Int
ern
at
i
onal
Journ
al of
P
ower E
le
ctr
on
i
cs a
n
d
Drive
S
ystem
(I
J
PE
D
S
)
Vo
l.
11
,
No.
4
,
Decem
be
r
2020
, p
p.
1719
~
17
30
IS
S
N:
20
88
-
8694
,
DOI: 10
.11
591/
ij
peds
.
v
1
1
.i
4
.
pp
17
19
-
17
30
1719
Journ
al h
om
e
page
:
http:
//
ij
pe
ds
.i
aescore.c
om
Hardw
are in
the loop
c
o
-
s
imulati
on of fi
nite set
-
m
odel
predicti
ve co
ntr
ol u
si
ng F
PGA for
a thr
ee level CH
B invert
er
Ma
i
V
an
Chu
ng
1
,
D
o
Tu
an
An
h
2
,
Ph
uong
V
u
3
,
M
anh Li
nh Ng
uyen
4
1
,2,3,
4
School
of E
le
c
tri
c
al E
ngin
eering
,
Hano
i
un
iv
ersit
y
of
sci
enc
e
and
t
ec
hnology
,
No.1,
Da
i
Co
Vi
et
,
Hai
Ba
Trung
,
Ha
Noi,
Viet Na
m
1
Hung Vuong Unive
rsity
,
Nguye
n
Tat
Tha
nh,
No
ng
Tra
ng
,
Vi
et Tri,
Phu
Tho, Viet
Nam
Art
ic
le
In
f
o
ABSTR
A
CT
Art
ic
le
history:
Re
cei
ved
Feb
9
, 2
0
20
Re
vised
M
a
y
26
, 2
0
20
Accepte
d
J
un
20
, 2
020
Along
with
the
dev
el
opm
e
nt
of
powerf
ul
mi
cro
pro
ces
sors
and
mi
cro
cont
rol
le
rs
,
the
appl
i
ca
t
ion
s
of
the
mode
l
pre
dictive
cont
r
oll
er
,
which
req
uire
s
high
co
mput
ational
cost
,
to
f
ast
dynami
ca
l
sys
tems
suc
h
as
power
conve
rt
ers
and
e
le
c
tri
c
driv
es
ha
ve
be
come
a
t
en
denc
y
r
ecent
ly
.
I
n
thi
s
p
ape
r
,
two
soluti
ons
ar
e
offe
r
ed
to
qu
i
ckl
y
dev
el
op
the
fini
t
e
set
pr
edic
ti
ve
cur
ren
t
cont
rol
for
indu
ct
ion
mot
or
fed
by
3
-
le
v
el
H
-
Bri
dge
c
asc
ad
ed
in
ver
te
r
.
First
,
the
field
prog
r
am
m
abl
e
g
at
e
arr
ay
(FP
GA
)
with
ca
pab
il
i
ty
of
par
al
l
el
com
put
at
ion
is
em
ploy
ed
to
m
i
nim
ize
th
e
computat
ion
al
ti
m
e.
Second,
th
e
har
dware
in
the
loop
(HIL)
co
-
simul
ation
is
u
sed
to
qu
ic
kly
ver
ify
th
e
deve
lop
ed
con
tr
ol
al
gor
it
hm
wi
thout
burde
n
of
ti
m
e
on
h
ard
ware
d
esign
since
th
e
mot
or
and
the
powe
r
sw
it
che
s
ar
e
e
mu
la
t
ed
on
a
r
ea
l
-
time
pl
at
form
with
high
-
fid
el
i
t
y
ma
th
em
a
ti
c
al
mode
ls.
The
implementation
pr
oce
dure
and
HIL
co
-
simu
lati
on
result
s
of
th
e
dev
el
op
ed
co
ntrol
al
gori
thm
show
s
the
eff
ective
n
ess of the
pro
posed
sol
uti
on.
Ke
yw
or
d
s
:
Ca
scaded H
-
bri
dg
e i
nv
et
e
r
FPGA
Hardwa
re i
n
th
e loop
Ind
uction m
otor c
on
tr
ol
Pr
e
dicti
ve
cu
rrent co
ntr
ol
This
is an
open
acc
ess arti
cl
e
un
der
the
CC
BY
-
SA
l
ic
ense
.
Corres
pond
in
g
Aut
h
or
:
M
a
nh Lin
h Nguy
e
n
,
P
huong
Vu
School
of Elec
tric
al
Engineer
ing
Hanoi
Un
i
ver
si
ty of
Scie
nce a
nd Tec
hnolog
y
No.1,
Dai Co
Viet
Roa
d,
Hai
Ba Tr
ung,
Ha
no
i,
Viet
Nam
Emai
l:
li
nh
.
nguy
e
nm
a
nh@
hu
st.ed
u.vn
,
phuo
ng.vu
ho
a
ng@
hust.e
du.vn
1.
INTROD
U
CTION
Nowa
day
s
,
m
os
t
co
mmercia
l
inv
erter
s
f
or
AC
mac
hin
es
are
instal
le
d
with
fiel
d
-
ori
e
nted
-
co
ntr
ol
(F
OC
).
T
o
f
ur
t
her
im
pro
ve
t
he
dy
namic
res
pons
e
of
the
F
OC,
ne
w
co
ntr
ol
st
rategies
ha
ve
bee
n
stu
di
ed
f
or
the
c
urren
t
co
nt
ro
l
l
oop
su
c
h
as
s
yn
c
hro
nous
vect
or
c
on
t
ro
l,
sta
te
fee
dbac
k
c
ontr
ol,
dea
dbeat
c
on
t
ro
l,
ne
ur
al
netw
ork
a
nd
f
uzzy
co
ntr
ol
[
1].
A
mon
g
th
em,
m
odel
pre
dicti
ve
co
ntr
ol
(MPC)
has
be
en
c
on
si
der
e
d
as
a
powe
rful
al
te
r
native
c
ontr
ol
method
f
or
power
c
onver
te
rs
an
d
el
ect
rical
dr
i
ves
[
2
-
8].
T
he
M
PC
,
w
hich
wa
s
early d
e
velo
pe
d
i
n
t
he
1960s,
is
a
n
a
ppli
cat
i
on o
f
the
opti
mal
co
ntr
ol
th
e
ory
[
2]
,
in
w
hi
ch
t
he
s
ys
te
m
model
is
us
e
d
to
pr
e
dic
t
the
fu
t
ur
e
be
hav
i
or
of
the
sy
ste
m
sta
te
s
i
n
a
pr
e
def
i
ned
ti
me
horizo
n
and
the
n
obta
in
t
he
op
ti
mal
co
ntr
ol
act
ion
w
hich
minimi
zes
a
giv
e
n
cost
fun
ct
ion
.
