Internati
o
nal
Journal of P
o
wer Elect
roni
cs an
d
Drive
S
y
ste
m
(I
JPE
D
S)
Vol
.
5
,
No
. 2, Oct
o
ber
2
0
1
4
,
pp
. 26
8~
28
2
I
S
SN
: 208
8-8
6
9
4
2
68
Jo
urn
a
l
h
o
me
pa
ge
: h
ttp
://iaesjo
u
r
na
l.com/
o
n
lin
e/ind
e
x.ph
p
/
IJPEDS
Fuzzy-PI Torque and Flux Cont
rollers for DTC with Multilevel
Inverter of Induction Machines
N. M.
N
o
rdin
1
,
N
.
R. N.
Idri
s
2
,
N
.
A. Az
li
3
, M.
Z
.
Pute
h
4
,
T. Su
tikn
o
5
1,2,3
F
acult
y of El
ectr
i
ca
l Engine
er
ing,
Univ
ersiti Teknologi Malay
s
ia, Johor
Bahru
,
Mala
y
s
ia
4
M
I
M
O
S
Berhad, T
echno
log
y
P
a
rk M
a
l
a
y
s
ia
, K
u
ala
Lum
pur, M
a
la
ys
i
a
5
Department of Electrical
Eng
i
n
eering
,
Un
iv
ersitas Ahmad Dahlan, Yog
y
ak
arta, I
ndonesia
Article Info
A
B
STRAC
T
Article histo
r
y:
Received Aug 19, 2014
Rev
i
sed
Sep
20
, 20
14
Accepte
d Oct 1, 2014
In this paper the performance
of flux
and torque controller for a Direct
Torque Contro
l
of Cascad
ed H-bridge
Multilev
e
l Inverter (DTC-
C
MLI) fed
induction mach
ines are
investigated
. A Fuzzy
-
PI with fix
e
d switching
frequency
is pr
oposed for both
torque
a
nd flu
x
controller
to
enhance th
e
DTC-CMLI performance. Th
e operational
conce
p
ts of the Fuzz
y-
PI with the
fixed
switch
i
ng frequency
contr
o
ller
of
a DTC-
M
L
I s
y
s
t
em
foll
owed b
y
the
s
i
m
u
lation res
u
l
t
s
and an
al
ys
is
are pr
es
ent
e
d.
The p
e
rform
a
n
ce of
the
proposed sy
stem is verified via
MA
TLAB/Simu
link©. Th
e proposed sy
stem
significantly
im
proves the DTC drive
in ter
m
s of dy
namic performance,
smaller torqu
e
and flux r
i
pple, and lo
wer
total h
a
rmonic distortio
n
(THD).
Keyword:
CMLI
Di
rect
t
o
rq
ue
c
ont
rol
Fuzzy logic c
o
ntrol
I
ndu
ctio
n m
a
c
h
in
es
Mu
ltilev
e
l in
v
e
rter
Copyright ©
201
4 Institut
e
o
f
Ad
vanced
Engin
eer
ing and S
c
i
e
nce.
All rights re
se
rve
d
.
Co
rresp
ond
i
ng
Autho
r
:
Nik R
u
m
z
i Nik
Id
ris,
Depa
rtem
ent of Electrical
P
o
wer
En
gi
nee
r
i
n
g,
Facu
lty of Electrical Eng
i
n
eerin
g
,
Un
iv
ersiti Tekn
o
l
o
g
i
Mal
a
ysia,
8
131
0 Sk
ud
ai,
Jo
hor
, Malaysia.
Em
a
il: nikrum
zi@ieee.org / nikrum
zi@fke.ut
m
.
m
y
1.
INTRODUCTION
The s
upe
ri
o
r
p
e
rf
orm
a
nce o
f
DTC
i
n
dy
na
m
i
c respo
n
se a
nd
si
m
p
l
e
cont
rol
c
o
n
f
i
g
urat
i
on
w
h
i
c
h i
s
ori
g
i
n
al
l
y
i
n
t
r
o
duce
d
i
n
[
1
]
,
h
a
s m
a
de i
t
one of t
h
e m
o
st
pop
ul
ar
resea
r
c
h
t
o
pi
cs i
n
el
e
c
t
r
i
cal
dri
v
e
sy
st
em
s.
Si
nce t
h
e a
p
pl
i
cat
i
on
of
hi
g
h
-
p
o
w
e
r
m
e
di
u
m
vol
t
a
ge i
n
AC
dri
v
es
has s
h
o
w
n
ra
pi
d
de
vel
opm
ent
,
t
h
e
us
e
o
f
m
u
l
tilev
e
l in
v
e
rters in
DTC
sch
e
m
e
h
a
s b
eco
m
e
an
im
p
o
r
tan
t
stru
ctu
r
e fo
r
further d
e
v
e
lop
m
e
n
t and
im
pro
v
em
ent
.
Vari
ous
t
ech
ni
cal
pape
rs
ha
v
e
sh
ow
n
bet
t
e
r pe
rf
o
r
m
a
nce of
D
T
C
sc
he
m
e
usi
n
g m
u
l
t
i
l
e
vel
inve
rters [2
-
3
1
]
.
By e
m
p
l
o
y
in
g
th
e
m
u
ltilev
e
l
in
v
e
rter, th
e ch
o
i
ces
o
f
vo
ltag
e
v
ectors th
at
can
b
e
u
s
ed
to
con
t
ro
l th
e
to
rq
u
e
and
flux
are in
creased
.
