ASF Average switching frequency 9. Karamanakos, P., & Geyer, T. (2020). Guidelines for the design of finite con‑
WMPC The FS-MPC without WF trol set model predictive controllers. IEEE Transactions on Power Electronics,
35(7), 7434–7450.
Acknowledgements 10. Yang, Y., Pan, J., Wen, H., Zhang, X., Wang, Y., & Perdikakis, W. (2021). Model
Not applicable. predictive control with autotuning weighting factors for single-phase
Zhou et al. Protection and Control of Modern Power Systems (2023) 8:64 Page 15 of 16

six-level hybrid-clamped converters. IEEE Transactions on Industrial Elec- 30. Ipoum-Ngome, P. G., Mon-Nzongo, D. L., Flesch, R. C. C., Song-Manguelle,
tronics, 68(9), 7946–7956. J., Wang, M., & Jin, T. (2021). Model-free predictive current control for mul‑
11. Aguirre, M., Kouro, S., Rojas, C. A., Rodriguez, J., & Leon, J. I. (2018). tilevel voltage source inverters. IEEE Transactions on Industrial Electronics.,
Switching frequency regulation for FCS-MPC based on a period control 68(10), 9984–9997.
approach. IEEE Transactions on Industrial Electronics., 65(7), 5764–5773. 31. Karamanakos, P., Geyer, T., & Kennel, R. (2016). A Computationally efficient
12. Ni, Z., Abuelnaga, A., & Narimani, M. (2020). A novel high-performance model predictive control strategy for linear systems with integer inputs.
predictive control formulation for multilevel inverters. IEEE Transactions on IEEE Transactions on Control Systems Technology, 24(4), 1463–1471.
Power Electronics, 35(11), 11533–11543. 32. Portillo, R., Vazquez, S., Leon, J. I., Prats, M. M., & Franquelo, L. G. (2013).
13. Xia, S., Wu, X., Zheng, J., Li, X., & Wang, K. (2021). A virtual space vector Model based adaptive direct power control for three-level NPC convert‑
PWM with active neutral point voltage control and common mode ers. IEEE Transactions on Industrial Informatics., 9(2), 1148–1157.
voltage suppression for three-level NPC converters. IEEE Transactions on 33. Vargas, R., Cortés, P., Ammann, U., Rodríguez, J., & Pontt, J. (2007). Predictive
Industrial Electronics, 68(12), 11761–11771. control of a three-phase neutral-point-clamped inverter. IEEE Transactions on
14. Yang, Y., Wen, H., Fan, M., Xie, M., Chen, R., & Wang, Y. (2019). A constant Industrial Electronics, 54(5), 2697–2705.
switching frequency model predictive control without weighting factors 34. Baidya R, Aguilera RP, Karamanakos P, Acuna P, Rojas C, Geyer T, Lu DD (2019)
for T-type single-phase three-level inverters. IEEE Transactions on Industrial Dealing with suboptimality in multistep model predictive control for transient
Electronics, 66(7), 5153–5164. operations, In: 2019 IEEE energy conversion congress and exposition (ECCE), pp.
15. Aguirre, M., Kouro, S., Rojas, C. A., & Vazquez, S. (2021). Enhanced switch‑ 3780-3785
ing frequency control in FCS-MPC for power converters. IEEE Transactions 35. Li, X., Xue, Z., Zhang, L., & Hua, W. (2021). A low-complexity three-vector-based
on Industrial Electronics., 68(3), 2470–2479. model predictive torque control for SPMSM. IEEE Transactions on Power Elec-
16. Liu, X. (2022). Neural predictor-based low switching frequency FCS-MPC tronics, 36(11), 13002–13012.
for MMC With online weighting factors tuning. IEEE Transactions on Power 36. Vazquez, S., Leon, J.I., Franquelo, L.G., Carrasco, J.M., Martinez, O., Rodriguez, J.,
Electronics, 37(4), 4065–4079. Cortes, P. & Kouro, S. (2009). Model predictive control with constant switching
17. Villarroel, F. (2013). Multiobjective switching state selector for finite-states frequency using a discrete space vector modulation with virtual state vectors.
model predictive control based on fuzzy decision making in a matrix In: 2009 IEEE international conference on industrial technology, pp. 1-6.
converter. IEEE Transactions on Industrial Electronics, 60(2), 589–599. 37. Yu, B., Song, W., Li, J., Li, B., & Saeed, M. S. R. (2021). Improved finite control set
18. Caseiro, L. M. A., Mendes, A. M. S., & Cruz, S. M. A. (2019). Dynamically model predictive current control for five-phase VSIs. IEEE Transactions on Power
weighted optimal switching vector model predictive control of power Electronics, 36(6), 7038–7048.
converters. IEEE Transactions on Industrial Electronics, 66(2), 1235–1245. 38. Yu, B., Song, W., Guo, Y., Li, J., & Saeed, M. S. R. (2021). Virtual voltage vector-
19. Habibullah, M., Lu, D. D., Xiao, D., & Rahman, M. F. (2016). Finite-state based model predictive current control for five-phase vsis with common-