Des
pite
of
a
dv
a
ntag
e
s
su
c
h
as
intu
it
ive
con
ce
pt,
quic
k
dyna
mic
res
ponse
,
a
bili
ty
to
handle
c
onstrai
ne
d
li
nea
r
a
nd
no
nliner
m
ulti
var
ia
ble
dy
namic
sy
ste
ms
[3],
t
he
ap
plica
ti
on
of
M
PC
is
re
stric
te
d
in
the
fiel
d
of
proce
s
s
con
t
ro
l
due
t
o
it
s
relat
ively
high
com
pu
ta
ti
onal
cost.
I
n
rece
nt
yea
rs,
with
th
e
de
velo
pme
nt
of
of
high
-
spe
ed
micr
ocontr
oller
a
nd
FP
G
A,
the
high
co
mputat
ion
al
c
os
t
pro
blem
of
th
e
M
P
C
can
be
so
l
ve
d
f
ollow
i
ng
t
ha
t
the
ap
plica
ti
on
of
the
M
P
C
has
been
e
xpan
de
d
to
fa
st
dyna
m
ic
al
sy
ste
ms
s
uch
as
powe
r
el
ect
ro
nic
c
onver
te
r
s
a
nd
el
ect
rical
dri
ves.
Du
e
t
o
the
fact
that
the
po
wer
el
ec
tro
nic
co
nv
e
rters
only
ha
ve
finite
switc
hi
ng
sta
te
s,
finit
e
con
t
ro
l
set
model
pr
e
dicti
ve
co
nt
ro
l
(
FCS
-
MPC
)
has
bee
n
devel
op
e
d
to
sel
e
ct
the
switc
hi
ng
sta
te
f
or
t
he
conve
rters
dir
ect
ly
Evaluation Warning : The document was created with Spire.PDF for Python.
IS
S
N
:
2088
-
8
694
In
t J
P
ow
Ele
c
&
Dr
i
S
ys
t
,
V
ol
.
1
1
, N
o.
4
,
D
ecembe
r
2020
:
17
19
–
17
30
1720
instea
d of u
sin
g
a
mod
ulator
wh
ic
h
is al
s
o
c
ompli
cat
ed,
es
pecial
ly in m
ulti
level i
nv
e
rter
s.
T
he
FCS
-
MPC
h
as
been
s
uccess
ful
ly
ap
plied
to
t
he
c
urre
nt
c
ontrol
i
n
t
hr
ee
-
phase
i
nverters
[9
-
14
],
matri
x
conve
rter
[
15]
.
A
n
d
inducti
on
mac
hin
e
[
16
-
18
].
An
imp
r
ov
e
d
FCS
-
M
PC
is
a
lso
propose
d
t
o
s
olv
e
the
vari
able
switc
hi
ng
pro
bl
e
m
ininducti
on
mot
or
(IM
)
fe
d b
y 2
-
le
vel
vo
lt
ag
e sour
ce
in
ver
t
er (VS
I) [
19].
M
ulti
le
vel
c
onver
te
r
s
is
wide
ly
us
ed
f
or
me
diu
m
-
vo
lt
age
(
HV)
a
nd
high
powe
r
a
pp
li
cat
ion
s
due
t
o
it
s
rema
rk
a
ble
ad
va
ntages
over
the
co
nven
ti
on
al
tw
o
-
le
ve
l
VSI
s
uc
h
a
s:
lo
wer
volt
age
stress
,
lo
we
r
r
at
e
of
vo
lt
age
ch
a
nge
(
dv/dt
)
, red
uce
d
total
har
m
oni
c d
ist
ort
ion (T
HD) a
nd lo
wer switc
hing
fr
e
quenc
y w
hich re
su
lt
s
in
imp
r
ov
e
d
s
witc
hing
l
os
s
[
9,20].
A
mon
g
the
m
ulti
le
vel
inv
e
rter
t
opol
ogie
s,
t
he
C
HB
is
the
m
os
t
s
uc
cessf
ul
config
ur
at
io
n
and
al
rea
dy
c
om
me
rcial
iz
ed
beca
us
e
of
it
s
mod
ularit
y
[
21
-
23].
T
ypic
al
ly,
a
m
otor
con
t
ro
l
sy
ste
m
c
onsist
s
of
a
c
on
ve
ntion
al
c
ontr
oller
su
c
h
as
pro
po
rtion
al
-
i
nteg
ral
-
dif
fer
e
ntial
(PID
),
sta
te
feedback
,
et
c.,
a
nd
a
pul
se
wi
dth
m
odul
at
ion
(PWM
).
F
or
m
ulti
le
vel
in
ver
te
rs
,
t
he
higher
the
le
ve
l,
the
m
or
e
co
mp
le
x
the
P
W
M
al
gorith
m
is,
e
spe
ci
al
ly
when
oth
e
r
in
her
e
nt
pro
blems
of
the
m
ulti
le
vel
topolo
gies
s
uc
h
as
capaci
tor
volt
age
bala
nce,
fa
ul
t
tolera
nt
a
bili
ty
a
re
co
ns
ide
r
ed. Hen
ce
,
t
his
re
searc
h
f
ocus
es o
n
t
he
a
ppli
cat
ion
of
FCS
-
M
PC
t
o
I
M
fed
by
m
ulti
le
vel
casca
ded
H
-
bri
dge
(
CHB)
in
ver
te
r
s.
I
ns
te
ad
of
usi
ng
t
he
co
nve
ntion
al
P
W
M
,
the
volt
age
vecto
r
w
hich
minimi
zes
t
he
pr
e
dicti
ve
c
urren
t
tracki
ng
er
ror
is
sel
ect
ed
directl
y
in
ever
y
consecuti
ve
sa
mp
li
ng
c
ycle.
Without
P
W
M
al
gorithm
a
nd
by
us
i
ng
FP
G
A
[
24
-
26]
as
t
he
co
ntro
ll
er
,
th
e
FCS
-
M
PC
al
go
rith
m
can
ea
sil
y
be
s
ov
le
in
ve
ry
s
hor
t
ti
me
f
ollow
i
ng
that
the
res
ponse
of
the
in
ner
c
urr
ent
is
sign
ific
a
ntly
imp
rove
d
i
n
co
mp
a
rison
with
the
c
onve
ntio
na
l
PID
co
ntr
oller.
Be
sides
,
m
ulti
ple
ob
je
ct
iv
es
ca
n
al
so
be
ac
hiev
ed
by
ap
pro
pr
i
at
el
y
ch
oosin
g
the
co
st
f
un
ct
i
on.
T
o
a
void
wasti
ng
ti
me
on
ha
r
dware
des
ign
a
nd
the
ris
ky
e
xp
e
riments
with
hi
gh
po
wer
sys
te
ms,
HI
L
platfo
rm
ma
nufact
ur
e
d
by
T
yphoon
is
em
ploy
ed
t
o
qu
ic
kly
ve
rif
y
the d
evel
op
e
d
con
t
ro
l
al
gorithm.
