electronics

Article
A Simplified Optimal-Switching-Sequence MPC with
Finite-Control-Set Moving Horizon Optimization for
Grid-Connected Inverter
Longyun Kang 1,2 , Jiancai Cheng 1,3, * , Bihua Hu 1,3 , Xuan Luo 1,3 and Jianbin Zhang 1,3
1 New Energy Research Center, School of Electric Power, South China University of Technology,
Guangzhou 510640, China; lykang@scut.edu.cn (L.K.); ephubihua2015@mail.scut.edu.cn (B.H.);
epluoxuan@mail.scut.edu.cn (X.L.); zhang_jianbin@outlook.com (J.Z.)
2 Dongguan DRN New Energy CO, LTD., Dongguan 523000, China
3 Guangdong Key Laboratory of Clean Energy Technology, School of Electric Power, South China University
of Technology, Guangzhou 510640, China
* Correspondence: cheng.jc@mail.scut.edu.cn; Tel.: +86-188-1946-6247




Received: 15 March 2019; Accepted: 22 April 2019; Published: 25 April 2019


Abstract: Optimal-switching-sequence model-predictive-control (OSS-MPC) strategy, a popular

kind of continuous-control-set MPC (CCS-MPC), has been used to address the variable frequency
of finite-control-set MPC (FCS-MPC). Since the digital signal processor (DSP) does not have a
dedicated divider, it requires a large amount of computation for the division operations in OSS
searching. Here, a simplified OSS-MPC absorbing the merits of FCS-MPC is proposed to reduce the
computational burden of conventional OSS-MPC. The proposed method uses a novel cost function
whose candidates are the finite center vectors instead of switching sequences, which can avoid duty
cycle computation and convert the moving horizon optimization of CCS-MPC into that of FCS-MPC.
Besides, the derivatives of the active and reactive powers are divided into the constant terms and
variable terms. The constant terms are extracted from the cost function. Experimental and simulation
results show that the computational amount of the proposed algorithm is only 36.34% of that of the
conventional OSS-MPC. Meanwhile, the simplified OSS-MPC still maintains the excellent dynamic
and steady state performance of conventional OSS-MPC.

Keywords: grid-connected inverter; model predictive control; cost function; complexity optimization

1. Introduction
Grid-connected inverter plays a key role in renewable energy systems, distributed generation
systems, power quality improvement units, and so on [1,2]. Since the performance of the inverter
depends largely on the control strategy, much effort has been done on the control strategy of the
grid-connected inverter.
Classical control strategies of the grid-connected inverter can be mainly divided into two
categories [2]. One is voltage oriented control (VOC) [3], which is broadly used for grid-connected
inverters due to its good steady-state performance. Nevertheless, the inner current control loop and
the proportional integral (PI) controllers may lower the dynamic performance of the system. The other
is direct power control (DPC), which is based on the direct torque control (DTC) of AC machines [4,5],
has been proposed to get better dynamic performance. However, it has variable switching frequency
and large power ripples due to the predefined switching table.
Recently, a novel and computationally complex control strategy, model predictive control (MPC),
has been proposed. The MPC shows a lot of advantages such as ease of implementation, flexibility in
the definition of control objectives, and fast dynamic response [6–8].

Electronics 2019, 8, 457; doi:10.3390/electronics8040457 www.mdpi.com/journal/electronics

Electronics 2019, 8, 457 2 of 18

Depending on the nature of the input constraints, model-predictive-control (MPC) can be divided
into continuous control set (CCS) MPC and finite control set (FCS) MPC [9]. In FCS-MPC schemes,
the inverter switching states are handled directly without using a modulator, and the input will belong
to an FCS. Once a modulator is used, the input will be constrained to belong to a bounded CCS,
which contains more than one switching state in a sampling period. Therefore, FCS-MPC only selects
one switching state in a sampling period, which causes high power ripple and variable switching
frequency [10,11]. Considering that CCS-MPC can acquire constant switching frequency by using
the modulator [12]. The best-known algorithm in CCS-MPC is the predictive direct power control
(P-DPC) based on the space vector modulation (SVM) whose vector sequence that contains two nonzero
vectors and one zero vector is employed during every control period. Although conventional P-DPC
can realize fixed switching frequency, its performance is deteriorated by incorrect vector sequence
selection [13,14]. This is because the two nonzero vectors in vector sequence are selected according
to the angular position of the grid voltage vector instead of the ideal output voltage vector. And the
incorrect vector sequence might lead to the calculation of negative duration time, which would be
clamped to zero in the practical system. To select the correct vector sequence, an improved vector
sequence table has been proposed for P-DPC in reference [14], which can modify the optimal switching
sequence when the duration time becomes negative. However, the control action of all these algorithms
still require the grid voltage angle and an extra SVM.
To solve this issue, the CCS-MPC with optimal switching sequence (OSS-MPC) has been presented
in references [15–17], substituting the vector sequences that contain more than one voltage vector into
the cost function. Compared with P-DPC, the advantage of OSS-MPC is that it does not need grid
voltage angle and can select the OSS even when the control action is saturated [15]. Since the OSS-MPC
in a way resembles a modulator in the optimization problem, the duration time of the switching
states are needed. However, the division of the duration time of the switching states will increase
the computational burden of the microprocessor. Considering other functions in renewable energy
sources that should be implemented in the sampling period, such as islanding detection protection
or maximum power point tracking algorithm [18], the computational burden of the conventional
OSS-MPC should be reduced.
Different from the classical OSS-MPC with two nonzero vectors and one zero vector in one
sampling period, there is also OSS-MPC with two voltage vectors in one sampling period, which is
called dual vector based MPC (D-MPC) [19–21]. These methods can mitigate the computational burden
due to the less voltage vectors in one sampling period, and have a faster speed of moving horizon
optimization. However, since D-MPC only contains two voltage vectors in the OSS, there is only one
degree of freedom for the duration time control and the active and reactive power errors cannot be
removed at the same time. In references [18,22], a constructed reference voltage vector in two-level and
three-level converter has been applied in the cost function optimization for OSS-MPC. The reference
voltage vector can select the OSS directly, thus reducing the cyclic number from 6 to 1. However, it needs
a grid voltage angle to construct a reference voltage vector when it is connected to the grid, which partly
deviates from the advantages of conventional OSS-MPC. There are many other algorithms that can
alleviate the computational burden of other kinds of CCS-MPC [23,24] and FCS-MPC [25–27] but few
about OSS-MPC. However, they are not suitable for our target (OSS-MPC for two-level grid-connected
inverter) because of the difference between moving horizon optimization [23,24], or may sacrifice some
advantages of OSS-MPC if they are introduced into OSS-MPC [25–27]. The lack of work concerning
the computational efficiency of classical OSS-MPC is partly because it often damages the merit of
OSS-MPC while reducing the computational burden. The related literature that can maintain the
merit of OSS-MPC is reduced OSS-DPC (ROSS-DPC) [15], which reduces the computational cost by
calculating the derivative of active and reactive power of the switching states before moving horizon
optimization. In this way, the calculating times of the derivative of the active and reactive power can
be reduced. However, this method has limited effect since the duration times of each vector need to be
calculated, which still occupies a lot of computational cost during the minimization of the cost function.
Electronics 2019, 8, 457 3 of 18
Electronics 2019, 8, x FOR PEER REVIEW 3 of 18