predictive torque control of induction motor supplied from a three- mode voltage reduction. IEEE Transactions on Transportation Electrification, 7(2),
level NPC voltage source inverter. IEEE Transactions on Power Electron- 706–717.
ics., 32(1), 479–489.
20. Antoniewicz, K., Jasinski, M., Kazmierkowski, M. P., & Malinowski, M.
(2016). Model predictive control for three-level four-leg flying capacitor Sheng Zhou was born in Fuzhou, China in 1983 . He received the
converter operating as shunt active power filter. IEEE Transactions on Master degree in Software Engineering from Nanjing University, Nan‑
Industrial Electronics., 63(8), 5255–5262.
jing, China, in 2007. He is Professor Senior Engineer, more than 15
21. Cortés, P., Rodríguez, J., Quevedo, D. E., & Silva, C. (2008). Predictive
current control strategy with imposed load current spectrum. IEEE
years of digital management and data analysis experience in power
Transactions on Power Electronics., 23(2), 612–618. industry. He is currently working at State Grid Fujian Electric Power
22. Donoso, F., Mora, A., Cardenas, R., Angulo, A., Saez, D., & Rivera, M. Co., LTD.
(2018). Finite-set model-predictive control strategies for a 3L-NPC
inverter operating with fixed switching frequency. IEEE Transactions on Minlong Zhu was born in Jiangxi province, China, in 1997. He
Industrial Electronics., 65(5), 3954–3965. received B.S. degrees (2020) in Electrical Engineering and Automation
23. Vazquez, S., Marino, D., Zafra, E., Peña, M. D., Rodríguez-Andina, J. J., from Yanshan University. Since 2020, He is working toward the M.S.
Franquelo, L. G., & Manic, M. (2022). An artificial intelligence approach degree in the Department of Electrical Engineering and Automation
for real-time tuning of weighting factors in FCS-MPC for power con‑
of Fuzhou University. His current research interests include the model
verters. IEEE Transactions on Industrial Electronics, 69(12), 11987–11998.
24. Tarisciotti, L., Zanchetta, P., Watson, A., Clare, J. C., Degano, M., & Bifaretti,
predictive control, renewable power generation and control of mul‑
S. (2015). Modulated model predictive control for a three-phase active tilevel inverters.
rectifier. IEEE Transactions on Industry Applications, 51(2), 1610–1620.
25. Vazquez, S., Marquez, A., Aguilera, R., Quevedo, D., Leon, J. I., & Fran‑ Jiaqi Lin was born in Fujian, China, in 1999. He received the B.S.
quelo, L. G. (2015). Predictive optimal switching sequence direct power degree in electrical engineering and automation from Huaqiao
control for grid-connected power converters. IEEE Transactions on University, Xiamen, China, in 2021. He is currently pursuing the M.S.
Industrial Electronics, 62(4), 2010–2020. degree with the Department of Electrical Engineering and Automa‑
26. Hu, B., Kang, L., Liu, J., Zeng, J., Wang, S., & Zhang, Z. (2019). Model tion, Fuzhou University, Fuzhou. His research interests include topo‑
predictive direct power control with fixed switching frequency and
logical structure of DC-DC converter and single-stage boost inverter.
computational amount reduction. IEEE Journal of Emerging and Selected
Topics in Power Electronics, 7(2), 956–966.
27. Wang, Y., Wang, X., Xie, W., Wang, F., Dou, M., Kennel, R. M., Lorenz, R. D., Paul Gistain Ipoum‑Ngome (Member, IEEE) was born in Douala,
& Gerling, D. (2017). Deadbeat model-predictive torque control with Cameroon, in 1989. He received the B.S., M.S. and DEA degrees in
discrete space-vector modulation for PMSM drives. IEEE Transactions on electrical engineering from Ecole Normal Supérieure D’ Enseigne‑
Industrial Electronics, 64(5), 3537–3547. ment Technique (ENSET) of Douala University, Douala, Cameroon in
28. Xing, X., Zhang, C., Chen, A., Geng, H., & Qin, C. (2018). Deadbeat control 2013, 2015, and 2016, respectively. And the Ph.D degree in electrical
strategy for circulating current suppression in multiparalleled three-level engineering from Fuzhou University, Fuzhou, China, in 2021. Since
inverters. IEEE Transactions on Industrial Electronics, 65(8), 6239–6249. May 2021, he has been a Postdoctoral Researcher Associated with
29. Li, X., Zhang, S., Cui, X., Wang, Y., Zhang, C., Li, Z., & Zhou, Y. (2022). Novel
Fuzhou University and Pearl-Electric Co. Ltd, Guangzhou, China. His
deadbeat predictive current control for PMSM With parameter updating
scheme. IEEE Journal of Emerging and Selected Topics in Power Electronics,
current research interests include MPC, Data-driven predictive con‑
10(2), 2065–2074. troller for grid-connected and motor drive applications.