By
usi
ng
t
he
T
ypho
on
H
I
L,
the
b
eha
vior
of
the
m
otor
a
nd
the
CHB
inv
e
rter
are
preci
sel
y
eval
uated
i
n
real
-
ti
me
by
hi
gh
-
fi
delit
y
mathe
mati
cal
mo
dels
.
Re
al
ti
me
simulat
ion
s
with
va
rio
us
scen
arios
a
re
c
onduct
ed
an
d
t
he
r
esults
s
how
t
ha
t
the
em
ploye
d
FC
S
-
M
PC
a
pp
li
ed
to
CHB
-
fe
d
I
M
dr
i
ve
ca
n
achieve
go
od
performa
nce
not
only
in
te
r
ms
of
tracki
ng
acc
ur
ac
y
bu
t
al
so
dynamic
res
pons
e.
2.
SY
STE
M DESC
RIPTIO
N
2.1.
Ov
er
view o
f CHB in
vert
er
The
ty
pical
c
onfig
ur
at
io
n
of
a
th
ree
-
pha
se
m
ulti
le
vel
CHB
is
sho
w
n
i
n
Fig
ure
1
(
a
)
.
T
he
fun
dame
ntal
c
ompone
nt
of
t
he
CHB
is
a
s
ing
le
-
phase
H
-
br
i
gd
e
in
ve
rter
w
hic
h
is
nor
mall
y
cal
l
ed
a
po
wer
cel
l. Each
phas
e of the
CHB c
on
sist
s
of se
ve
ral cel
ls co
nne
ct
ed
in
series.
HB
-
A
1
HB
-
A
2
HB
-
An
Z
A
Z
A
Z
B
Z
C
B
C
HB
-
B
1
HB
-
B
2
HB
-
Bn
HB
-
C
1
HB
-
C
2
HB
-
Cn
(
a)
v
dc
v
ac
S
2
S
1
S
3
S
4
(
b)
Figure
1
.
(a
)
St
ru
ct
ur
e
of a C
HB
in
ver
te
r; (
b) T
opolog
y of
a cel
l
Assume
that
each
phase
c
onsist
s
of
n
ce
ll
s
con
necte
d
in
series
an
d
each
cel
l
is
fed
with
a
n
ind
e
pende
nt d
c
volt
age
s
ource
V
dc.
T
he
powe
r
s
witc
hes
S
1
-
S
4
in
cel
l
i
of
a
phase
ca
n
be
c
ontr
olled
to
ge
ne
rate
Evaluation Warning : The document was created with Spire.PDF for Python.
In
t J
P
ow Elec
& Dri S
ys
t
IS
S
N:
20
88
-
8
694
Ha
r
dw
ar
e i
n
t
he
lo
op
c
o
-
si
m
ula
ti
on
of fi
nite set
-
model
pre
dicti
ve co
ntrol
u
sin
g
F
PGA
…
(
Mai
Van C
hung
)
1721
three
vo
lt
age
le
vels
{+V
dc
,
0,
-
V
dc
}
c
orres
pondin
g
t
o
cel
l
sta
te
S
xi
={1,
0,
-
1}
w
her
e
x
re
pr
ese
nts
ph
ase
a,
b
or
c
. T
he vo
lt
age
le
vel of a
phas
e is de
fine
d
as
the s
um
of all
cel
l st
at
es in tha
t ph
a
se as
f
ollow
i
ng
:
0
n
x
x
i
i
SS
(1)
The n
umber o
f
possible
volt
age levels
for o
ne ph
a
se is:
21
mn
=+
(2)
The o
utput v
ol
ta
ge
of eac
h p
ha
se is cal
culat
e
d by
0
0
0
;;
n
n
n
A
N
d
c
a
i
B
N
d
c
b
i
C
N
d
c
c
i
i
i
i
v
V
S
v
V
S
v
V
S
(3)
2.
2.
Spa
ce
volt
age
vector
of 3
-
le
ve
l CHB in
verte
r
A
volt
age
vect
or
is
f
ormed
by
a
c
ombinati
on
of
s
witc
hi
ng
sta
te
s.
T
he
t
ota
l
com
bi
nations
of
the
CHB
inv
e
rter is as
foll
ow
:
3
(
2
1
)
m
Kn
=+
(4)
2
1
2
6
1
v
K
n
n
=
+
+
(5)
In
mu
lt
il
evel
CHB
co
nverte
rs,
sa
me
vo
lt
a
ge
vecto
r
ca
n
be
im
plemente
d
by
seve
ral
c
ombinati
on
of
switc
hing
sta
te
s.
He
nce,
the
numb
e
r
of
volt
age
vect
or
s
ar
e
normall
y
le
s
s
than
the
tota
l
switc
hing
sta
te
s
as
fo
ll
owin
g
[
9].
In
this
re
searc
h,
a
th
ree
phase
thr
ee
le
vel
CH
B
in
ver
te
r
is
us
ed
with
2
7 po
s
sible vo
lt
age
vec
tors
in
sta
ti
c
αβ
co
ordinate.
By
el
imi
nating
the
s
witc
hing
sta
te
s
w
hich
ge
ner
at
e
high
co
mm
on
-
mode
volt
ag
e,
the
numb
e
r
of
em
ployed
volt
age
vecto
rs
a
re
r
edu
ce
d
t
o
19
as
sho
wn
in
Fig
ure
2.
The
se
volt
age
vet
or
are
employe
d
i
n
th
e FCS
-
M
PC
in
the
nex
t
sect
io
n.
(
1
,
-
1
,
-
1
)
V
7
(
1
,
1
,
-
1
)
V
9
(
-
1
,
1
,
-
1
)
V
11
(
-
1
,
1
,
1
)
V
13
(
-
1
,
-
1
,
1
)
V
15
(
1
,
-
1
,
1
)
V
17
(
0
,
-
1
,
-
1
)
(
1
,
0
,
0
)
V
1
(
1
,
1
,
0
)
(
0
,
0
,
-
1
)
V
2
(
-
1
,
0
,
-
1
)
(
0
,
1
,
0
)
V
3
(
0
,
1
,
1
)
(
-
1
,
0
,
0
)
V
4
(
-
1
,
-
1
,
0
)
(
0
,
0
,
1
)
V
5
(
1
,
0
,
1
)
(
0
,
-
1
,
0
)
V
6
(
1
,
0
,
-
1
)
V
8
(
0
,
1
,
-
1
)
V
10
(
-
1
,
1
,
0
)
V
12
(
-
1
,
0
,
1
)
V
14
(
0
,
-
1
,
1
)
V
16
(
1
,
-
1
,
0
)
V
18
(
0
,
0
,
0
)
(
1
,
1
,
1
)
(
-
1
,
-
1
,
-
1
)
V
0
2
1
Figure
2. S
pac
e volt
age
vecto
r of
a
3
-
le
vel C
HB
in
ver
te
r
2.3.