Diff
eren
t app
r
o
a
ch
es
h
a
v
e
b
e
en
p
r
op
o
s
ed
for
DTC sche
m
e
u
s
ing
m
u
ltilev
e
l
inve
rter; hysteresis-base
d
controlle
r and non-hysteresis-based controlle
r such as s
p
ac
e vector m
o
dulation
(SVM
)[
8,
13
, 16
, 1
8
, 2
2
, 2
7
-
2
9
, 3
1
]
,
p
r
edi
c
t
i
v
e cont
r
o
l
st
rat
e
gy
[1
0, 1
2
, 3
0
]
and f
u
zzy
l
ogi
c cont
rol
l
e
r
(
F
LC)[7
,
9
,
11
, 22
].
The i
m
pl
em
ent
a
t
i
on o
f
t
h
e
hy
st
eresi
s
-
b
ase
d
c
ont
rol
st
ra
t
e
gy
has
l
ead
t
o
a
hi
g
h
t
o
rq
ue
ri
p
p
l
e
esp
ecially in
d
i
screte im
p
l
e
m
en
tatio
n
ev
en
with
sm
a
ll
h
y
steresis
b
a
nd
. Th
is is
du
e to
th
e
d
e
lay
in
the
sam
p
l
i
ng t
i
m
e
. On t
o
p
o
f
t
h
at
, t
h
e va
ri
abl
e
swi
t
c
hi
n
g
fr
eque
ncy
of t
h
e swi
t
c
hi
n
g
de
vi
ces w
h
i
c
h l
e
ads t
o
un
p
r
edi
c
t
a
bl
e
harm
oni
cs c
u
r
r
e
nt
i
s
al
so
p
r
od
uced
by
im
plementing the
hysteresis-based
cont
rol st
rategy.
As a result, s
o
me researche
r
s
have
chose
n
t
o
use
non-hysteresis-base
d control strategies
to overc
o
m
e
these
dra
w
ba
c
k
s. Significant
im
provem
ent
s
i
n
t
e
rm
s of
fl
ux
an
d t
o
r
q
ue
ri
p
p
l
e
an
d
swi
t
chi
n
g
f
r
e
que
n
c
y
are
accom
p
lished
by using these
control strate
gies; howe
ve
r the use of com
p
lex
m
a
the
m
atical equations and
Evaluation Warning : The document was created with Spire.PDF for Python.
I
J
PED
S
I
S
SN
:
208
8-8
6
9
4
Title o
f
ma
nu
scrip
t
is sh
o
r
t
an
d clea
r, imp
lies resea
r
ch resu
lts (First Au
tho
r
)
26
9
alg
o
rith
m
s
h
a
s led
to
th
e co
m
p
u
t
ation
a
l bu
rden
and
co
m
p
lex
ity o
f
th
e
DTC-MLI sch
e
m
e
esp
ecially when
th
e
lev
e
l of
v
o
ltage is in
creased.
In
[6
] th
e m
u
ltilev
e
l in
v
e
rter in
th
e DTC
sch
e
m
e
e
m
p
l
o
y
s a
m
u
ltile
v
e
l h
y
steresis co
n
t
ro
ller.
Al
t
h
o
u
gh t
h
e
resul
t
s
ha
ve s
h
o
w
n som
e
signi
fi
cant
im
pr
ovem
e
nt
s, as m
e
nt
i
oned ea
r
l
i
e
r, by
usi
n
g
t
h
e
hy
st
eresi
s
-
b
ase
d
co
nt
r
o
l
l
e
r t
h
e swi
t
c
hi
n
g
fr
eque
ncy
o
f
t
h
e po
wer
devi
c
e
s vari
es w
h
i
l
e
t
h
e t
o
rq
ue a
nd
fl
u
x
ripp
le can
still
b
e
co
n
s
i
d
ered as h
i
g
h
.
A fuzzy-PI b
a
sed
co
n
t
ro
ller fo
r
DTC was in
itiall
y in
tro
d
u
c
ed
in
[32
]
.
The c
ontroller
has been
use
d
to re
place the
hysteresis co
nt
roller
while maintaining th
e
use of a look-up table.
Ho
we
ver
t
h
e
p
r
o
p
o
sed
co
nt
r
o
l
l
e
r has
bee
n
a
ppl
i
e
d
t
o
a
3-
p
h
ase c
o
nve
nt
i
o
nal
i
n
vert
e
r
.
In
th
is
work
a
fu
zzy-PI
with
th
e fix
e
d
switch
i
ng
fr
equ
e
n
c
y co
n
t
ro
ller is utilized
in
th
e l
o
ok-up
tab
l
e
base
d DTC
d
r
i
v
e. The
pr
o
p
o
se
d co
nt
rol
l
e
r co
nsi
s
t
s
of
a fuzzy
l
ogi
c
cont
rol
l
e
r a
n
d
a t
r
i
a
ngul
a
r
carri
er
wave
f
o
rm
. The
fi
xe
d swi
t
c
hi
n
g
f
r
eq
ue
ncy
i
s
obt
ai
ne
d
by
co
m
p
ari
ng t
h
e
fu
zzy
l
ogi
c o
u
t
p
ut
wi
t
h
t
h
e t
r
i
a
ng
ul
a
r
wav
e
fo
rm
s. A 5
-
lev
e
l cascad
e
d
H-bridg
e
m
u
l
tilev
e
l in
v
e
rter (CMLI)
is e
m
p
l
o
y
ed
in
th
is sch
e
m
e
. The
m
u
l
tip
le iso
l
ated
inp
u
t
DC
sou
r
ces
o
f
th
e CMLI
a
r
e
part
i
c
ul
a
r
l
y
sui
t
a
bl
e f
o
r e
l
ect
ri
c vehi
cl
e
(EV
)
appl
i
cat
i
o
ns si
nce t
h
e
p
o
we
r
so
urce
f
o
r a
n
EV ca
n
be
ob
t
a
i
n
ed f
r
o
m
t
h
e bat
t
e
ry
m
odul
es.