Daniel Legrand Mon‑Nzongo (Member, IEEE) received the B.S.

Zhou et al. Protection and Control of Modern Power Systems (2023) 8:64 Page 16 of 16

degree from the Ecole Normale Supérieure d’Enseignement Tech‑

nique (ENSET) of Douala University, Douala, Cameroon, in 2010 and
the Ph.D. degree from Fuzhou University, Fuzhou, China, in 2018, all
in electrical engineering. From 2018 to 2021, he was a Postdoctoral
Research Fellow with Pearl Electric, Co. Ltd, Guangzhou, China. Dur‑
ing this time, he was the team leader of the Power Electronics Lab. He
was involved on several industry funded projects related to medium-
voltage converters and energy router technologies applied in railway
applications. He is currently an Associate Research Professor with the
College of Electrical Engineering and Automation of Fuzhou Univer‑
sity. His current research interests include bidirectional AC and DC
converters for medium voltage applications.

Tao Jin (FIET, SM’2019) received B.S. and M.S. degrees in Electri‑
cal Engineering from Yanshan University respectively in 1998 and
2001, and the Ph.D. degree in Electrical Engineering from Shang‑
hai Jiaotong University in 2005. From 2005 to 2007, he worked as a
postdoctor in Shanghai Jiaotong University. From 2008 to 2009, he
held research scientist position with Virginia Tech, Blacksburg, USA.
In 2010, he joined Imperial College London, UK, as European Union
Marie Curie Research Fellow. Now he is a professor in College of
Electrical Engineering & Automation, Fuzhou University, China. Prof.
Tao Jin has published about 200 papers, and he is members of IEEE
Power and Energy Society and IEEE Industrial Electronics Society, and
special committee members of Chinese Society of Electrical Engi‑
neering, China Electrotechnical Society, and etc. He currently serves
as associate editors for MPCE, PCMP, High Voltages, Scientific Reports,
and other journals.