This
This paper
paper aimsaims to propose a simplified OSS-MPC strategy, strategy, which
which cancan reduce
reduce the
the computational
computational
burden while maintaining
burden while maintaining all advantages of OSS-MPC. We
We introduce the core idea of FCS-MPC into
introduce the core idea of FCS-MPC into
OSS-MPC
OSS-MPC by constructing
constructing a new new costcost function
function with with center
center vectors.
vectors. The new cost cost function
function uses
uses the
the
center
center vectors
vectorstotosearch
searchthe theOSS
OSSand and avoid
avoid thethe
division
divisionoperation
operationin duration
in duration timetime
calculation. AndAnd
calculation. the
duration timetime
the duration of OSS willwill
of OSS be becalculated
calculated after moving
after moving the
thehorizon
horizonoptimization.
optimization.At Atthe
the same
same time,
the
the active
active and
and reactive
reactive power
power predictive model are divided into various parts and a constant part.
Then
Then the
the constant
constant part part isis extracted
extracted from from the the moving
moving horizon
horizon optimization
optimization thus as to to decrease
decrease thethe
computational
computational amount further. further. In this way, way, the the proposed
proposed algorithm
algorithm cancan alleviate
alleviate thethe computational
computational
burden
burden of of conventional
conventional OSS-MPCOSS-MPC withoutwithout sacrificing
sacrificingthe theexcellent
excellentperformance
performanceof ofit.
it.
The
The main
mainadvantages
advantagesofofthe theproposed
proposed method
method arearesummarized
summarized as follows:
as follows: (i) Compared
(i) Compared withwith
the
D-MPC
the D-MPC[19–21], the proposed
[19–21], the proposedstrategy maintains
strategy three voltage
maintains vectors in
three voltage one sampling
vectors period, and
in one sampling thus
period,
has
andathus
betterhas
steady statesteady
a better performance. (ii) Compared
state performance. (ii)with the strategy
Compared withintroduced
the strategy in reference
introduced [18],
in
the proposed strategy does not require the grid voltage sector information.
reference [18], the proposed strategy does not require the grid voltage sector information. The OSS The OSS can be selected
even
can be when the control
selected even when actiontheiscontrol
saturatedaction[15]. (iii)Compared
is saturated with ROSS-DPC
[15]. (iii)Compared [15],
with the proposed
ROSS-DPC [15],
strategy can reduce
the proposed strategy thecan
computational burden further.burden further.
reduce the computational
The
The hypotheses
hypotheses of the the study
study are are given
given as as follow.
follow. We We assume
assume thatthat (i)
(i) the
the insulated
insulated gategate bipolar
bipolar
transistors
transistors (IGBTs)
(IGBTs) are are ideal
ideal switches;
switches; (ii)(ii) the
the resistance
resistance of of the
the wires
wires and
and the
the inductors
inductors (AC(AC filter)
filter) can
can
be
be ignored;
ignored; (iii)
(iii) the
the inverter
inverter is is connected
connectedto toaabalanced
balancedand andideal
idealgrid.
grid.
The
The rest
restofofthis paper
this paperis organized
is organizedas follows. Conventional
as follows. OSS-MPC
Conventional for grid-connected
OSS-MPC inverter
for grid-connected
is introduced
inverter in Section 2.
is introduced in In Section2.3,In
Section theSection
proposed algorithm
3, the proposedis analyzed
algorithm andisitsanalyzed
superiorities
and areits
summarized.
superiorities are In Section 4, the experimental
summarized. In Section 4, the platform is established
experimental platformto verify the effectiveness
is established to verifyandthe
advantages
effectivenessofand the advantages
proposed algorithm. Finally,algorithm.
of the proposed the conclusion is given
Finally, in Sectionis5.given in Section 5.
the conclusion

2.
2. Conventional
Conventional OSS-MPC
OSS-MPC

2.1.
2.1. Model
Model of
of Two-Level
Two-Level Grid-Connected
Grid-Connected Inverter
Inverter
Figure
Figure 1 shows an exemplary two-level grid-tied
1 shows an exemplary two-level grid-tied inverter,
inverter, which
which contains
contains six
six insulated
insulated gate
gate
bipolar
bipolar transistors
transistors(IGBTs)
(IGBTs)with
withantiparallel
antiparalleldiodes. V dc
diodes. Vdcisisthe direct
the current
direct (dc)
current power
(dc) supply,
power e is the
supply, e is
grid voltage,
the grid C1 and
voltage, C2 are
C1 and thethe
C2 are dc-link capacitors.
dc-link Additionally,
capacitors. Additionally, L isL the inductance
is the inductanceof of
thethe
AC filter.
AC filter.

Sa1 Sb1 Sc1

C1 L e
i
Vdc
O
C2

Sa2 Sb2 Sc2

Figure 1. The topology of two-level grid-connected inverter.

Figure 1. The topology of two-level grid-connected inverter.
As shown in Figure 1, SX1 and SX2 represent the upper and lower switching states of each phase,
x =shown
with As a, b, c.inDifferent
Figure 1, switching
SX1 and SX2states
represent
of eachthe phase
upper in and lower switching
inverter can produce states of eachoutput
different phase,
with x = a, b, c. Different switching states of each phase in inverter can produce
voltages ui . And there are two switching states of each phase, which can be denoted as “P” and “N”. different output
voltages
The ui. And
switching there
state “P”are two switching
means statesleg
that the upper of switch
each phase,
SX1 iswhich
on andcan thebelower
denoted as “P” Sand
leg switch X2 is“N”.
off,
The switching state “P” means that the upper leg switch S is on and the lower
and “N” is the opposite. Therefore, the two-level three-phase inverter has eight switching states that
X1 leg switch S X2 is off,
andgenerate
can “N” is the opposite.
seven validTherefore, the two-level
voltage vectors, i.e., six three-phase
nonzero vectorsinverter
(i.e.,has
u1 , eight
u2 , u3switching
, u4 , u5 , u6states that
) and one
can generate
zero seven
vector (i.e., u0 ).valid voltage vectors
The voltage vectors,are
i.e.,illustrated
six nonzeroin vectors
Figure 2. (i.e. u1, u2, u3, u4, u5, u6) and one zero
vector (i.e. u0). The voltage vectors are illustrated in Figure 2.
Electronics2019,
Electronics 8,x457
2019,8, FOR PEER REVIEW 44of
of18
18

β
u3 u2

II
III I
u4 u0 u1 α
IV VI
V

u5 u6
Figure 2. Space vector diagram of the two-level inverter.
Figure 2. Space vector diagram of the two-level inverter.
Under the stationary αβ framework, the model of the three-phase inverter can be expressed as
Under the stationary αβ framework, the model of the three-phase inverter can be expressed as
di 1
di= 1(ui − e), (1)
dt = L ( ui − e ) , (1)
dt L
where ui , e, and i are the inverter voltage vector, grid voltage vector, and grid current vector, respectively.
whereAccording
ui, e, andtoi different
are the controlling
inverter voltage vector, grid voltage vector, and grid current vector,
goals, the MPC can be classified into model predictive current
respectively.
control (MPCC) and model predictive power control (MPPC). We take MPPC as an example to illustrate
According to different controlling goals, the MPC can be classified into model predictive current
the proposed method. On the basis of the instantaneous power theory, the apparent powers can be
control (MPCC) and model predictive power control (MPPC). We take MPPC as an example to
expressed by
illustrate the proposed method. On the basis of the instantaneous
3  power theory, the apparent powers
can be expressed by S = P + jQ = e · î , (2)
2
where “ˆ” denotes complex conjugate, P and Q are active
3 and reactive power, respectively.
S = P + jQ = e
Under the balanced three-phase system, the derivative
2
⋅ ˆ ,
i ( )
of grid voltage vector can be obtained(2)
as

where “^” denotes complex conjugate, P and Q

deαare active and reactive power, respectively.

dt = −ωeβ



Under the balanced three-phase system,  the
 deβ derivative
 , of grid voltage vector can be obtained as (3)


dt = ωe α
 deα
 = −ω e β
where ω is the grid angular frequency. By combining dt Equations (1)–(3), the derivatives of the active
 , (3)
and reactive powers in the stationary αβ framework de β = ωcan be calculated as
eα
 h u −e  dt uβi −eβ i
α
 dt = 2 eα ( αiL + ωiβ ) + eβ ( L − ωiα )
dP 3

where ω is the grid angular

frequency.
h By combining Equations (1), (2),
i . and (3), the derivatives (4)
of

u −e uβi −eβ
 dQ

3
eβ ( αiL α +αβ ωiframework − ωi

in = β ) − eα ( L can )
beαcalculated

the active and reactive powers dt the2 stationary as

2.2. FCS-MPC  dP 3  uα i − eα uβ i − eβ 
 =  eα ( + ω iβ ) + e β ( − ω iα ) 
 dtmethod.
MPC is a nonlinear control
2 
As shown in Figure 3, it uses thesampling
L L
value x(k) and (4)
the
 .
discrete mathematical model dQ  uα i to
of the3system − epredict its u
future
βi
− e
behavior
β
 x(k + 1) within a finite time
 dt =action,
 eβ (
α
+ ω iβ ) − eα ( − ω iα ) 
horizon for all possible switching 2 
 represented
L by n. The Lcost function of MPC, considering the

reference value x*(k) and other constraints, selects the optimal switching action S that minimizes the
cost function to be applied in the next sampling instant [28]. Such calculations are repeated in each
2.2. FCS-MPC
sampling period.
MPC is a nonlinear control method. As shown in Figure 3, it uses the sampling value x(k) and
the discrete mathematical model of the system to predict its future behavior x(k+1) within a finite
time horizon for all possible switching action, represented by n. The cost function of MPC,
considering the reference value x*(k) and other constraints, selects the optimal switching action S that
minimizes the cost function to be applied in the next sampling instant [28]. Such calculations are
repeated in each sampling period.
Electronics
Electronics2019,
2019,8,8,x457
FOR PEER REVIEW 55ofof18
18

Converter
x*(k)
Minimize S
x(k+1) cost Grid
Predictive discrete
function
mathematical n

{
model
x(k)
x(k)

Figure 3. Flow chart of model-predictive-control (MPC).

Figure 3. Flow chart of model-predictive-control (MPC).
Assuming the sampling time is relatively small, the grid current and grid voltage can be seen as a
constant value the
Assuming andsampling time is of
the derivatives relatively
the active small,
andthe grid current
reactive power and grid voltageconstant
are considered can be seen as
during
athe
constant
periodvalue and the
in which derivatives
a certain outputofvoltagethe active and is
vector reactive
active.power Thus, are considered
Equation (4) atconstant
instant kduring
can be
the period
rewritten as in which a certain output voltage vector is active. Thus, Equation (4) at instant k can be
rewritten as
 dP
 fpi (k) = dti 

uα = uαi , uβ = uβi . (5)

fqi (k) = dtdPi
dQ

 f pi ( k ) =


dt u = u , u = u .