In t
h
e
pr
op
ose
d
strategy, the
fuzzy-PI
with the fi
xe
d switc
hing fre
que
nc
y controller
w
ill replace the
m
u
lt
ilevel hysteresis
cont
rol
l
e
r f
o
r t
o
r
q
ue an
d fl
u
x
cont
rol
.
B
a
se
d on t
h
e
pr
o
pose
d
co
nt
r
o
l
l
e
r o
u
t
put
t
oget
h
er
w
i
t
h
t
h
e fl
ux
po
s
i
t
i
on,
an a
p
p
r
o
p
r
i
a
t
e
vol
t
a
ge
vect
or
can
be sel
ect
ed
fr
om
t
h
e l
o
o
k
-
u
p
t
a
bl
e.
Fi
g
u
r
e
1 s
h
ows
a
pr
op
ose
d
sy
st
em
bl
o
c
k
di
ag
ram
.
In
th
is
p
a
p
e
r,
th
e o
p
e
ration
a
l co
n
c
ep
ts of fu
zzy
-P
I with t
h
e fixe
d s
w
itchin
g
fre
q
u
e
n
c
y
contr
o
ller
fol
l
o
we
d by
t
h
e sim
u
l
a
t
i
on re
sul
t
s
an
d anal
y
s
i
s
on t
h
e pe
rf
orm
a
nce of t
h
e
pr
op
ose
d
sy
st
em
are prese
n
t
e
d. T
h
e
resul
t
s
ha
ve s
h
o
w
n t
h
at
bet
t
er dy
nam
i
c perf
orm
a
nce, s
m
al
l
e
r t
o
rq
ue
an
d fl
ux
ri
p
p
l
e
s, c
o
nst
a
nt
de
vi
ce
switch
i
ng
frequ
en
cy and
lower THD in
t
h
e
pha
se c
u
rrent a
r
e ac
hieve
d
.
Fi
gu
re
1.
Pr
o
p
o
se
d sy
st
em
bl
ock
di
a
g
ram
2.
BASI
C P
R
I
N
CIPLE O
F
DI
RECT T
O
R
Q
UE C
O
NT
RO
L (DT
C
)
In
Fi
g
u
re
2
t
h
e
bl
oc
k
di
ag
ram
o
f
D
T
C
basi
c
cont
rol
w
h
i
c
h
ori
g
i
n
al
l
y
i
n
t
r
o
duce
by
[
1
]
i
s
s
h
o
w
n. B
y
havi
ng
an
i
n
st
a
n
t
a
ne
ou
s
val
u
e
o
f
t
o
rq
ue a
n
d
fl
u
x
whi
c
h are
cal
cul
a
t
e
d f
r
o
m
t
h
e m
easured t
e
rm
i
n
al
vari
abl
e
s
of
i
n
duct
i
o
n m
achi
n
e
(IM
),
b
o
t
h
t
o
r
q
ue a
n
d
fl
u
x
er
r
o
rs
can
be
det
e
rm
i
n
ed.
B
a
sed
o
n
t
h
e
e
r
r
o
rs
, a
n
opt
i
m
um
require
d
volta
ge vectors a
r
e se
lected from
th
e lo
ok-up
tab
l
es to
d
r
i
v
e th
e IM.
The
fl
u
x
est
i
m
at
i
on i
s
base
d
on
st
at
or
v
o
l
t
a
ge
vect
o
r
e
quat
i
on i
n
st
at
i
o
nar
y
refe
rence
f
r
a
m
e whi
c
h
can
b
e
written
as
dt
d
i
R
v
s
s
s
s
(1
)
Ov
er t
h
e sm
all
p
e
ri
o
d
of tim
e,
it is assu
m
e
d
th
at
th
e vo
ltag
e
d
r
op
acro
s
s
th
e
stator resistance
can be
negl
ect
e
d
. T
h
e
r
ef
ore
t
h
e e
q
ua
t
i
on ca
n
be
rew
r
i
t
e
as
Evaluation Warning : The document was created with Spire.PDF for Python.
I
S
SN
:
2
088
-86
94
I
J
PED
S
Vo
l. 5
,
No
. 2
,
O
c
tob
e
r
201
4
:
2
68–
2
82
27
0
t
v
s
s
(2
)
It’s clearly s
h
ows
that the
c
h
ange
i
n
st
at
o
r
fl
ux
l
i
nka
ge
vect
o
r
,
∆
is di
rectly affected by the
sel
ect
i
on o
f
st
a
t
or
vol
t
a
ge
vec
t
or.