Model
ing
of the i
nduc
tio
n
motor
In stat
ion
a
ry fr
ame
α
β, t
he be
hav
i
or of the
IM ca
n be
repre
sented
by t
he
f
ollow
i
ng equat
ion
s:
;0
s
s
s
s
s
s
s
r
s
s
s
r
r
p
r
d
d
R
R
j
z
d
t
d
t
=
+
=
+
−
Ψ
Ψ
u
Ψ
ii
(6)
Evaluation Warning : The document was created with Spire.PDF for Python.
IS
S
N
:
2088
-
8
694
In
t J
P
ow
Ele
c
&
Dr
i
S
ys
t
,
V
ol
.
1
1
, N
o.
4
,
D
ecembe
r
2020
:
17
19
–
17
30
1722
s
s
s
s
s
s
s
s
s
m
r
r
m
s
r
r
L
L
L
L
Ψ
;
Ψ
=
+
=
+
i
i
i
i
(7)
I
n w
hich
s
s
r
r
R
L
R
L
,,
are r
e
sist
ance and i
nducta
nce
of the
stat
or
a
nd
ro
t
or,
res
pecti
vely.
The
c
urre
nt and
flu
x of
t
he
st
at
or
a
nd roto
r
a
re e
xpr
esse
d
in
v
ect
or
form
as
foll
ow
:
[
]
[
]
[
]
[
]
[
]
s
T
s
T
s
T
s
T
s
T
s
s
s
s
s
s
r
r
r
s
s
s
r
r
r
u
u
i
i
i
i
=
=
=
=
=
u
;
i
;
i
(8)
Def
i
ning
new
var
ia
bles as:
2
1
;
,
;
1
ss
sm
r
r
m
r
r
s
m
r
s
s
r
LL
L
TT
L
C
R
L
L
=
=
=
=
−
(9)
A
f
unda
me
ntal mani
pu
la
ti
on
on (1) an
d (
2)
r
esults i
n:
11
22
s
s
s
s
s
s
s
r
m
s
ss
rm
r
m
s
d
i
dt
d
dt
=
+
+
=+
i
A
B
u
P
A
B
i
(10)
with
1
1
2
2
1
1
1
1
(
1
)
1
1
(
)
0
0
0
,,
1
1
1
(
1
)
1
1
1
0
(
)
0
0
s
r
s
r
r
r
s
r
s
r
r
r
T
T
L
T
T
T
T
T
L
T
T
T
−
−−
−+
−−
=
=
=
=
=
−
−
−
−
+
−
−
A
B
,
P
,
A
B
Fo
r
MPC
co
ntr
ol
desi
gn,
the
con
ti
nu
ous
-
ti
m
e
model
(
10)
ne
eds
to
be
t
ransforme
d
int
o
di
screte
-
ti
me
model wit
h sa
mp
li
ng
per
i
od
T
s
by
us
i
ng forwar
d
-
Euler
me
thod as
foll
ow
i
ng
:
,
1
1
,
1
,
,
,
1
2
,
2
,
s
s
s
s
s
k
s
k
s
k
k
r
m
k
s
s
s
r
m
k
r
m
k
s
k
+
+
=
+
+
=+
i
i
u
D
i
(11)
in which
1
2
1
2
2
2
1
1
2
2
2
10
01
s
s
s
s
k
s
T
T
T
TT
=
+
=
+
=
=
=
=
I
A
I
A
B
B
;
D
P
;
I
Ba
sed on (
11),
the pre
dicti
ve c
urren
t i
n N st
ep
a
head ca
n b
e comp
uted b
y:
11
,
1
,
1
2
1
1
1
1
1
,
2
,
2
,
12
12
1
1
1
1
1
1
1
1
1
,,
0
.
.
.
0
0
.
.
.
0
.
.
.
0
.
.
.
0
.
.
.
.
.
.
...
...
ss
k
s
k
s
k
ss
kk
s
k
s
k
s
sk
NN
N
N
N
ss
k
k
k
s
k
N
s
k
N
++
++
−−
−−
++
++
=
+
+
D
iu
DD
iu
i
D
D
D
iu
,
,1
1
,1
...
s
r
m
k
s
r
m
k
s
N
r
m
k
N
+
−
+−
(12)
3.
CONTR
OL
D
ESIGN
3.1.
FCS
-
M
PC
des
ign
The
blo
c
k
diagr
a
m
of
the
I
M
c
on
t
ro
l
s
ys
t
em
is
sho
wn
i
n
Fig
ure
3
(
a
)
wh
ic
h
c
on
ist
s
of
t
wo
oute
r
loops
i
nclu
ding
a
spe
ed
re
gu
la
tor,
a
fl
ux
re
gu
la
to
r
a
nd
a
n
inn
e
r
c
urren
t
loop
us
i
ng
FCS
-
MPC
.
T
he
tw
o
ou
te
r
loops
desi
gne
d
i
n
sync
hro
nous
ref
e
re
nce
f
rame
dq
ge
ner
at
e
the
re
f
eren
ce
sta
to
r
currents
.
The
n,
t
hese
ref
e
ren
ce
c
urre
nts
a
re
t
ran
s
f
orme
d
i
nto
α
β
c
oor
din
at
e
a
nd
us
e
d
for
the
F
CS
-
MPC
desi
gn.
Ba
sed
on
th
e
I
M
Evaluation Warning : The document was created with Spire.PDF for Python.
In
t J
P
ow Elec
& Dri S
ys
t
IS
S
N:
20
88
-
8
694
Ha
r
dw
ar
e i
n
t
he
lo
op
c
o
-
si
m
ula
ti
on
of fi
nite set
-
model
pre
dicti
ve co
ntrol
u
sin
g
F
PGA
…
(
Mai
Van C
hung
)
1723
model,
the
opti
mal
vo
lt
age
ve
ct
or
is
c
ho
s
en
to
minimi
ze
a
cost
f
unct
ion
in
w
hich
one
or
se
ve
ral
obje
ct
ives
may
be
c
on
si
de
red
s
uc
h
as:
minimal
tracki
ng
e
rro
r,
ca
pa
ci
tor
vola
ge
ba
la
nce,
mi
n
im
al
switc
hing
lo
ss,
et
c.