There are 8 switching states in two-level inverter,
α and
α i they
β are
β i corresponding to 7 output voltage (5)
 dQ
vectors. Different output voltage vectors fqi ( k )generate
= different fpi and fqi . Since there are one output
 dt
voltage vector implemented in the sampling period Ts , the active and reactive power at instant k + 1 in
FCS-MPC
There arecan8be predicted
switching as in two-level
states  inverter, and they are corresponding to 7 output voltage
vectors. Different output voltage vectors
 P ( k + 1 ) = P(kdifferent
generate ) + fpi (k)T fpis and fqi. Since there are one output
, (6)

voltage vector implemented in the sampling
 Q(k + 1period

 T s, the active and reactive power at instant k+1 in
) = Q(k) + fqi (k)Ts
FCS-MPC can be predicted as
where Ts is the sampling period.
 P ( k + 1) = P ( k ) +off piFCS-MPC,
Considering the delay of digital implementation ( k ) Ts the control program has a delay
 , (6)
 ( k + 1) = Q ( k ) + fqi ( k ) Ts
time tdelay between the sampling instantQ and the implementation of the switching action [29]. The active
and reactive power in conventional FCS-MPC should be predicted as
where Ts is the sampling period.
      
tdelay tdelay tdelay
Considering the delay of
P kdigital
+ 1 + implementation = P k + of FCS-MPC,
+ f k + the control program has a delay
Ts Ts



 Ts Ts pi
time tdelay between the sampling instant and the implementation of the switching , action [29]. The
(7)
      
 Q k + 1 + tdelay = Q k + tdelay + f k + tdelay T


active and reactive power in conventional Ts FCS-MPC should Ts beqi predictedTs as
s

where P(k + 1 + tdelay /Ts ), Q(k+ 1 + tdelaytdelay /Ts )are the

 active
tdelay and reactive
 
tdelaypower of instant k + 1 + tdelay /Ts ,
P  k + 1 +  = Pk +  + f pi  k +  Ts

and fpi (k + tdelay /Ts ), fqi (k + tdelay
  s/T ) are the
T  
derivative T of active  and T 
reactive power of instant k + 1 +
s   s   s 
tdelay /Ts .  , (7)
  tdelay   tdelay   tdelay 
Q  k + 1 + T P(k
However, in practical applications,  =+Q1 k++tdelay /T
 
 +s ),f Q(k
Ts  qi 
 k ++ 1 +tTdelay
Ts  s
/Ts ), fpi (k + tdelay /Ts ) and
  by sampling
fqi (k + tdelay /Ts ) cannot be obtained s   calculation,
and thus Equation (7) cannot be used as a
predictive model for active power and reactive power in applications. Then, various kinds of delay
where P(k+1+tdelay/Ts), Q(k+1+tdelay/Ts) are the active and reactive power of instant k+1+tdelay/Ts, and
compensation have been proposed. Here we choose a normal two steps ahead prediction to achieve
fpi(k+tdelay/Ts), fqi(k+tdelay/Ts) are the derivative of active and reactive power of instant k+1+tdelay/Ts
delay compensation [30]. We set tdelay as Ts , and the prediction equation are written as
However, in practical applications, P(k+1+tdelay/Ts), Q(k+1+ tdelay/Ts), fpi(k+tdelay/Ts) and fqi(k+tdelay/Ts)
cannot be obtained by sampling  and calculation, thus Equation (7) cannot be used as a predictive
 P(k + 2) = P(k + 1) + fpi (k + 1)Ts

model for active power and  reactive power in applications. Then, , various kinds of delay (8)
compensation have been proposed.  Q(Here

k + 2)we Q(k + 1a) normal
= choose + fqi (k + 1)Tsteps
two s ahead prediction to achieve
delay compensation [30]. We set tdelay as Ts, and the prediction equation are written as
where P(k + 1) and Q(k + 1) are computed by grid voltage and current sampled at instant k and
fpi (k), fqi (k) of the optimal output voltage  P ( k +vector
2 ) = Pcalculated
( k + 1) + fpi at( kinterval
+1) Ts between k − 1 and k. As shown
 , (8)k.
Q ( k + 2 ) = Q ( k + 1) + fqi ( k +1) Ts
in Figure 4, the power trajectory from k + 1 to k + 2 can be predicted by Equation (8) at instant
Though we chose a normal two steps ahead prediction, the prediction horizon in the model is still
where
one step. P(k+1) and Q(k+1) are computed by grid voltage and current sampled at instant k and fpi(k),
fqi(k) of the optimal output voltage vector calculated at interval between k-1 and k. As shown in Figure
Electronics 2019, 8, x FOR PEER REVIEW 6 of 18

Electronics
4, the power 8, 457
2019, trajectory 6 of 18
from k+1 to k+2 can be predicted by Equation (8) at instant k. Though we chose
a normal two steps ahead prediction, the prediction horizon in the model is still one step.
P(k)

P(k)
Pref P(k+2)
P(k+1)

Q(k)

Q(k+2)
Qref
Q(k+1)
Q(k)
k-1 k k+1 k+2

Figure4.4.Control
Figure Controldelay
delay compensation
compensation based on the
based on the two
two steps
steps ahead
aheadprediction
predictionin
infinite-control-set
finite-control-set
(FCS)-MPC.
(FCS)-MPC.

The optimal voltage vector in FCS-MPC from the previous iteration is used to estimate the active
The optimal voltage vector in FCS-MPC from the previous iteration is used to estimate the active
and reactive power at instant k + 1. The estimated active power P(k + 1), reactive power Q(k + 1) and
and reactive power at instant k+1. The estimated active power P(k+1), reactive power Q(k + 1) and all
all the switching states are used to predict the active and reactive power at instant k + 2.
the switching states are used to predict the active and reactive power at instant k + 2.
The cost function of FCS-MPC with delay compensation should be expressed as
The cost function of FCS-MPC with delay compensation should be expressed as
2 2
J = [JP=re[f P− P−(kP+
( k 2+)]2)]+ +[Q
[Qrereff −
−Q 2)]2 2, ,
Q((kk ++2)] (9)
(9)
ref

where PPref
where ref and QQref are the
ref are the reference
reference values of the
the active
active and
andreactive
reactivepower, respectively.P(k
power,respectively. + 2) and
P(k+2)
Q(k + 2)are
Q(k+2) arethe
thepredicted
predictedactive
activeand
andreactive
reactivepower
poweratatinstant k + 2.
instantk+2.
FCS-MPC has the disadvantage of implementing only voltage
FCS-MPC has the disadvantage of implementing only one vectorvector
one voltage in eachinsampling period,
each sampling
which
period,leads
which to leads
poor steady-state performance
to poor steady-state and variable
performance switching
and variable frequency.
switching frequency.

2.3. Conventional OSS-MPC

2.3. Conventional OSS-MPC
In conventional
In conventional OSS-MPC,
OSS-MPC,there thereare arethree
threevoltagevoltage vectors
vectors in in oneone sampling
sampling period.
period. Hence,
Hence, the
the predicted active and reactive power at instant
predicted active and reactive power at instant k+2 are predicted as k + 2 are predicted as

 P2()k= + 2P)(k= +
P (1k) ++1)fp1
+ (fkp1+ ( k1+1)t)1t+ f p 2(k( k+
+ fp2 +11))tt22++f pf3p3( (kk+1+) t13)t3
(
P(k + 1
 2) = Q(k + 1) + fq1 (k + 1)t1 + fq2 (k + 1)t2 + fq3 (k + 1),t3 , (10)
(10)
Q(k +
Q ( k + 2 ) = Q ( k + 1 ) + f q1 ( k +1 ) t 1
+ f q2 ( k +1 ) t 2
+ f q3 ( k +1 ) t 3

where ffp1
where (k + 1), fp2 (k + 1) and fp3 (k + 1) are the derivatives of the active power for two nonzero vectors
p1(k+1), fp2(k+1) and fp3(k+1) are the derivatives of the active power for two nonzero vectors and
and one zero
one zero vector vector
in one incontrol
one controlperiod, period, respectively,
respectively, fq1(k+1), fq1f(k + 1), fq2 (k + 1) and fq3 (k + 1) are the
q2(k+1) and fq3(k+1) are the derivatives of
derivatives
the reactiveofpower
the reactive power
similarly. t1, tsimilarly.
2 and t3 are
t1 , the
t2 anddurationt3 are the time duration
for these time for these
vectors. In vectors.
order toIn order
predict
to P(k + Q(k +
P(k+2) and Q(k+2), the voltage vectors implemented at instant k should be calculated at instant k-1. at
predict 2) and 2), the voltage vectors implemented at instant k should be calculated
instant k − 1.
In reference [31], by setting the dJ/dt to zero, the optimal duration time t1, t2 and t3 can be obtained
as In reference [31], by setting the dJ/dt to zero, the optimal duration time t1 , t2 and t3 can be
obtained as
 ( )
fq 2 (− q2fq 3 q3E)p + ( ) (
f − f Ep +( fp3 − fp2 )Eq +( fp2 fq3 − fp3 fq2 )Ts
)

t1 = f − f 2 Eq + f p 2 fq 3 − f p 3 fq 2 Ts
t

= ( fq2 − fq3 ) fp1 +p(3 fq3 −pfq1 ) fp2 +( fq1 − fq2 ) fp3


( ) ( ) ( )