He
nce t
h
e
st
at
or fl
ux l
o
cu
s can
be co
nt
r
o
l
l
e
d by
sel
ect
i
n
g a s
u
i
t
a
bl
e v
o
l
t
age
vectors.
As
fo
r t
h
e t
o
r
que
est
i
m
at
i
on, i
t
base
d
o
n
t
h
e st
at
o
r
-
r
ot
or
fl
u
x
a
n
gl
e m
ovem
e
nt
. T
h
e
rel
a
t
i
ons
hi
p
b
e
tween
stator-ro
t
o
r
flux
ang
l
e and
electro
mag
n
e
tic t
o
r
q
u
e
in
statio
n
a
ry
referen
ce fram
e
can
b
e
written
as
sr
r
s
s
s
e
P
i
P
T
sin
2
3
2
3
(3
)
Whe
r
e |
ψ
s
| and |
ψ
r
|
are t
h
e
m
a
gni
t
u
des
of
st
at
or fl
ux a
n
d r
o
t
o
r fl
u
x
l
i
n
ka
ges res
p
ect
i
v
el
y
and
θ
sr
is
th
e stato
r
-ro
to
r flu
x
ang
l
e.
Wh
en
a
v
o
ltag
e
v
ector is a
p
p
l
i
e
d
,
th
e stator flu
x
link
a
g
e
will
m
o
v
e
faster than
ro
t
o
r flux
lin
kag
e
, wh
ere th
e ro
tor flux
m
o
t
i
o
n
is lag
b
e
h
i
n
d
th
e stator flu
x
ro
tation
.
This is d
u
e
to
th
e to
the
rot
o
r a
n
d stator leaka
g
e i
n
ductances.
Th
e
r
e
f
ore the stator-rot
or
flux a
n
gle,
θ
sr
(he
n
ce t
o
rque) is
affec
t
ed by
the selection of appropriate voltage vect
or.
The abovem
entioned
pri
n
cipl
e of both
stator flux and stat
or
-r
ot
o
r
an
gl
e m
ovem
e
nt are affected
by
t
h
e va
ri
at
i
o
n
o
f
st
at
or
v
o
l
t
a
ge
i
s
use
d
i
n
D
T
C
schem
e
t
o
ach
i
e
ve a
desi
re
d
f
l
ux t
r
aject
ory
a
n
d
t
o
rq
ue
res
p
ons
e.
Fi
gu
re 2.
C
o
nv
ent
i
onal
DTC
bl
oc
k di
ag
ram
3.
5
LEVEL CASCDED H-B
R
IDGE
MULTILEVEL
INVERTER
5
-
lev
e
l CMLI
co
nsist o
f
2 H-bridg
e
i
n
ve
rt
er (cel
l
)
c
o
nnec
t
ed i
n
casca
de
d f
o
rm
wi
t
h
separat
e
d
D
C
source as
shown i
n
Fi
gure
3.
As
for three-pha
se
IM, each
pha
se is fed by
one
5-le
vel CMLI. T
h
e
co
nf
igu
r
ation of
thr
ee-ph
ase
IM f
e
d b
y
5-
level CMLI
is sh
ow
n in
Figu
r
e
4.
Th
e
nu
m
b
er of vo
ltag
e
lev
e
l,
L
f
o
r
C
M
LI ca
n
be
obt
ai
ne
d
b
y
(4
)
Whe
r
e
n
is a
num
b
er of cell
per
phase. Eac
h
cell is
conn
ected
in
series.
Th
erefo
r
e t
h
e t
o
tal ou
tpu
t
voltage
of eac
h phase ca
n
be
determined
by
∑
(5
)
∑
(6
)
∑
(7
)
Evaluation Warning : The document was created with Spire.PDF for Python.
I
J
PED
S
I
S
SN
:
208
8-8
6
9
4
Title o
f
ma
nu
scrip
t
is sh
o
r
t
an
d clea
r, imp
lies resea
r
ch resu
lts (First Au
tho
r
)
27
1
V
aN
,
V
bN
and
V
cN
is
th
e p
h
a
se ou
tpu
t
vo
ltag
e
with
resp
ect
t
o
th
e
n
e
u
t
ral,
N
.
By consi
d
eri
n
g each cell
pr
o
duce
{-
V
DC
, 0 ,
V
DC
}
,
base
d
on
(
5
)
,
(6
) a
n
d
(7
),
a
5-l
e
vel
o
u
t
p
ut
v
o
l
t
a
ge
f
o
r eac
h
p
h
ase
i
s
,
,
,
,
(8
)
The output
vol
t
ages,
V
S
gene
r
a
t
e
d by
t
h
e i
n
v
e
rt
er ca
n
be e
x
press
e
d
i
n
s
p
ac
e p
h
as
or
f
o
rm
as gi
ven
i
n
(9
).
V
S
(t)
= 2/3
(
V
aN
(t
)
+
aV
bN
(t
)
+
a
2
V
cN
(t
)
)
(9
)
Whe
r
e
a
=
e
j2
π
/3
and
a
2
=
e
j4
π
/3
.