The
detai
l
FC
S
-
M
PC
al
gorithm
w
hich
is
act
ivate
d
i
n
e
very
co
ns
ec
utive
sam
pling
per
i
od
is
sho
wn
in
Fig
ure
3
(
b
)
,
i
n detai
ls.
w
*(
k
)
Ψ
rd
(
k
)*
w
(
k
)
+
-
S
p
e
e
d
c
o
n
t
r
o
l
l
e
r
i
s
q
*
F
l
u
x
c
o
n
t
r
o
l
l
e
r
i
s
d
*
i
αβ
(
k
)*
F
l
u
x
m
o
d
e
l
α
β
d
q
Ψ
r
_
α
β
(
k
)
i
αβ
(
k
)
i
αβ
(
k
)
w
(
k
)
θ
s
w
(
k
)
S
a
,
S
b
,
S
c
C
o
s
t
f
u
n
c
t
i
o
n
P
r
e
d
i
c
t
i
v
e
c
u
r
r
e
n
t
a
t
i
n
s
t
a
n
t
k
+
2
i
αβ
(
k
+
2
)
F
C
S
-
M
P
C
+
Ψ
rd
(
k
)
-
IM
c
b
a
v
dc
v
dc
v
dc
N
3
-
l
e
v
e
l
c
o
n
v
e
r
t
e
r
C
H
B
a
b
c
i
a
α
β
IE
i
b
i
c
A
p
p
l
y
o
p
ti
m
al
v
o
l
tag
e
v
e
c
to
r
V
(
k
)
S
tar
t
Kn
o
w
n
f
r
o
m
o
u
te
r
l
o
o
p
C
al
c
u
l
a
te
s
tato
r
c
u
r
r
e
n
t
F
l
u
x
r
o
to
r
,
a
n
g
l
e
C
a
l
c
u
l
a
te
r
e
f
e
r
e
n
c
e
c
u
r
r
e
n
t
x
=
0
x
=
x
+
1
Me
as
u
r
e
S
to
r
e
o
p
ti
m
a
l
v
a
l
u
e
x
<
19
Y
N
Wa
i
ti
n
g
f
o
r
n
e
x
t
s
am
p
l
i
n
g
i
n
s
tan
t
(a)
(
b)
Figure
3. FCS
con
t
ro
l st
rateg
y
,
(
a)
IM
c
ontr
ol sc
heme,
(
b)
Pr
e
dicti
ve
cu
rrent co
ntr
ol alg
or
it
hm
3.2.
Co
s
t func
tio
n
sel
ection
In
this
resea
rc
h,
the
cost
f
unct
ion
is
ch
os
e
n
base
d
on
the
predict
ive
cu
rr
e
nt
er
r
or
in
αβ
c
oor
din
at
e
of
the in
du
ct
i
on
mo
to
r.
3.2.
1.
Form
ulat
i
on
of the cu
rrent err
or
The
e
rror
between
the
predi
ct
ive
cu
rr
e
nt
a
nd
it
s
re
fer
e
nc
e
can
be
ex
pre
ssed
a
s
a
n
a
bsolt
ute
value,
s
quare
value
or
integ
ral
valu
e
[3].
If
only
the
pr
ese
nt
trac
king
er
r
or
is
use
d
by
the
c
ost
functi
on,
a
bsolute
error
a
nd
squa
re
er
ror
gi
ve
s
ame
pe
rforma
nce.
W
he
n
t
he
pr
ese
nt
a
nd
pa
st
values
of
t
he
trac
king
er
r
or
a
re
consi
der
e
d,
the
square
of
e
rro
r
gi
ves
bette
r
t
rack
i
ng
pe
rformance
t
han
the
abs
olu
te
e
rror
wh
il
e
the
integ
ral
of
error
gi
ves
t
he
best
performa
nc
e.
Howe
ver
,
a
cost
f
un
ct
io
n
with
i
ntegr
al
of
trac
king
e
rro
r
al
so
re
qu
i
res
hi
gh
e
r
com
pu
ta
ti
onal
cost.
Evaluation Warning : The document was created with Spire.PDF for Python.
IS
S
N
:
2088
-
8
694
In
t J
P
ow
Ele
c
&
Dr
i
S
ys
t
,
V
ol
.
1
1
, N
o.
4
,
D
ecembe
r
2020
:
17
19
–
17
30
1724
3.2.
2.
Delay c
om
pens
at
i
on
In
the
ideal
c
ase,
th
e
ti
me
need
e
d
f
or
ca
lc
ulati
on
is
ne
gligible.
A
ssume
that
the
c
urren
ts
are
measu
red
at
ti
me
insta
nce
t
k
,
the
opti
mal
vo
lt
age
vecto
r
that
mi
nimize
s
the
e
r
ror
at
ti
me
instance
t
k+1
is
chosen
a
nd
a
pp
li
ed
imme
diate
ly
at
t
k
.
T
her
e
fore,
t
he
l
oad
cu
rr
e
nt
tr
acks
the
pre
dicte
d
cu
rr
e
nt
a
t
t
k+1
.
Howe
ver,
if
t
he
cal
culat
ion
ti
me
is
sig
nifica
nt
co
mp
a
re
d
t
o
the
sam
plin
g
ti
me,
the
re
is
a
delay
be
twee
n
the
measu
red
c
urr
ent
insa
nt
a
nd
the
a
ppli
ed
ne
w
volt
age
ve
ct
or
i
ns
ta
nt.
D
ur
i
ng
the
dela
y
ti
me,
the
pr
e
vious
vo
lt
age
vecto
r
is
sti
ll
ap
plied,
w
hich
makes
t
he
l
oa
d
c
urre
nt
m
ov
e
a
wa
y
from
the
re
fe
rence
a
nd
inc
reas
es
the
current
rip
ple.
A
sim
ple
s
olu
ti
on
to
c
omp
ensate
this
del
ay
is
to
ta
ke
t
he
cal
culat
io
n
ti
me
into
acc
ount
a
nd
app
l
y
t
he
sel
e
ct
ed
vo
lt
age
ve
ct
or
a
fter
the
ne
xt
sam
plin
g
in
sta
nt
[3
,
14
,
16]
.
T
he
l
oa
d
c
urre
nt
rea
ches
t
he
pr
e
dicte
d value
at t
k+2
.
3.2.
3.
Predi
cti
on
of futur
e r
eferences
In
ge
ne
ral,
th
e
f
uture
ref
e
rence
s
are
un
known
a
nd
nee
de
d
t
o
be
est
imat
ed
by
us
in
g
a
seco
nd
-
or
der
extra
po
la
ti
on
[
10]
.
H
ow
e
ve
r,
for
s
uffici
ently
small
sam
pli
ng
ti
me
,
the
f
uture
re
fer
e
nce
value
at
ti
me
t
k+2
is
assume
d
to
b
e
ap
pr
ox
imat
el
y equ
al
to
th
e pr
esent ref
e
re
nce
v
al
ue
at
ti
me
t
k
[
3]
:
**
)
2
)
((
kk
+
ii
.
Ba
se
d
on
the
aforeme
ntio
ne
d
a
nalysis,
the
cost
f
unct
ion i
s
chose
n
as:
**
(
)
(
2
)
(
)
(
2
)
J
k
k
k
k
=+
−
+
−
+
i
i
i
i
(13)
*
)
(
k
i
:
Re
fer
e
nce
sta
tor
c
urre
nt of t
he
m
otor i
n αβ
coor
din
at
e at t
ime i
ns
ta
nce
t
k
.
2)
(
k
+
i
:
Pr
e
dicti
ve
sta
tor
c
urre
nt of t
he
m
otor i
n αβ
coor
din
at
e at t
ime i
ns
ta
nce
t
k+2
.
4.
FPGA
I
MPL
EMENT
ATION
4.1.