 1
− + − f 2 f+p3 − ffq3 − fq 2s f, p 3

 f f
q(2fq3 − fqq1
f
3 )Epp+ 1 ( fp1 −
f f
q 3fp3 )Eqq1+ ( fpq1 q 1 fp1 )T (11)
 t =
( ) ( ) ( )
2



 ( f − f ) f + ( f − f ) f + (
fq 3 − fq1 Ep + f p1 − f p 3 Eq + fq1 f p 3 − fq 3 f p1 Ts
q2 q3 p1 q3 q1 p2 f q1 − fq2 ) fp3


t2 =t3 = Ts − t1 − t2

, (11)
 ( ) ( ) (
f q 2 − fq 3 f p 1 + f q 3 − fq 1 f p 2 + fq 1 − f q 2 f p 3 )
where fp1 , fp2 and fp3 represent  fp1 (k + 1), fp2 (k + 1) and fp3 (k + 1), fq1 , fq2 and fq3 represent fq1 (k + 1),
t = T − t1 − t2
fq2 (k + 1) and fq3 (k + 1), Ep =3Pref s− P(k + 1), Eq = Qref − Q(k + 1).

There are six combinations  of the voltage vector sequences for OSS-MPC with two nonzero vectors
and one zero vector in one control period. Therefore, through six calculations of Equations (4) and
(8)–(11), the optimal switching sequence with minimal cost function J can be obtained. Particularly,
Electronics 2019, 8, x FOR PEER REVIEW 7 of 18

where fp1, fp2 and fp3 represent fp1(k+1), fp2(k+1) and fp3(k+1), fq1, fq2 and fq3 represent fq1(k+1), fq2(k+1) and
fq3(k+1), Ep = Pref−P(k+1), Eq = Qref−Q(k+1).
Electronics 2019, 8, 457 7 of 18
There are six combinations of the voltage vector sequences for OSS-MPC with two nonzero
vectors and one zero vector in one control period. Therefore, through six calculations of Equations
(4) (8),
the (9), (10), and
calculation (11), the
amount optimal switching
of Equation (11) in thesequence with minimal
six calculations cost function
is dramatically can be
large, Jsince obtained.
there are two
Particularly, the calculation
division operations. Figureamount
5 showsofthe
Equation (11) in the
flow diagram six calculations
of the conventionalisOSS-MPC.
dramatically large, since
there are two division operations. Figure 3 shows the flow diagram of the conventional OSS-MPC.
Start
Calculate P(k+1) and Q(k+1)
similarly as Equation (10) with
vectors calculated at instant k-1
Pref, Qref, J0=∞, i0=1

(u0, ui, ui+1)

Calculate (fp0, fpi, fp(i+1)), (fq0, fqi,

fq(i+1)) as Equation (5)

Calculate (t0, ti, ti+1) as

Equation (11)

Calculate P(k+2), Q(k+2)

as Equation (10)

Calculate J as Equation (9)

J<J0
No Yes
J0=J, i0=i

Yes
i++ i<6
No
u∈{u1, ui , u(i +1)}
0 0

t∈{t1, t2, t3}

End

Figure 5. Flow diagram of the conventional optimal-switching-sequence (OSS)-MPC.

Figure 5. Flow diagram of the conventional optimal-switching-sequence (OSS)-MPC.
3. Simplified OSS-MPC
3. Simplified OSS-MPC
To improve the efficiency of minimizing the cost function, the ideal voltage vector and the center
vectorsimprove
To the efficiency
are introduced to reduceof minimizing the cost
the calculation times function, the ideal
of Equation (11) voltage
from sixvector and
to one. the centerto
Analogous
vectors
FCS-MPC, the center vectors in the six sectors are applied to minimize the cost function instead of to
are introduced to reduce the calculation times of Equation (11) from six to one. Analogous the
FCS-MPC, the center vectors in the six sectors are applied to minimize the cost function instead of
vector sequences.
the vector sequences.
3.1. Equivalent Ideal Voltage Vector of Output Voltage Vector Sequence
3.1. Equivalent Ideal Voltage Vector of Output Voltage Vector Sequence
Since the gradient of the active and reactive power during the duration of the voltage vector is
Since theconstant,
considered gradienttheof active
the active
and and reactive
reactive powerpower duringduring
trajectories the duration of the voltage
the sampling vector
period are is
shown
considered constant, the active and reactive power trajectories during the sampling period are
as Figure 6. Equation (9), the cost function, can choose the vector sequence that minimizes the deviation shown
asbetween
Figure P(k + 2) and P(9),
6. Equation ref . the cost function, can choose the vector sequence that minimizes the
deviation between P ( k+2)
Therefore, we can use andanPref .
equivalent ideal voltage to track the reference power. When the ideal
voltage vector u*equi that can perfectly track the reference power is located in sector I, u1 and u2 in
Figure 2 are selected as the two nonzero voltage vectors of vector sequence. With a zero voltage vector,
they can equivalently generate the ideal voltage vector u*equi , as shown in Figure 7.
 P
P
u u u u u
u11 u22 u33 u22 u11
Pref
P(k+1)
x FOR PEER REVIEW P(k+1) Pref }min
Electronics 2019,8,
Electronics 2019, 8,457 P(k+2)}min 88 of
of 18
18
P(k+2)
t
00
t t t t
t11 t22 t t22 t11 t
2 2 t33 2 2
P2 2 2 2

Q TTss
Q u1 u2 u3 u2 u1
P(k+1)uu11 u u u2 u1 Pref }min
u22 u33 u2 u 1 Q
Q(k+1) QP(k+2)
ref
ref }min
Q(k+1) Q(k+2)}min
0 t1 t2 t2 t1 Q(k+2)t
t3
2 2 2
0 t12 t2 t t t
0 t1 t2 t3 t22 t11 t
2 2 t3Ts 2 2
Q 2 2 2 2
TTss
u1 u2 u3 u2 u1
Figure 6.
Figure 6. Active
Active and
and reactive
reactive trajectoriesQwithin
power trajectories
Q(k+1) power ref
within one sampling period.
}min
Q(k+2) one sampling period.

Therefore, we
Therefore, we can
can use
use an
an equivalent0 t ideal
equivalent ideal
t voltage
voltage t to track the reference
t the
tot track reference power.
power. When
When the
the ideal
ideal
2
t3 1 2 2 1

2 2 2
voltage vector u *equi that can perfectly track the reference power is located in sector I, u1 and u2 in
voltage vector u*equi that can perfectly track the Treference power is located in sector I, u1 and u2 in
s
Figure 2 are selected as the two nonzero voltage
Figure 2 are selected as the two nonzero voltage vectors of vectors of vector
vector sequence.
sequence. With
With aa zero
zero voltage
voltage
vector, they can Figure 6. Active
equivalently and reactive
generate the powervoltage
ideal trajectories within
vector u * one
equi , sampling
as shown period.
in Figure 7.
Figure 6. Active
vector, they can equivalently and reactive
generate powervoltage
the ideal trajectories within
vector u*equione
, assampling
shown inperiod.
Figure 7.
ββ
Therefore, we can use an equivalent ideal uu22 voltage to track the reference power. When the ideal
voltage vector u*equi that can perfectly track the reference power is located in sector I, u1 and u2 in
*
uuequi
Figure 2 are selected as the two nonzero voltage equi* vectors of vector sequence. With a zero voltage
vector, they can equivalently generate the ideal voltage vector u*equi, as shown in Figure 7.
ωLL
iiω
β ee
u2
u •t
u2•t2/Ts
2 2 /T s
00 uu1•t
•t1/Ts uu11 αα
1 1/Ts *
uequi
Figure 7.
Figure Equivalentideal
7. Equivalent idealvoltage
voltagevector
vectorof
of output
output vector
vector sequence.
sequence.
Figure 7. Equivalent ideal voltage vector
iωL of output vector sequence.
3.2. Center Vectors in Six Sectors e
3.2. Center
3.2. Center Vectors
Vectors in
in Six
Six Sectors
Sectors
u2•t2/Ts
In this section, the concept of the center vector is introduced.
u1 α TheThe ends of the center vectors are
In this
In this section,
section, the
the concept
concept of the0 center
of the center vector is
u1•t1/Tsvector is introduced.
introduced. The ends of the center vectors are
located at the centroids of six triangular sectors, as shown in Figure 8. ends of the center vectors are
located at the centroids of six triangular sectors, as shown
located at the centroids of six triangular sectors, as shown in Figure 8.in Figure 8.
Figure 7. Equivalent ideal voltage vector of output vector sequence.
ββ
3.2. Center Vectors in Six Sectors
uuIIII uequi**
In this section, the concept of the center vector isuequi
introduced. The ends of the center vectors are
located at the centroids of six triangularuuIII
III uuII in Figure 8.
sectors, as shown
θ
θ
u β α α
u00
uuIV
IV
uuVI
VI
uII
uequi*
uuVV
uIII uI
θ
Figure 8.
Figure Center
8. Center vectors in
u0vectors
vectors in six α
six sectors.
sectors.
Figure 8. Center in six sectors.
In Figure 8, u
Figure 8, I , uII , uIII III, ,uuIVIV , uV , VI
uVI uIVthe center vectors
are uVI of the six sectors, respectively. They are
In
In Figure 8, uuII,, uuIIII,, uuIII , uIV,, uuVV,, uuVI are
are the center
the center vectors
vectors of the
of the six
six sectors,
sectors, respectively.
respectively. They
They are
are
the finite
the finite control
control set
set (FCS)
(FCS) of of the
the cost
cost function
function optimization.
optimization.
uV
the finite control set (FCS) of the cost function optimization.
3.3. FCS Moving Horizon Optimization
3.3. FCS
3.3. FCS Moving
Moving Horizon
Horizon Optimization
Optimization Figure 8. Center vectors in six sectors.
With the definition of the equivalent ideal voltage vector and center vector, we can use them to
improve the efficiency
In Figure 8, uI, uIIof , uthe
III, umoving
IV, uV, uVIhorizon
are theoptimization. Combining
center vectors of the six Equation (4) with cost They
sectors, respectively. function,
are
the cost function to track the reference
the finite control set (FCS) of the cost function optimization. power can be rewritten as
 