I
n
d-
q f
o
rm
,
t
h
e out
put
v
o
l
t
a
ges
ca
n be defi
ned
as
(
1
0)
√
(1
1)
Based
on
5-level
m
u
lti
lev
e
l i
n
v
e
rter,
5
3
= 1
25 c
o
m
b
i
n
at
i
ons o
f
p
h
ase
vo
l
t
a
ge wi
t
h
3L(
L
– 1)
+ 1
=
61
v
o
l
t
a
ge
vec
t
ors ca
n b
e
ge
nerat
e
d. T
h
i
s
c
a
n gi
ve m
o
re d
e
grees
o
f
f
r
eed
om
i
n
cho
o
si
n
g
v
o
l
t
a
ge
vect
ors
f
o
r
co
n
t
ro
l
p
u
rpo
s
es co
m
p
ared
t
o
th
e co
nv
en
tio
n
a
l
3
-
ph
ase in
v
e
rter.
Th
e
h
i
g
h
e
r th
e lev
e
l o
f
m
u
ltilev
e
l i
n
v
e
rter,
th
e
m
o
re th
e vo
ltag
e
v
ect
o
r
s
g
e
n
e
rated. Figu
re 5
illu
strates th
e v
o
ltag
e
vecto
r
s g
e
n
e
rated
b
y
5
-
lev
e
l CMLI o
n
a d-
q p
l
an
e.
Fi
gu
re 3.
5
-
l
e
v
e
l
cascade
d
H-
bri
dge
m
u
l
t
i
l
e
vel
i
nve
rt
er (C
M
L
I)
Fi
gu
re 4.
Th
re
e-p
h
ase IM
fed
by
5-l
e
vel
CMLI
Evaluation Warning : The document was created with Spire.PDF for Python.
I
S
SN
:
2
088
-86
94
I
J
PED
S
Vo
l. 5
,
No
. 2
,
O
c
tob
e
r
201
4
:
2
68–
2
82
27
2
Fi
gu
re 5.
V
o
l
t
a
ge vect
o
r
gene
rat
e
d by
5-l
e
v
e
l
CMLI
Figu
re
6.
Ge
ne
ral str
u
cture
o
f
the F
u
zzy
-P
I c
ont
roller
4.
THE PROPOSED
CONTROLLER
The p
r
op
ose
d
cont
rol
l
e
r i
s
e
m
pl
oy
ed as an al
t
e
rn
ativ
e to th
e h
y
steresis-b
ased
co
n
t
ro
ller with
th
e
bene
fi
t
of
ope
r
a
t
i
ng at
a fi
xe
d
swi
t
c
hi
n
g
f
r
eq
uency
wi
t
h
l
o
w t
o
r
q
ue an
d f
l
ux ri
ppl
e.
Pri
o
r k
n
o
w
l
e
d
g
e
of
rot
o
r
param
e
t
e
rs i
s
not
re
qui
re
d i
n
desi
gni
ng
a f
u
zzy
-P
I
base
d
cont
rol
l
e
r.
H
o
weve
r,
t
h
e
fuz
z
y
-
PI
pe
rf
orm
a
nce i
s
hi
g
h
l
y
affect
e
d
by
t
h
e
n
o
r
m
a
lizat
i
on
gai
n
sel
ect
i
on
of
a t
y
pi
cal
fuzzy
l
ogi
c
cont
rol
l
e
r.
Th
e fu
zzy-PI co
n
t
ro
ller stru
ct
u
r
e is
b
a
sed
on th
e trad
ition
a
l
PI co
n
t
ro
ller
[3
3
]
. Th
e
g
e
n
e
ral stru
ct
u
r
e
o
f
th
e Fu
zzy-PI co
n
t
ro
ller is as sh
own
in
Fig
u
re 6
.
It h
a
s two
in
pu
ts; syste
m
error,
E
rr
and t
h
e cha
nge
of
sy
stem
erro
r,
dE
rr
, whe
r
e
E
rr
=
Y
ref
– Y
est
and
dE
rr
=
E
rr
(t)
–
E
rr
(t-
Δ
t)
.
Befo
re th
ese inp
u
t
v
a
lu
es are
fed in
to
t
h
e f
u
zzy
co
nt
r
o
l
l
e
r,
bot
h
val
u
es nee
d
t
o
be
n
o
rm
al
i
zed by
u
s
i
ng t
h
e
inpu
t n
o
r
ma
liza
tion
g
a
i
n
(G
0
a
nd
G
1
). A
s
fo
r the
f
u
zzy
c
ont
roller
o
u
tp
u
t
,
dO
u
t
, th
e
valu
e n
e
ed
t
o
be d
e
n
o
rm
alize
d
u
s
ing
th
e
o
u
t
put
den
o
r
m
al
i
z
at
i
o
n
gai
n
, G
2
before
it can
be
use
d
as a c
o
m
p
arator input.
4.
1
F
u
zzy
-
P
I
t
o
rq
u
e
co
n
t
r
o
lle
r
The p
r
o
p
o
sed
fuzzy
-
P
I to
rq
u
e
contr
o
ller (F
TC) con
s
ists o
f
6
triangu
lar wav
e
form g
e
n
e
rat
o
rs, 6
com
p
arat
ors a
nd a f
u
zzy
-P
I cont
rol
l
e
r. T
h
e
si
x t
r
i
a
ng
ul
ar wave
f
o
rm
generat
o
r
s
ge
nerat
e
3 pai
r
s o
f
t
r
i
a
ng
ul
a
r
wav
e
fo
rm
s with
the sam
e
m
a
g
n
itud
e
bu
t
with
d
i
fferen
t DC
o
f
fset.