Functi
onal int
ernal cir
cuit
(IC)
desig
n.
A
c
ompli
cat
ed
al
gorith
m
ca
n
be
di
vid
e
d
i
nto
man
y
small
er
cal
culat
io
n
ste
ps
a
nd
a
n
i
nt
ern
al
ci
rcu
it
needs
to
be
de
sign
e
d
an
d
pro
gr
a
med
func
ti
on
al
y
to
perf
orm
each
ste
p.
A
n
IC
inclu
des
tw
o
f
unda
mental
blo
c
ks
:
a
fi
nite
sta
te
machin
e
(F
S
M)
an
d
a
processi
ng
un
it
(P
U)
a
s
sho
wn
i
n
Fig
ure
4
(
a
)
.
T
he
F
SM
blo
c
k
con
ta
in
s
al
l
fin
it
e
sta
te
s
of
the
IC
an
d
co
ntr
ol
sign
al
s
to pro
c
essing
unit
.
Me
anly
or
Moor
e
method
a
re
usual
ly
us
e
d
to
desi
gn
the FS
M
with
s
ever
al
t
yp
ic
al
s
ign
al
s:
cl
k
,
r
st
,
init
and
do
ne.
I
n wh
ic
h,
Clk
a
nd
rst
are
ope
r
at
ing
cl
ock
a
nd
reset
sig
nal
f
or
the
ci
rcu
it
,
res
pect
ively.
I
nit
a
nd
done
sig
nal
is
set
to
‘
1’
for
only
one
cl
oc
k
per
i
od
wh
e
n
t
he
IC
s
t
arts
an
d
finis
hes
it
s
operati
on.
T
he
done
sign
al
is
co
nn
ect
ed
to
the
i
ni
t
sign
al
of
th
e
othe
r
ci
rcu
it
s
o
that
seq
uen
ti
al
cal
c
ulati
on
ca
n
be
imple
mente
d.
Since
a
n
inter
nal
ci
rc
uit
only
operates
in
a
fi
xed
small
ti
me
an
d
is
inact
ive
m
os
t
of
th
e
ti
me
,
this
de
sig
n
a
vo
i
ds
the
pr
opagati
on
of
un
e
sp
ect
ed
glit
che
s
an
d
reduces
the
F
P
GA
powe
r
c
on
su
m
ptio
n.
T
he
PU
bl
ock
c
onta
ins
operat
or
s
+,
-
,
x,
/
a
nd
ta
kes
D
ata_in
as
in
pu
t
data
to
cal
culat
e
an
d
t
hen
se
nd
the
re
su
lt
s
t
o
the
D
ata_out
unde
r
the
co
ntr
ol
of
t
he
F
S
M
.
The
s
tr
uctu
re
of
the
PU
is
desig
ne
d
by
t
he
pi
pelined
str
uctu
re
as
s
how
n
i
n
F
igure
4
(
b
)
t
o
s
yn
c
hro
nize
t
he
data
a
nd
sig
na
ls.
c.
Pr
e
dicti
ve
cu
rrent co
ntr
ol im
pl
ementat
ion ba
sed o
n
FP
G
A p
la
tfor
m
.
The
fl
ow
c
har
t
s
how
n
i
n
Fig
ure
5
(
a
)
il
lustrat
es
how
to
im
plement
the
FCS
-
MPC
for
cu
r
r
ent
loop
of
the
I
M
on
a
F
PGA
platf
orm.
To
so
l
ve
the
al
gorithm
desc
ribe
d
in
the
previo
us
sect
io
n,
there
are
nin
e
ste
ps
corres
pondin
g
to
ni
ne
inte
rn
a
l
ci
rcu
it
s.
T
he
name
of
Ste
p1
to
Step
9
are:
ADC_re
ad
,
ab
c_to
_α
β
,
Dq_t
o_αβ
,
Is_
t
o_
fl
ux
,
Pr
e_mo
del
,
J
_c
al
c
ci
rcu
it
,
Fin
d_mindJ
,
αβ_
t
o_dq
a
nd
Flu
x_mo
del,
res
pe
ct
ively
.
Since
Ste
p2
requires
the
da
ta
cal
culat
ed
in
Step
1
w
hile
Step
3
re
qu
ire
s
the
data
from
Step
2
,
the
ope
r
at
ion
of
ci
rc
uits
1,2
,3
is
sequ
e
ntial
.
S
imi
la
rly,
the
ci
rcu
it
s
3,5
,
6,7
a
nd
8,9
mu
st
run
se
qu
e
ntial
ly.
M
ea
nwhile
,
St
ep5
re
qu
i
res
th
e
data
from
bo
t
h
Ste
p3
a
nd
Ste
p4
,
so
that
ci
rc
uit
3
an
d
4
m
us
t
op
e
rate
in
paral
le
l.
Be
cause
ci
rcu
it
4
needs
more
cal
culat
ion
ti
me
tha
n
ci
rcu
it
3,
ci
rc
uit
5
only
ta
ke
the
i
nput
data
wh
e
n
ci
rcu
i
t
4
e
nds
it
s
operati
on,
wh
ic
h
means
the
done
sig
nal
of
ci
rc
uit
4
is
the
init
sig
nal
of
ci
rc
uit
5.
On
the
ot
her
ha
nd,
Step
9
on
l
y
re
qu
ire
s
data
from
Ste
p8
w
hi
ch
mea
ns
ci
rc
uit
8
an
d
9
ope
rate
in
pa
rall
el
with
ci
rc
uit
3,5
,6
,
7.
Finall
y,
t
he
e
xec
ution
ti
me
of
th
e
FC
S
-
M
PC
is
sho
wn
i
n
Fig
ure
5b.
As
ca
n
be
ob
se
r
ved,
se
ver
al
ste
ps
can
be
ca
rr
ie
d
ou
t
in
par
al
le
l
wh
ic
h
reduces
the
c
omp
utati
on
al
ti
me
of
the
im
plemented
al
gori
thm.
T
he
par
al
le
l
computat
io
n
abili
ty
of
the
FPGA,
wh
ic
h
is
not
a
vaila
ble
in
c
onve
ntio
nal
D
S
P/mi
cro
c
ontr
oller,
al
lows
the
M
PC
to
be
a
pp
li
ed
to
ma
ny
ot
her
fiel
ds
without
hav
i
ng to w
orr
y
a
bout the
c
omp
utati
on
al
ti
me.
Evaluation Warning : The document was created with Spire.PDF for Python.