3.3. FCS Moving Horizon Optimization
h i2 h i2 9Ts 2 e2α + e2β l2
J = fp∗ Ts − fpi Ts + fq∗ Ts − fqi Ts = , (12)
2L2
Electronics 2019, 8, 457 9 of 18

where l is the distance between u*equi and the voltage vectors.

As shown in Figure 8, according to cosine theorem, the square of the distance between u*equi and
the six center vectors can be calculated as
 2   2
l2 = u∗equi  + |uci |2 − 2u∗equi |uci | cos θi = (u∗α − uαi )2 + u∗β − uβi ,

(13)

where uci is the center vector, indicating uI , uII , uIII , uIV , uV and uVI .
It can be seen from Equation (13) that the distance l is determined by θ, since |u*equi | and |uci | are
constant. If u*equi is located in the sector I as shown in Figure 8, θ gets minimum with uI than the other
center vectors. Inspired by FCS-MPC, we can substitute the center vectors into the cost function to
find out which sector the ideal voltage vector u*equi is in without calculating the optimal duration time.
The cost function to obtain the sector where the ideal voltage vector u*equi is located is

2 2
J0 = [Pre f − P(k + 1) − f pi Ts ] + [Qre f − Q(k + 1) − f qi Ts ] , (14)

where 
dPi
 f pi =
 

 dt u0 + ui + ui+1
uci = . (15)

dQi 3

 f qi =


dt

Assuming that the ideal voltage vector u*equi is as shown in Figure 8. Then uI in FCS will be chosen
as the center vector that minimizes the cost function J0 . According to the above analysis, vector sequence
(u0 , u1 , u2 ) will be selected as the optimal switching sequence. After that, the duration time (t0 , t1 , t2 )
for (u0 , u1 , u2 ) will be calculated. In this way, we can avoid the duration time calculation in minimizing
the cost function. The FCS moving horizon optimization is the core idea of the simplified OSS-MPC.

3.4. Further Reduce the Computational Expense

Based on FCS moving horizon optimization of simplified OSS-MPC, we can further reduce the
computational expense by rewriting the Equation (4) as
eβ uβi

3 eα uαi dP0
 
dP


 dt = 2 L + L + dt
, (16)

 dQ 3 eβ uαi
 eα uβi dQ0

= − +


dt 2 L L dt


where dP0 /dt and dQ0 /dt are the constant term of the derivatives of the active and reactive powers,
respectively. They depend on the grid current and grid voltage, which are considered as constant values
during the sampling period. The definition of dP0 /dt and dQ0 /dt can be obtained from Equation (4) as

dP0
 −e   −eβ 


 dt = eα L + ωi
α
β + eβ L − ωi α
 . (17)

 dQ0
 −e   −eβ
= eβ L + ωiβ − eα L − ωiα
α


dt


According to the above analysis, we can extract the constant term (i.e., dP0 /dt and dQ0 /dt) of
Equation (17) outside the moving horizon optimization. In this way, Equation (17) will not need to be
calculated during moving horizon optimization. Since one cycle calculation of P(k + 1), Q(k + 1) and
6 cycle calculations of Equation (16) for the center vectors, the extraction of Equation (17) can save
6 cycle calculations of Equation (17). Meanwhile, this modification of Equation (4) during moving
horizon optimization only change the calculation order without changing the calculation results, and
will not affect the control performance. The flow diagram of the whole simplified OSS-MPC is shown
in Figure 9. During the simplified OSS-MPC, the constant term dP0 /dt and dQ0 /dt are calculated outside
the moving horizon optimization. With delay compensation, P(k + 1) and Q(k + 1) are predicted as
Equation (16) with the OSS calculated at instant k − 1. Afterwards, the center vectors are used to select
 6 cycle calculations of Equation (16) for the center vectors, the extraction of Equation (17) can save 6
cycle calculations of Equation (17). Meanwhile, this modification of Equation (4) during moving
horizon optimization only change the calculation order without changing the calculation results, and
will not affect the control performance. The flow diagram of the whole simplified OSS-MPC is shown
in Figure 9. During the simplified OSS-MPC, the constant term dP0/dt and dQ0/dt are calculated
Electronics 2019, 8, 457 10 of 18
outside the moving horizon optimization. With delay compensation, P(k+1) and Q(k+1) are predicted
as Equation (16) with the OSS calculated at instant k-1. Afterwards, the center vectors are used to
select
the OSSthe OSS through
through FCShorizon
FCS moving movingoptimization.
horizon optimization.
Eventually,Eventually, the OSS
the OSS is applied is applied
to the inverterto the
with
inverter
the optimal with the optimal
duration duration
time (t , t
1 2 3, t ) time (t
obtained1, t ,
by
2 t 3) obtained
Equation by
(11).Equation (11).

Start
Calculate dP0/dt and dQ0/dt
as Equation (17)
Calculate P(k+1) and Q(k+1)
similarly as Equation (16) with
vectors calculated at instant k-1
Pref, Qref, J0=∞, i0=1
(u0+ui+ui+1)/3
Calculate⎯ fpi, ⎯fqi as Equation
(15) and Equation (16)
Calculate J′ as Equation (14)

J′ <J0
No Yes
J0=J′, i0=i

Yes
i++ i<6
No
Calculate duration time
(t1, t2, t3) as Equation (11)

u∈{u1, ui , u(i +1)}

0 0

t∈{t1, t2, t3}

End

Figure9.9.Flow
Figure Flowdiagram
diagramof
ofsimplified
simplifiedOSS-MPC.
OSS-MPC.

Comparing
ComparingFigure
Figure99with
withFigure
Figure5,5,the
theunderlying
underlyingprinciple
principleofofthe
thesimplified
simplifiedOSS-MPC
OSS-MPCisisthat
that
itituses
uses the switching states to search the optimal switching sequence (OSS) ratherthan
the switching states to search the optimal switching sequence (OSS) rather thanswitching
switching
sequences
sequencesdirectly.
directly. Thus,
Thus, the
thesimplified
simplified OSS-MPC
OSS-MPC onlyonly needs
needs to
togo
goover
overall
all66switching
switchingstates
statesand
and
while the conventional one needs to go over all 6 switching sequences. And using
while the conventional one needs to go over all 6 switching sequences. And using switching switching sequences
means calculating
sequences means Equation
calculating(11) during 6(11)
Equation cycles,
duringwhile the simplified
6 cycles, while theOSS-MPC
simplified only need to only
OSS-MPC calculate
need
Equation (11)Equation
to calculate once after finding
(11) the OSS
once after without
finding containing
the OSS without the duration the
containing time.
duration time.
4. Simulation and Experimental Results
In order to verify the effectiveness of the simplified OSS-MPC, a comparative study with
conventional OSS-MPC is carried out. The grid side line voltage 220 V is obtained by the Y/∆ isolation
transformer. The other parameters of the simulation and experimental systems are shown in Table 1.

Table 1. Parameters of simulation and experimental system.