Each p
a
ir (C
Upper
and
C
Lower
)
is 18
0°
o
u
t
of
p
h
a
se. In
p
r
in
cip
l
e, th
e pro
posed
con
t
ro
ller
will p
r
o
d
u
ce the sa
m
e
o
u
t
pu
t as an
8-lev
e
l
hysteresis co
n
t
ro
ller in
[6]
,
whi
c
h ca
n
be ei
t
h
er
o
n
e
o
f
t
h
e
fol
l
o
wi
n
g
t
o
r
que e
r
r
o
r st
at
us;
3,
2,
1,
+
0
.
5
,
-0
.5
, -
1
,
-2
, -
3
. T
h
e
num
ber o
f
l
e
vel
s
, h
o
w
eve
r
, m
u
st
be real
i
s
t
i
c
eno
u
g
h
f
o
r
im
pl
em
ent
a
t
i
o
n p
u
r
p
oses.
I
n
ot
he
r w
o
r
d
s, t
h
e hi
g
h
er t
h
e n
u
m
ber
o
f
lev
e
ls, t
h
e
faster in term
s
o
f
pro
c
essor
requ
irem
en
t is n
e
ed
ed fo
r i
m
p
l
e
m
en
tatio
n
.
By co
m
p
aring
the
t
r
i
a
ng
ul
ar
wa
v
e
fo
rm
s wi
t
h
t
h
e f
u
zzy
-P
I c
ont
rol
l
e
r
o
u
t
p
ut
, a
fi
xe
d s
w
i
t
c
hi
n
g
fre
que
ncy
ca
n
be ac
hi
eve
d
.
Based
on
th
e
g
e
n
e
ral stru
cture of fu
zzy-PI
co
n
t
ro
ller, th
e
FTC h
a
s t
w
o
i
n
pu
ts; to
rqu
e
erro
r,
T
err
and
the change of torque error,
dT
err
, where
T
err
= T
ref
– T
est
and
dT
err
= T
err
(t
)
–
T
err
(t-
Δ
t)
and one out
put
,
dT
O
.
Fi
gu
re
7 s
h
ows
t
h
e
bl
oc
k
di
ag
ram
of t
h
e
FT
C
.
4.
2
F
u
zzy
-
P
I
f
l
u
x
co
nt
ro
lle
r
The p
r
o
p
o
se
d
fuzzy
-P
I flu
x
contr
o
ller (F
FC) co
nsis
t
s
of
4 t
r
i
a
ng
ul
a
r
wa
vef
o
rm
generat
o
rs
, 4
com
p
arators a
nd a fuzzy-PI cont
roller. T
h
e f
o
u
r
t
r
i
a
n
gul
a
r
wave
f
o
rm
generat
o
r
s
gene
rat
e
2 pai
r
s of t
r
i
a
ng
ul
ar
wave
form
s with the sam
e
ma
gnitude but wi
th diffe
re
nt
DC offset and si
milar to the F
T
C, each pair
(C
Upper
and C
Lower
)
is also
180
° o
u
t
o
f
ph
ase. In
prin
cip
l
e,
th
e pr
op
o
s
ed
con
t
ro
ller will produ
ce th
e sam
e
o
u
t
p
u
t
as
a
5
-
lev
e
l h
y
steresis con
t
ro
ller i
n
[6
], wh
ich can
b
e
eith
er
on
e
o
f
t
h
e
f
o
l
l
o
wi
ng
t
o
rq
ue e
r
r
o
r
st
at
us;
2,
1,
0
,
-1
, -
2
.
Agai
n, t
h
e
num
b
er of le
vels
chosen m
u
st
be p
r
act
i
cal
fo
r i
m
pl
em
ent
a
t
i
on p
u
r
p
oses.
Based on
the
g
e
n
e
ral stru
cture of
fu
zzy
-PI
co
n
t
ro
ller, th
e
FFC h
a
s two
i
n
pu
ts; fl
u
x
erro
r,
F
lx
err
and
the cha
n
ge of
flux error,
dF
lx
err
, whe
r
e
F
lx
err
= F
lx
ref
– F
lx
est
and
dF
lx
err
= F
lx
err
(t
) –
F
lx
err
(t-
Δ
t)
and
o
n
e
out
put
,
dF
lx
O
.
Fi
gu
re
8 s
h
ows
a bl
ock
di
a
g
ra
m
of FFC
.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
J
PED
S
I
S
SN
:
208
8-8
6
9
4
Title o
f
ma
nu
scrip
t
is sh
o
r
t
an
d clea
r, imp
lies resea
r
ch resu
lts (First Au
tho
r
)
27
3
Fi
gu
re
7.
F
u
zz
y
t
o
r
que
co
nt
r
o
l
l
e
r (FTC
)
bl
oc
k
di
ag
ram
Fi
g
u
re
8
.
Fuzzy
fl
u
x
c
o
nt
r
o
l
l
e
r (
FFC
)
bl
oc
k
di
ag
ram
4.
3
Fuz
z
i
fication o
f
inp
u
ts
a
n
d o
u
tp
uts
B
o
t
h
FTC
a
nd
FFC
are u
s
i
n
g t
h
e sam
e
fuzzy
cont
r
o
l
p
r
ope
r
t
i
e
s. The u
n
i
v
e
r
se o
f
di
sc
ou
rs
e (U
OD
)
fo
r t
h
e
i
n
p
u
t
s
and
o
u
t
put
s a
r
e di
vi
de
d i
n
t
o
7
ove
rl
ap
pi
n
g
sub
s
et
s, +
3
t
o
-3
. Ta
bl
e
1 s
h
ows
t
h
e
de
fi
ni
t
i
ons
o
f
the fuzzy s
u
bs
ets in
UOD.