In
t J
P
ow Elec
& Dri S
ys
t
IS
S
N:
20
88
-
8
694
Ha
r
dw
ar
e i
n
t
he
lo
op
c
o
-
si
m
ula
ti
on
of fi
nite set
-
model
pre
dicti
ve co
ntrol
u
sin
g
F
PGA
…
(
Mai
Van C
hung
)
1725
F
S
M
Pr
o
cessi
n
g
Un
i
t
C
ont
r
ol
s
i
g
nal
I
nit
D
o
n
e
r
s
t
c
l
k
I
nit
r
s
t
D
a
t
a
_
in
D
a
ta
_
out
Done
(a)
RE
G
RE
G
RE
G
RE
G
RE
G
RE
G
RE
G
X
X
X
X
RE
G
RE
G
RE
G
RE
G
RE
G
RE
G
MUX
sel
+
-
+
-
A
11
K
11
A
12
Ψ
K
21
A
22
K
22
(
b)
Figure
4
.
F
un
ct
ion
al
,
(a) IC
str
uctu
re, (b)
Pip
el
ined
str
uct
ure
1
.
R
e
a
d
A
D
C
-
M
C
P
3
2
0
8
(
S
P
I
s
t
a
n
d
a
r
d
)
2
.
C
l
a
r
k
t
r
a
n
s
f
o
r
m
a
t
i
o
n
o
f
c
u
r
r
e
n
t
a
b
c
_
to
_
αβ
4
.
C
a
l
c
u
l
a
t
e
f
l
u
x
r
o
t
o
r
i
n
α
β
c
o
o
r
d
i
n
a
t
e
Is
_
t
o
F
lu
x
7
.
F
e
c
t
h
v
o
l
t
a
g
e
v
e
c
t
o
r
t
o
m
i
n
i
m
i
z
e
t
h
e
c
o
s
t
f
u
n
c
t
i
o
n
F
i
n
d
_
m
i
n
J
A
D
C
_
r
e
a
d
R
e
f
e
r
e
n
c
e
c
u
r
r
e
n
t
f
r
o
m
s
p
e
e
d
a
n
d
f
l
u
x
r
e
g
u
l
a
t
o
r
5
.
P
r
e
d
i
c
t
i
v
e
c
u
r
r
e
n
t
m
o
d
e
l
a
t
i
n
s
t
a
n
t
k
+
1
,
k
+
2
P
r
e
_
m
o
d
e
l
G
e
n
e
r
a
t
i
n
g
t
r
i
g
g
e
r
s
i
g
n
a
l
s
6
.
C
o
s
t
f
u
n
c
t
i
o
n
J
_
c
a
lc
C
u
r
r
e
n
t l
oop
F
C
S
-
M
P
C
8
.
T
r
a
n
s
f
o
r
m
i
αβ
i
n
t
o
i
d
q
αβ
_
to
_
dq
9
.
F
l
u
x
r
o
t
o
r
m
o
d
e
l
F
lu
x
_
m
o
d
e
l
3
.
T
r
a
n
s
f
o
r
m
i
*
αβ
i
n
t
o
i
*
d
q
Dq
_
to
_
αβ
(a)
A
D
C
_
r
e
a
d
αβ
_
to
_
dq
dq
_
to
_
αβ
Is
_
t
o
F
lu
x
αβ
_
to
_
dq
P
r
e
_
m
o
d
e
l
J
_
c
a
l
Fi
n
d
_
m
i
n
J
Flu
x
_
m
o
d
e
l
E
xe
c
ut
i
on t
i
m
e
(
s
)
C
i
r
c
ui
t
(
s
t
e
p
)
(b)
Figure
5.
Flo
w
char
t,
(
a
)
F
PGA im
pleme
ntati
on
(
b)
IC’s
exe
cution t
ime
Evaluation Warning : The document was created with Spire.PDF for Python.
IS
S
N
:
2088
-
8
694
In
t J
P
ow
Ele
c
&
Dr
i
S
ys
t
,
V
ol
.
1
1
, N
o.
4
,
D
ecembe
r
2020
:
17
19
–
17
30
1726
4.2.
HIL
-
FP
GA pl
atfo
r
m
To
quic
kl
y
veri
fy
the
FCS
-
MPC
f
or
IM
without
risky
ex
pe
riments,
real
-
ti
me
simulat
io
n
base
d
o
n
HI
L
platfo
rm
a
re
co
nducte
d.
The
vi
rtual
I
M
mo
to
r
wit
h
pa
rameters
pro
vid
ed
i
n
Table
1
and
t
he
3
-
le
vel
CHB
inv
e
rter
a
re
si
mu
lt
ed
by
T
yphoon
HI
L
402
de
vice.
T
he
c
on
t
ro
l
al
go
rith
m
is
impleme
nted
by
a
F
PGA
car
d
name
d
ZYB
O
-
27.
The
blo
c
k
diag
r
am
an
d
the
e
xperime
nt
al
de
vices
of
t
he
real
-
ti
me
si
mu
la
ti
on
s
ys
te
m
a
re
sh
ow
n
in
Fi
g
ure
6
a
nd Fi
g
ure
7,
res
pecti
vely
.
w
*(
k
)
w
(
k
)
+
-
S
p
e
e
d
r
e
g
u
l
a
t
o
r
i
s
q
*
F
l
u
x
r
e
g
u
l
a
t
o
r
i
s
d
*
Ψ
rd
(
k
)*
i
αβ
(
k
)*
F
l
ux m
ode
l
α
β
d
q
Ψ
r
_
α
β
(
k
)
i
αβ
(
k
)
i
αβ
(
k
)
w
(
k
)
θ
s
w
(
k
)
C
o
s
t
f
u
n
c
t
i
o
n
P
r
e
d
i
c
t
i
v
e
c
u
r
r
e
n
t
a
t
i
n
s
t
a
n
t
k
+
2
F
C
S
-
M
P
C
+
Ψ
rd
(
k
)
-
IM
c
b
a
v
dc
v
dc
v
dc
N
3
-
l
e
v
e
l
c
o
n
v
e
r
t
e
r
C
H
B
ZY
BO
-
Z
7
IE
i
a
i
b
i
c
w
V
d
c
V
dc
(
k
)
V
d
c
a
b
c
α
β
Typ
h
oon
H
I
L
402
S
i
g
n
a
l
m
e
a
s
u
r
e
m
e
n
t
Tr
i
g
g
e
r
s
i
g
n
a
l
s
Figure
6. Bl
oc
k diag
ram of
th
e HIL
-
base
d
si
mu
la
ti
on s
ys
te
m
Table
1
. Para
m
et
ers
of t
he
I
M
Sy
m
b
o
l
Qu
an
tity
Valu
e
P
r
ate
Rate p
o
wer
2
.2k
W
T
r
ate
Rate to
rqu
e
7
.3Nm
I
r
ate
Rate p
h
ase current
4
.7A
V
r
ate
Rate p
h
ase v
o
ltag
e
400V
f
r
ate
Rate freq
u
en
cy
5
0
Hz
N
r
ate
Rate sp
eed
2
8
8
0
r
p
m
R
s
Stato
r
resistan
ce
1
.99
Ω
R
r
Ro
to
r
resistan
ce
1
.84
Ω
L
m
Mutu
al
in
d
u
ctan
ce
0
.37
m
H
Figure
7. Ex
pe
riment
de
vices
5.