System Parameter Symbol Value

Filter inductance L 9 mH
Line voltage frequency f 50 Hz
DC-link capacitor C1 , C2 1000 µF
DC-link voltage Udc 350 V
Sampling frequency fs 10 kHz
 environment of MATLAB/Simulink. The active power reference steps from 0 to 2000 W at 0.02 s.
Meanwhile, the reactive power reference is held at 0 to achieve unity power factor. The proposed
algorithm not only reduces the computational burden but also completely maintains the original
excellent dynamic and steady-state response of conventional OSS-MPC. The minimum value of J is
always equal to 0 during the steady-state, since the output voltage vector sequence has the same effect
Electronics 2019, 8, 457 11 of 18
as the ideal voltage vector at instant k+2. During each sampling period, both algorithms choose the
same voltage vector sequence, and the calculated optimal duration time is the same. Since these two
algorithms
4.1. Simulationhave the same simulation results, only one of them is shown below.
Results
The dynamic simulation results of the system are presented in Figure 10. Figure 10 illustrates
Both
that the simplified
proposed OSS-MPC
algorithm hasand
fastconventional OSS-MPC
dynamic response, strategy
which are simulated
only needs 2.2 ms tointrack
the environment
the reference
of MATLAB/Simulink. The
active power as shown in Figure 10a. active power reference steps from 0 to 2000 W at 0.02 s. Meanwhile,
the reactive power
Meanwhile, reference
both is held
algorithms at 0 to excellent
express achieve unity power factor.
steady-state The proposed
performance after thealgorithm not
step change.
only reduces the computational burden but also completely maintains the
The harmonic content of the output current is presented in Figure 11. The corresponding totaloriginal excellent dynamic
and steady-state
harmonic distortion response
(THD) of conventional
of phase A output OSS-MPC.
current inThe minimum
steady-state value
is only of J when
2.33% is always equal to
the reference
0active
during the steady-state,
power since the
is 2000 W. It should beoutput
noticedvoltage
that thevector
current sequence
harmonics hasisthe sameconcentrated
mainly effect as the at
ideal
the
voltage vector at instant k + 2. During each sampling period, both algorithms
sampling frequency of 10 kHz. Figure 11b presents the harmonic content of the output current at choose the same voltage
vector
varioussequence,
operatingand the calculated
points, which haveoptimal
very low duration
THD from time1000
is the
Wsame.
to 3000Since
W. these two algorithms
have The
the same
activesimulation
and reactive results,
poweronly oneofofthe
ripple them is shown
simplified below. and conventional OSS-MPC are
OSS-MPC
The dynamic simulation results of the system are
depicted in Figure 12 under various reference active power. Figures presented in Figure
10–1210. Figurethat
indicate 10 illustrates that
the proposed
the proposed
algorithm algorithm
maintains has fast
the high dynamic response,
performance which only
of the conventional needs 2.2
OSS-MPC ms totransient
during track theandreference
steady
active
states. power as shown in Figure 10a.

Electronics 2019, 8, x FOR PEER REVIEW 12 of 18

(a)

(b)
Figure 10.
Figure 10. System response for step change of
of (a)
(a) active
active and
and reactive
reactive power;
power; (b)
(b) output
output current.
current.

Meanwhile, both algorithms express excellent steady-state performance after the step change.
The harmonic content of the output current is presented in Figure 11. The corresponding total harmonic
distortion (THD) of phase A output current in steady-state is only 2.33% when the reference active
power is 2000 W. It should be noticed that the current harmonics is mainly concentrated at the sampling
frequency of 10 kHz. Figure 11b presents the harmonic content of the output current at various
operating points, which have very low THD from 1000 W to 3000 W.

(a)

5.00%
 (b)
Electronics Figure
2019, 8, 10.
457 System response for step change of (a) active and reactive power; (b) output current. 12 of 18

(a)

5.00%
4.00% 4.64%
3.00%
THD

2.00%
2.33%
1.00% 1.55%
0.00%
1000 2000 3000
Reference Active Power (W)

(b)
Figure 11.Harmonic
Figure11. Harmoniccontent
contentofofthe
theoutput
outputcurrent (a)PP==2000
currentatat(a) 2000W;
W;(b)
(b)various
variousoperating
operatingpoints.
points.

The active and reactive power ripple of the simplified OSS-MPC and conventional OSS-MPC
are depicted in Figure 12 under various reference active power. Figures 10–12 indicate that the
proposed algorithm maintains the high performance of the conventional OSS-MPC during transient
Electronics
and steady 2019,states.
8, x FOR PEER REVIEW 13 of 18

P Q
Power ripple (W&Var)

150

133 135 126

100
103
92 91
50

0
1000 2000 3000
Reference Active Power (W)
Figure 12. Active and reactive power ripple at various operating points.
Figure 12. Active and reactive power ripple at various operating points.
4.2. Experimental Results
4.2. Experimental Results
The experimental control board is based on a 32 bit 150MHz DSP (TMS320F28335). The insulated
gate The experimental
bipolar control
transistors board
(IGBTs) usedis are
based on a 32 bit 150MHz
IKW50N60T, and the DSP
dead(TMS320F28335). The
time is set to 3µs. insulated
The output
gate bipolar
current transistors
spectrum (IGBTs)
is measured byused are IKW50N60T,
the power and the
quality analyzer dead time is set to 3μs. The output
Fluke435.
current spectrum
First is measured
of all, we measured bythe
the execution
power quality
timeanalyzer Fluke435.
of the conventional OSS-MPC and simplified
First of all, we measured the execution time of the conventional OSS-MPC
OSS-MPC. As shown in Figure 13, the execution time of conventional FCS-MPC was and18.6us.
simplified OSS-
However,
MPC. As shown in Figure 13, the execution time of conventional FCS-MPC was 18.6us. However, it
cannot achieve constant switching frequency. At the same time, the conventional OSS-MPC requires
82 μs execution time, while the simplified OSS-MPC only needs 29.8 μs. For one thing, the shorter
execution time enables the inverter to enhance the switching frequency or to extend prediction
horizons (with more than one prediction step), which can improve the steady-state performance. For
4.2. Experimental Results
The experimental control board is based on a 32 bit 150MHz DSP (TMS320F28335). The insulated
gate bipolar transistors (IGBTs) used are IKW50N60T, and the dead time is set to 3μs. The output
current spectrum is measured by the power quality analyzer Fluke435.
Electronics 2019, 8, 457 13 of 18
First of all, we measured the execution time of the conventional OSS-MPC and simplified OSS-
MPC. As shown in Figure 13, the execution time of conventional FCS-MPC was 18.6us. However, it
it cannot
cannot achieve
achieve constant
constant switching
switching frequency.AtAtthe
frequency. thesame
sametime,
time,the
theconventional
conventionalOSS-MPC
OSS-MPCrequires
requires
82 µs execution time, while the simplified OSS-MPC only needs 29.8 µs. For
82 μs execution time, while the simplified OSS-MPC only needs 29.8 μs. For one thing, the shorterone thing, the shorter
execution time
execution time enables
enablesthe
theinverter
inverterto enhance the switching
to enhance frequency
the switching or to extend
frequency or to prediction horizons
extend prediction
(with more
horizons than
(with onethan
more prediction step), which
one prediction can
step), improve
which the steady-state
can improve performance.
the steady-state For another,
performance. For
we can use the spare time to implement other functions in renewable energy sources,
another, we can use the spare time to implement other functions in renewable energy sources, such such as islanding
detection
as islanding protection
detectionorprotection
maximumorpower pointpower
maximum tracking of the
point wind turbine
tracking system.
of the wind turbine system.

Conventional 82 us
OSS-MPC

Simplified 29.8 us
OSS-MPC

Conventional
18.6 us
FCS-MPC

Figure13.
Figure Executiontime
13.Execution timeof
ofconventional
conventionalOSS-MPC,
OSS-MPC,simplified
simplifiedOSS-MPC,
OSS-MPC,and
andFCS-MPC.
FCS-MPC.

Figures14
Figures 14toand15 15 show
show the the steady-state
steady-state experimental
experimental results
results of conventional
of the the conventional
OSS-MPCOSS-MPC
and
simplified OSS-MPC. The active power is set to 2000 W while the reactive power is set toto0 0Var.
and simplified OSS-MPC. The active power is set to 2000 W while the reactive power is set Var.
Figure 14a,b depict the steady-state waveforms of the active and reactive power
Figures 14a and 14b depict the steady-state waveforms of the active and reactive power under under conventional
OSS-MPC and
conventional the simplified
OSS-MPC OSS-MPC,
and the respectively.
simplified OSS-MPC, Figure 15 shows
respectively. the steady-state
Figure 15 shows thewaveforms
steady-stateof
grid voltage and current. The steady-state experimental results show that both control
waveforms of grid voltage and current. The steady-state experimental results show that both control strategies can
track the power
strategies can track reference accurately.
the power Additionally,
reference accurately.both control strategies
Additionally, have the
both control same low
strategies THDthe
have as
shown in
Electronics Figure
2019, 8, x 16.
FOR PEER
same low THD as shown in Figure 16.REVIEW 14 of 18

P(500W/div) P(500W/div) Q(500Var/div)

Q(500Var/div)

0 0

t(10ms/div) t(10ms/div)

(a) (b)
Figure 14. Steady-state
Steady-state waveforms
waveforms of
of the
the active
active and
and reactive
reactive power
power under
under (a)
(a) conventional
conventional OSS-MPC;
OSS-MPC;
(b) simplified OSS-MPC.
OSS-MPC.