Fi
gu
re
9 s
h
o
w
s t
h
e i
n
p
u
t
s
an
d
out
put
m
e
m
b
ers
h
i
p
f
unct
i
o
ns
of t
h
e f
u
zz
y
cont
r
o
l
l
e
r.
B
o
t
h
i
n
p
u
t
s
and
out
puts are connected through the
rule
s that e
m
ulate
an ideal second
order syste
m
response which is
ex
p
ected
to
h
a
v
e
go
od
p
e
rfo
r
man
ce (resp
onse ti
m
e
,
o
v
e
rsho
o
t
,
d
a
m
p
in
g
facto
r
and
etc.)
an
d
system
s
t
a
b
ility.
By u
s
ing
t
h
e
IF-
T
H
E
N ru
les, th
e i
n
pu
ts and ou
tpu
t
s ar
e con
n
ected
u
s
ing
4
9
ru
les
sin
ce
7
ov
er
lapp
ing
su
bsets
are
used. T
h
es
e rules are
sum
m
ari
zed and
presented in t
h
e
decision
tab
l
e
as show
n in
Tab
l
e 2.
Tabl
e 1. Fuzzy
su
bset
s defi
ni
t
i
ons
Sy
m
bol
L
i
nguistic
T
e
r
m
s
Fuzzy
Subsets
PL
Positive
Large
+3
PM
Positive Mediu
m
+2
PS Positive
S
m
all
+1
Z
Zero
0
NS Negative
S
m
a
l
l
-1
NM
Negative
Mediu
m
-2
NL
Negative
La
rge
-3
Tabl
e 2. Deci
si
on
t
a
bl
e of
f
u
z
z
y
cont
rol
rul
e
s
T
er
r
/Flx
err
dT
er
r
/Flx
err
PL
PM
PS
Z
NS
NM
NL
PL
PL
PL
PM
PM
PS
PS
Z
PM
PL
PM
PM
PS
PS
Z
NS
PS
PM
PM
PS
PS
Z
NS
NS
Z
PM
PS
PS
Z
NS
NS
NM
NS
PS
PS
Z
NS
NS
NM
NM
NM
PS
Z
NS
NS
NM
NM
NL
NL Z
NS
NS
NM
NM
NL
NL
Evaluation Warning : The document was created with Spire.PDF for Python.
I
S
SN
:
2
088
-86
94
I
J
PED
S
Vo
l. 5
,
No
. 2
,
O
c
tob
e
r
201
4
:
2
68–
2
82
27
4
Fi
gu
re
9.
M
e
m
b
ers
h
i
p
f
u
nct
i
o
ns
of
t
h
e
f
u
zzy
co
nt
r
o
l
l
e
r
(a)
First in
p
u
t
varia
b
le (
b
) Se
con
d
i
n
p
u
t vari
able
(c
) Out
put
va
riables
4.
4
Fuzzy
co
ntro
ller ga
in selectio
n
In o
r
der t
o
get
a sm
al
l
e
r
t
o
rq
u
e
and fl
ux ri
pp
l
e
and fi
xe
d s
w
i
t
c
hi
n
g
fre
q
u
e
ncy
,
t
h
e f
u
zzy
cont
r
o
l
l
e
r
g
a
in
m
u
st b
e
prop
erly selected
to fu
lfill two
main
criteria of torqu
e
and
flu
x
con
t
ro
ller:
i)
Pr
odu
ce a good
d
y
n
a
m
i
cs o
f
f
l
ux
an
d to
rqu
e
r
e
sp
on
se.
ii)
Ab
so
lu
te slop
e o
f
th
e to
rqu
e
an
d
fl
ux
con
t
ro
ller ou
tpu
t
(
T
O
and
F
lx
O
) must not exceed the absol
u
te
slo
p
e
o
f
th
e trian
g
u
l
ar carrier to
av
o
i
d
m
u
ltip
le in
tersectio
n
and
h
e
n
ce
ex
trem
ely h
i
g
h
switch
i
ng
fre
que
ncy
.
By u
s
in
g trial an
d
error m
e
t
h
od
,
sev
e
ral co
m
b
in
atio
n
s
of g
a
i
n
(G
0
, G
1
, G
2
)
ha
ve bee
n
f
o
un
d t
o
achi
e
ve t
h
e des
i
red res
p
o
n
se
. In ge
ne
ral
,
l
o
w
e
r val
u
es o
f
G
0
and G
1
m
u
st be accom
p
anied with a highe
r value
of
G
2
a
n
d vi
ce versa
.