RESU
LT
S
A
ND D
I
SCUS
S
ION
In
this
sect
io
n,
va
rio
us
real
-
ti
me
simulat
io
ns
a
re
ca
rr
ie
d
out
to
ve
rif
y
the
de
velo
pe
d
FC
S
-
M
PC
app
li
ed
t
o
I
M
con
t
ro
l.
Fi
rst,
the
tran
sie
nt
re
sp
onse
of
t
he
con
t
ro
l
s
ys
te
m
with
the
loa
d
disturba
nce
is
te
ste
d.
At
ti
me
instan
ce
t
=
{0.0
5s
,
0.5s,
0.7
5s
}
,
th
e
load
to
rque
is
su
dde
nly
c
ha
ng
e
d
with
co
rr
es
pondin
g
va
lues
T
L
=
{0
,
0
.
5
T
rate
,
T
rate
}
w
hile
the
ref
e
re
nc
e
sp
ee
d
is
fix
ed
at
ω
ref
=
300(
−
1
).
As
can
be
ob
se
r
ved
i
n
Fig
ure
8,
the
e
le
ct
ro
ma
gn
et
ic
tor
que
gen
e
rat
ed
by
the
I
M
quic
kly
trac
ks
t
he
loa
d
t
orq
ue
to
mainta
i
n
the
r
oto
r
Evaluation Warning : The document was created with Spire.PDF for Python.
In
t J
P
ow Elec
& Dri S
ys
t
IS
S
N:
20
88
-
8
694
Ha
r
dw
ar
e i
n
t
he
lo
op
c
o
-
si
m
ula
ti
on
of fi
nite set
-
model
pre
dicti
ve co
ntrol
u
sin
g
F
PGA
…
(
Mai
Van C
hung
)
1727
sp
ee
d
at
it
s
de
sired
val
ue.
Fr
om
ti
me
in
sta
nce
t
=
1.5s
,
s
peed
re
ve
ral
of
the
I
M
w
it
h
co
ns
ta
nt
l
oad
is
inv
est
igate
d.
T
he
re
fere
nce
s
peed
is
cha
nge
d
at
ti
m
e
insta
nce
t
=
1.5s
a
nd
t
=
4.3s
with
c
orres
pondin
g
va
lues
N
ref
=
-
N
rate
and
N
ref
=
0
.
I
n
this
case,
the
r
otor
sp
ee
d
al
so
qu
i
ckly
trac
kes
it
s
ref
e
ren
ce
with
ne
gligible
tr
ackin
g
error,
i.e.
,
ab
out
0.2
rad/s
.
I
n
the
sec
ond
sc
enar
i
o,
t
he
a
bili
ty
to
gen
e
rate
el
ect
ro
ma
gn
et
i
c
tor
que
at
sta
nd
sti
ll
of
t
he
I
M
c
ontr
ol
sy
ste
m
i
s
cl
arified.
W
it
h
employe
d
FCS
-
M
PC
f
or
the
cu
rr
e
nt
loop,
t
he
ge
ne
rated
el
ect
ro
ma
gn
et
i
c
tor
que
quic
kl
y
trac
ks
the
load
to
rque
at
zero
s
peed
in
j
us
t
3ms
as
s
hown
in
Fi
g
ure
9.
This
qu
ic
k
respo
nse
is
suffici
ent
for
mo
st
pra
ct
ic
al
app
li
cat
ion
s
an
d
c
ompara
ble
to
the
well
-
known
direct
tor
qu
e
contr
ol
(D
TC
).
S
p
e
e
d
m
o
t
o
r
T
o
r
q
u
e
m
o
t
o
r
(a)
T
w
S
p
e
e
d
m
o
t
o
r
T
o
r
q
u
e
m
o
t
o
r
(
b)
Figure
8.
E
le
ct
romag
netic
to
r
qu
e
,
(
a)
Time
-
varyin
g
s
pee
d moto
r
a
nd loa
d t
orq
ue,
(
b) Z
oom
-
in
res
ult
Evaluation Warning : The document was created with Spire.PDF for Python.
IS
S
N
:
2088
-
8
694
In
t J
P
ow
Ele
c
&
Dr
i
S
ys
t
,
V
ol
.
1
1
, N
o.
4
,
D
ecembe
r
2020
:
17
19
–
17
30
1728
S
p
e
e
d
m
o
t
o
r
T
o
r
q
u
e
l
o
a
d
(a)
(
b)
Figure
9.
Ge
ne
rated t
orqu
e
,
(
a
)
L
oad to
rque a
nd loa
d
c
urren
t
at sta
nd
sti
ll
,
(
b) Tor
qu
e
step
respo
ns
e
6.
CONCL
US
I
O
N
In
t
his
pap
e
r,
a
n
a
dv
a
nce
d
s
ol
ution
to
quic
kl
y
dev
el
op
the
FCS
-
M
PC
for
IM
fe
d
by
m
ulti
le
vel
CHB
inv
e
rter
is
dis
cusse
d.
First,
t
he
beh
a
vi
or
of
the
sta
to
r
c
ur
ren
t
i
n
fu
t
ur
e
is
pr
e
dicte
d
w
it
h
each
of
po
ssible
vo
lt
age
vect
or
base
d
on
the
e
xp
li
ci
t
discrete
-
ti
me
m
at
hem
a
ti
cal
mo
del
of
the
I
M
.
T
he
opti
mal
vo
lt
age
ve
ct
or
wh
ic
h
fu
l
fill
the
obje
ct
ive
of
a
prede
fine
d
c
os
t
f
unct
io
n
is
sel
ect
ed
an
d
a
pp
li
ed
to
the
CHB
in
ver
te
r.
S
econd,
the
FP
G
A
with
par
al
le
l
cal
c
ulati
on
abili
ty
is
us
e
d
to
mi
ni
mize
the
c
ompu
ta
ti
onal
ti
m
e
of
the
co
mpl
ic
at
ed
con
t
ro
l
al
gorithm
li
ke
M
PC
.
Finall
y,
to
el
imi
nate
the
bur
den
of
ti
me
on
ha
rdwa
re
desi
gn
as
well
as
t
o
a
void
po
te
ntial
risks
with
high
pow
er
e
xp
e
rime
ntal
sy
ste
ms
,
real
-
ti
me
simulat
i
on
s
wit
h
hi
gh
-
fideli
ty
mat
he
mati
cal
models
im
plemented
by
H
I
L
platf
orm
are
co
nducted
to
ver
i
fy
the
c
ontr
ol
al
gorith
m.
Ach
ie
ved
r
eal
-
ti
me
simulat
ion
s
show
t
hat
the
de
velo
ped
FCS
-
M
PC
giv
es
good
perf
or
m
anc
e
no
t
on
l
y
in
s
te
ady
-
sta
te
bu
t
al
so
i
n
transient
-
sta
te
.
Evaluation Warning : The document was created with Spire.PDF for Python.