Ua(50V/div) Ua(50V/div)
Ia(5A/div) Ia(5A/div)

0 0

t(10ms/div) t(10ms/div)

(a) (b)
Figure 15. Steady-state waveforms of grid voltage and current under (a) conventional OSS-MPC; (b)
simplified OSS-MPC.
 t(10ms/div) t(10ms/div)

(a) (b)
Figure 14. Steady-state waveforms of the active and reactive power under (a) conventional OSS-MPC;
Electronics 2019, 8, 457 14 of 18
(b) simplified OSS-MPC.

Ua(50V/div) Ua(50V/div)
Ia(5A/div) Ia(5A/div)

0 0

t(10ms/div) t(10ms/div)

(a) (b)
Figure 15. Steady-state
Steady-statewaveforms
waveformsofofgrid voltage
grid and
voltage current
and under
current (a) (a)
under conventional OSS-MPC;
conventional (b)
OSS-MPC;
simplified OSS-MPC.
(b) simplified OSS-MPC.

(a) (b)
Figure 16.
16.Harmonic spectrumspectrum
Harmonic of grid current undercurrent
of grid (a) conventional
under (a)OSS-MPC; (b) simplified
conventional OSS-
OSS-MPC;
(b) simplified OSS-MPC.
MPC.

To further
To furthervalidate
validatethe
thedynamic
dynamic performance,
performance, thethe
active power
active poweris stepped fromfrom
is stepped 0 W to
0W2000
toW while
2000 W
the reactive power maintains 0 Var. Figure 17a,b presents the transient waveforms
while the reactive power maintains 0 Var. Figures 17a and 17b presents the transient waveforms of of the active and
reactive
the activepower under conventional
and reactive power underOSS-MPC
conventionalandOSS-MPC
simplifiedandOSS,simplified
respectively.
OSS,Figure 18a,b presents
respectively. Figure
the transient
18a waveforms
and 19b presents of grid voltage
the transient and current
waveforms of grid under
voltageconventional
and current OSS-MPC and simplified
under conventional OSS-
OSS-MPC, respectively. Since the experiment has different trigger phase
MPC and simplified OSS-MPC, respectively. Since the experiment has different trigger phase angle angle with simulation,
the transient
with simulation, waveforms are slightly
the transient waveforms different from the
are slightly simulation
different from theresults in Figure
simulation 10. However,
results in Figure
Figures
10. 17 and Figure
However, 18 illustrate thatFigure
17 and the simplified OSS-MPC
18 illustrate has simplified
that the the same fast dynamichas
OSS-MPC performance
the same with
fast
conventional
dynamic OSS-MPC,
performance
Electronics 2019, 8, x FOR PEER which
with can track the reference active power within 5 ms as
conventional OSS-MPC, which can track the reference active power
REVIEW shown in Figure
15 of17.
18
within 5 ms as shown in Figure 17.

P(500W/div) P(500W/div)
Q(500Var/div) Q(500Var/div)

0 0

t(10ms/div) t(10ms/div)

(a) (b)
Figure 17. Transient waveforms of the active and reactive power under (a) conventional
conventional OSS-MPC;
OSS-MPC;
(b) simplified OSS-MPC.

Ua(50V/div) Ua(50V/div)
Ia(5A/div) Ia(5A/div)

0 0

t(10ms/div) t(10ms/div)
 t(10ms/div) t(10ms/div)

(a) (b)
Figure
Electronics 2019,17. Transient waveforms of the active and reactive power under (a) conventional OSS-MPC;
8, 457 15 of 18
(b) simplified OSS-MPC.

Ua(50V/div) Ua(50V/div)
Ia(5A/div) Ia(5A/div)

0 0

t(10ms/div) t(10ms/div)

(a) (b)
Figure Transientwaveforms
Figure 18. Transient waveformsofofgrid
gridvoltage
voltage and
and current
current under
under (a) (a) conventional
conventional OSS-MPC;
OSS-MPC; (b)
(b) simplified
simplified OSS-MPC.
OSS-MPC.

5. Conclusions
5. Conclusion
In order to improve the computation efficiency of OSS-MPC, this paper proposes a simplified
In order to improve the computation efficiency of OSS-MPC, this paper proposes a simplified
OSS-MPC strategy for a grid-connected inverter. The simplified OSS-MPC improves moving horizon
OSS-MPC strategy for a grid-connected inverter. The simplified OSS-MPC improves moving horizon
optimization
optimization by by introducing
introducingthe theideal
idealvoltage
voltage vector
vector andand
thethe center
center vectors,
vectors, whichwhich is similar
is similar to
to FCS-
FCS-MPC. Besides, the equation structure of the derivatives of the active and reactive
MPC. Besides, the equation structure of the derivatives of the active and reactive powers is modified powers is
modified to further reduce the computational
to further reduce the computational expense. expense.
As
As the
the simulation
simulation and
and experiment
experiment results
results show,
show, the
the execution
execution time
time ofof the
the simplified
simplified OSS-MPC
OSS-MPC
is
is reduced by 63.7% compared to conventional OSS-MPC. Simultaneously, the proposed
reduced by 63.7% compared to conventional OSS-MPC. Simultaneously, the proposed algorithm
algorithm
maintains the superb performance of the OSS-MPC, which not only has a low
maintains the superb performance of the OSS-MPC, which not only has a low power ripple power ripple and current
and
THD, but also a very fast dynamic response. Therefore, the simplified OSS-MPC has
current THD, but also a very fast dynamic response. Therefore, the simplified OSS-MPC has a good a good application
prospect
applicationin industrial
prospect in applications with a low-cost
industrial applications withmicroprocessor.
a low-cost microprocessor.
In
In addition, the proposed algorithm is conducive to
addition, the proposed algorithm is conducive to achieving
achieving the
the longer
longer prediction
prediction horizons,
horizons,
and its extension to systems with additional levels is also under
and its extension to systems with additional levels is also under study. study.

Author Contributions: All

Author Contributions: All authors
authors contributed
contributed toto this
this work
work by
by collaboration.
collaboration. conceptualization,
conceptualization, L.K. and J.C.;
methodology, B.H.; software,
methodology, B.H.; software,J.C.;
J.C.;validation,
validation,B.H.,
B.H., J.C.;
J.C.; formal
formal analysis,
analysis, J.C.;J.C.; investigation,
investigation, B.H.;B.H.; resources,
resources, L.K.;
L.K.; data curation, X.L.; writing—original draft preparation, J.C.; writing—review and editing, X.L. and J.Z.;
data curation, X.L.; writing—original draft preparation, J.C.; writing—review and editing, X.L. and J.Z.; project
project administration, L.K.; funding acquisition, L.K.
administration, L.K.; funding acquisition, L.K.
Funding: This research was funded by Science and Technology Development Special Foundation of Guangdong,
Funding:
China This research was funded by Science and Technology Development Special Foundation of Guangdong,
(2017B010120001).
China (2017B010120001).
Acknowledgments: I would like to give my sincere gratitude to Longyun Kang, my tutor who gave me great
help during the experiments
Acknowledgments: I wouldand
likethe paper-writing.
to give my sincere gratitude to Prof. Longyun Kang, my tutor who gave me
great helpofduring
Conflicts the The
Interest: experiments and theno
authors declare paper-writing.
conflict of interest.
Conflicts of Interest: The authors declare no conflict of interest.
Nomenclature
Nomenclature
MPC Model predictive control
OSS-MPC Optimal-switching-sequence model-predictive-control
MPC Model predictive control
CCS-MPC Continuous-control-set model-predictive-control
FCS-MPC Finite-control-set model-predictive-control
DSP Digital signal processor
VOC Voltage oriented control
PI controller Proportional integral controller
DPC Direct power control
DTC Direct torque control
P-DPC Predictive direct power control
SVM Space vector modulation
D-MPC Dual vector based model-predictive-control
ROSS-DPC Reduced optimal-switching-sequence direct power control
IGBT Insulated gate bipolar transistor
Electronics 2019, 8, 457 16 of 18

e Grid voltage
V dc Direct current power supply
L Inductance of AC filter
C1 , C2 Dc-link capacitors
SX1 , SX2 Upper and lower switching states of each phase
ui (u0 , u1 , u2 , u3 , u4 , u5 , u6 ) Output voltage vector of switching state
i Grid current vector
ˆ Complex conjugate
S Apparent power
P Active power
Q Reactive power
MPPC Model predictive power control
MPCC Model predictive current control
ω Grid angular frequency
fpi , fqi Derivatives of active and reactive power
Ts Sampling period
tdelay Delay time between sampling instant and implementation
J Cost function
t1 , t2 , t3 Optimal duration time of optimal switching sequence
uci (uI , uII , uIII , uIV , uV , uVI ) Center vector
u*equi Ideal voltage vector
l Distance between u*equi and center vector
Cost function to obtain the sector where ideal voltage vector u*equi is
J0
located
f pi , f qi Derivatives of active and reactive power of center vector
dP0 /dt, dQ0 /dt Constant term of derivatives of active and reactive power
f Line voltage frequency
fs Sampling frequency
THD Total harmonic distortion

References
1. Cheng, C.; Nian, H.; Wang, X.; Sun, D. Dead-beat predictive direct power control of voltage source inverters
with optimised switching patterns. IET power Electron. 2017, 10, 1438–1451. [CrossRef]
2. Song, Z.; Tian, Y.; Chen, W.; Zou, Z.; Chen, Z. Predictive Duty Cycle Control of Three-Phase Active-Front-End
Rectifiers. IEEE Tran. Power Electron. 2016, 31, 698–710. [CrossRef]
3. Blasko, V.; Kaura, V. A new mathematical model and control of a three-phase AC-DC voltage source converter.
IEEE Trans. Power Electron. 1997, 12, 116–123. [CrossRef]
4. Malinowski, M.; Kazmierkowski, M.P.; Hansen, S.; Blaabjerg, F.; Marques, G.D. Virtual-Flux-Based Direct
Power Control of Three-Phase PWM Rectifiers. IEEE Trans. Ind. Appl. 2001, 37, 1019–1027. [CrossRef]
5. Noguchi, T.; Tomiki, H.; Kondo, S.; Takahashi, I. Direct power control of PWM converter without power-source
voltage sensors. IEEE Trans. Ind. Appl. 1998, 34, 473–479. [CrossRef]
6. Young, H.A.; Perez, M.A.; Rodriguez, J.; Abu-Rub, H. Assessing Finite-Control-Set Model Predictive Control:
A Comparison with a Linear Current Controller in Two-Level Voltage Source Inverters. IEEE Ind. Electron.
2014, 8, 44–52. [CrossRef]
7. Kouro, S.; Cortes, P.; Vargas, R.; Ammann, U.; Rodriguez, J. Model Predictive Control—A Simple and
Powerful Method to Control Power Inverters. IEEE Trans. Ind. Electron. 2009, 56, 1826–1838. [CrossRef]
8. Dragičević, T. Model Predictive Control of Power Inverters for Robust and Fast Operation of AC Microgrids.
IEEE Trans. Power Electron. 2018, 33, 6304–6317. [CrossRef]
9. Panten, N.; Hoffmann, N.; Fuchs, F.W. Finite Control Set Model Predictive Current Control for Grid-Connected
Voltage-Source Inverters with LCL Filters: A Study Based on Different State Feedbacks. IEEE Trans. Power
Electron 2016, 31, 5189–5200. [CrossRef]
10. Zhang, Y.; Peng, Y.; Yang, H. Performance Improvement of Two-Vectors-Based Model Predictive Control of
PWM Rectifier. IEEE Trans. Power Electron 2016, 31, 6016–6030. [CrossRef]
Electronics 2019, 8, 457 17 of 18

11. Nikhil, P.; Sonam, K.; Monika, M.; Wagh, S. Finite control set model predictive control for two level inverter
with fixed switching frequency. In Proceedings of the 2018 SICE International Symposium on Control
Systems (SICE ISCS), Tokyo, Japan, 9–11 March 2018; pp. 74–81.
12. Donoso, F.; Mora, A.; Cardenas, R.; Angulo-Cárdenas, A.; Sáez, D. Finite-Set Model Predictive Control
Strategies for a 3L-NPC Inverter Operating with Fixed Switching Frequency. IEEE Trans. Ind. Electron. 2018,
65, 3954–3965. [CrossRef]
13. Hu, J. Improved Dead-Beat Predictive DPC Strategy of Grid-Connected DC–AC Inverters with Switching
Loss Minimization and Delay Compensations. IEEE Trans. Ind. Inf. 2013, 9, 728–738. [CrossRef]
14. Hu, J.; Zhu, Z.Q. Improved Voltage-Vector Sequences on Dead-Beat Predictive Direct Power Control of
Reversible Three-Phase Grid-Connected Voltage-Source Inverters. IEEE Trans. Power Electron. 2013, 28,
254–267. [CrossRef]
15. Vazquez, S.; Marquez, A.; Aguilera, R.; Quevedo, D. Predictive Optimal Switching Sequence Direct Power
Control for Grid-Connected Power Inverters. IEEE Trans. Ind. Electron. 2015, 62, 2010–2020. [CrossRef]
16. Tarisciotti, L.; Zanchetta, P.; Watson, A.; Bifaretti, S.; Clare, J.C. Modulated Model Predictive Control for
a Seven-Level Cascaded H-Bridge Back-to-Back Inverter. IEEE Trans. Ind. Electron. 2014, 61, 5375–5383.
[CrossRef]
17. Tarisciotti, L.; Zanchetta, P.; Watson, A.; Clare, J.C.; Degano, M.; Bifaretti, S. Modulated Model Predictive
Control for a Three-Phase Active Rectifier. IEEE Trans. Ind. Appl. 2015, 51, 1610–1620. [CrossRef]
18. Yang, Y.; Wen, H.; Li, D. A Fast and Fixed Switching Frequency Model Predictive Control with Delay
Compensation for Three-phase Inverters. IEEE Access. 2017, 5, 17904–17913. [CrossRef]
19. Zhang, Y.; Xie, W.; Li, Z.; Zhang, Y. Low-Complexity Model Predictive Power Control: Double-Vector-Based
Approach. IEEE Trans. Ind. Electron. 2014, 61, 5871–5880. [CrossRef]
20. Zhang, Z.; Hackl, C.M.; Kennel, R. Computationally Efficient DMPC for Three-Level NPC Back-to-Back
Inverters in Wind Turbine Systems with PMSG. IEEE Trans. Power Electron. 2017, 32, 8018–8034. [CrossRef]
21. Bozorgi, A.M.; Gholami-Khesht, H.; Farasat, M.; Mehraeen, S.; Monfared, M. Model Predictive Direct Power
Control of Three-Phase Grid-Connected Inverters with Fuzzy-Based Duty Cycle Modulation. IEEE Trans.
Ind. Appl. 2018, 54, 4875–4885. [CrossRef]
22. Yang, Y.; Wen, H.; Fan, M.; Xie, M.; Chen, R. Fast Finite-Switching-State Model Predictive Control Method
Without Weighting Factors for T-Type Three-Level Three-Phase Inverters. IEEE Trans. Ind. Inf. 2019, 15,
1298–1310. [CrossRef]
23. Mariethoz, S.; Morari, M. Explicit Model-Predictive Control of a PWM Inverter with an LCL Filter. IEEE Trans.
Ind. Electron. 2009, 56, 389–399. [CrossRef]
24. Beccuti, A.G.; Kvasnica, M.; Papafotiou, G.; Morari, M. A Decentralized Explicit Predictive Control Paradigm
for Parallelized DC-DC Circuits. IEEE Trans. Control Syst. Technol. 2013, 21, 136–148. [CrossRef]
25. Xia, C.; Tao, L.; Shi, T.; Song, Z. A Simplified Finite-Control-Set Model-Predictive Control for Power
Converters. IEEE Trans. Ind. Inf. 2014, 10, 991–1002.
26. Liu, X.; Wang, D.; Peng, Z. Improved finite-control-set model predictive control for active front-end rectifiers
with simplified computational approach and on-line parameter identification. ISA Trans. 2017, 69, 51–64.
[CrossRef]
27. Kwak, S.; Park, J.C. Switching Strategy Based on Model Predictive Control of VSI to Obtain High Efficiency
and Balanced Loss Distribution. IEEE Trans. Power Electron. 2014, 29, 4551–4567. [CrossRef]
28. Rodriguez, J.; Kazmierkowski, M.P.; Espinoza, J.R.; Zanchetta, P.; Abu-Rub, H.; Young, H.A.; Rojas, C.A.
State of the Art of Finite Control Set Model Predictive Control in Power Electronics. IEEE Trans. Ind. Inf.
2013, 9, 1003–1016. [CrossRef]
29. Hu, B.; Kang, L.; Cheng, J.; Zhang, Z.; Zhang, J.; Luo, X. Double-step model predictive direct power control
with delay compensation for three-level converter. IET Power Electron. 2019, 12, 899–906. [CrossRef]
Electronics 2019, 8, 457 18 of 18

30. Ngo, V.-Q.-B.; Nguyen, M.-K.; Tran, T.-T.; Choi, J.-H.; Lim, Y.-C. A Modified Model Predictive Power Control
for Grid-Connected T-Type Inverter with Reduced Computational Complexity. Electronics 2019, 8, 217.
[CrossRef]
31. Aguilera, R.P.; Quevedo, D.E.; Vazquez, S.; Franquelo, L.G. Generalized Predictive Direct Power Control
for AC/DC converters. In Proceedings of the IEEE ECCE Asia Downunder (ECCE Asia), Melbourne, VIC,
Australia, 3–6 June 2013; pp. 1215–1220.

© 2019 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access
article distributed under the terms and conditions of the Creative Commons Attribution
(CC BY) license (http://creativecommons.org/licenses/by/4.0/).