The c
h
aracte
r
istic of eac
h
gain towards t
h
e syst
em
’s response
can be clearly unde
rstood by
refe
rri
n
g
t
o
Ta
bl
e 3. I
n
t
h
i
s
t
a
bl
e, t
h
e cha
nge
s of G
0
and G
1
will affect th
e to
rqu
e
and
flux
resp
on
se time an
d
the controller out
put slope. There
f
ore the G
2
value
needs an
opposite adju
stm
e
n
t
to
co
m
p
en
sate the
increm
ental or decrem
enta
l
effect
pr
od
uce
d
by
G
0
an
d G
1
i
n
o
r
de
r t
o
ac
hi
eve t
h
e desi
r
e
d res
p
on
se. F
i
g. 1
0
shows
an exam
ple of
fu
zzy
c
o
nt
r
o
l
l
e
r o
u
t
put
(
T
O
or
F
lx
O
) wi
t
h
p
r
ope
r gai
n
val
u
e.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
J
PED
S
I
S
SN
:
208
8-8
6
9
4
Title o
f
ma
nu
scrip
t
is sh
o
r
t
an
d clea
r, imp
lies resea
r
ch resu
lts (First Au
tho
r
)
27
5
Tabl
e 3.
Gen
e
ral respon
se
o
f
fu
zzy con
t
ro
ller g
a
in with th
e
increasi
n
g and
decreasi
n
g
val
u
es
of
G
0
,
G
1
and
G2
.
↑↑
and
↓↓
Denotes highly
effected
↑
an
d
↓
Deno
tes m
o
d
e
rately effected
Fig
u
r
e
10
.
Fu
zzy co
n
t
r
o
ller ou
tpu
t
w
ith prop
er g
a
i
n
5.
SIMULATION RESULTS
The p
r
o
p
o
se
d sy
st
em
and a hy
st
e
r
esi
s
-base
d
sy
st
em
have been si
m
u
l
a
ted usi
n
g
MATLAB/Simu
lin
k fo
r
v
a
lidatio
n
p
u
rpo
s
es. Bo
th system
s
u
s
e t
h
e sam
e
in
du
ctio
n m
a
c
h
in
es
p
a
ram
e
te
rs as
sho
w
n i
n
Ta
bl
e 4.
Fo
r t
h
e hy
st
er
esi
s
-base
d
sy
st
em
, t
h
e fl
ux a
nd t
o
r
que
hy
st
eresi
s
ba
nd i
s
set
t
o
10% o
f
t
h
e rat
e
d
values
.
As for the
Fuzzy-PI-base
d co
nt
r
o
l
l
e
r
,
t
h
e val
u
es o
f
G
0
, G
1
a
nd
G
2
are set
t
o
0.
1
5
,
0.
01
, 4
00 a
nd
4,
0.
00
8,
30
0
0
, f
o
r f
u
zzy
t
o
r
q
ue an
d fuzzy
fl
u
x
cont
ro
ller resp
ectiv
ely. Th
e triang
u
l
ar carrier wav
e
fo
rm
s
fre
que
nci
e
s
fo
r
fl
u
x
a
n
d t
o
rq
u
e
f
u
zzy
-P
I
bas
e
d c
ont
rol
l
e
r
ar
e 5
k
H
z a
n
d
1
0
k
H
z
respect
i
v
el
y
.
5.
1
Torque
and
flux s
t
ep
resp
o
nse
A step refe
re
nc
e torq
ue an
d fl
ux at
t
= 0.0
1
s has bee
n
ap
pl
i
e
d t
o
t
h
e DTC
-
C
M
LI sy
st
em
.
Fi
gu
re 1
1
an
d
Figur
e 1
2
sh
ow
a
step
r
e
sp
on
se
of
f
l
ux
an
d
tor
q
u
e
fo
r
bot
h c
ont
rol
l
e
r
s
res
p
ect
i
v
el
y.
It’s clearly i
n
dicated
that the proposed controller
responds
fast
e
r
t
h
an t
h
e hy
st
eresi
s
co
nt
r
o
l
l
e
r w
h
ere
bot
h
t
o
r
que a
nd
fl
u
x
wi
t
h
pr
o
pose
d
c
o
nt
r
o
l
l
e
r ac
hi
eve
d
st
eady
st
at
e at
0.
01
7
1
s a
n
d
0.
14
5s
res
p
ect
i
v
el
y
whi
l
e
bot
h
t
o
r
que
an
d
fl
u
x
wi
t
h
hysteresis c
ont
roller ac
hie
v
ed steady stat
e
at
0.
01
7
6
s
a
n
d 0.
01
5
1
s resp
ect
i
v
el
y
.
In
t
h
e pr
op
ose
d
c
ontr
o
ller (D
TC-CM
L
I with
f
u
zzy
-
P
I c
o
n
t
roller)
, the
de
signate
d f
u
zzy
rules
first
filtered
an
d
pro
cessed
th
e error (t
o
r
q
u
e and
fl
u
x
) to
en
su
re t
h
e larg
e
erro
r
will app
r
o
p
riately trig
ger th
e
correct triangular. Conseque
ntly, the
v
o
ltage v
ector with
h
i
gh
est to
rqu
e
an
d
flux
in
cre
m
en
t (d
ecrem
en
t) is
selected for large e
r
rors
he
nce
produce
f
a
ster to
rqu
e
an
d f
l
ux
r
e
spon
se.
As for
DTC-CMLI with
hysteresis con
t
ro
ller, t
h
e
errors are
directly co
nt
rol
l
e
d by
hy
st
eresi
s
cont
rol
l
e
r w
h
i
c
h pr
o
duces
a
sl
owe
r
res
p
o
n
se com
p
ared
t
o
pr
op
ose
d
c
o
nt
rol
l
er.
Gain
Torque &
Flux
Response
Slope of
T
O
and
Fl
x
O
Response time
Ripple
G
0
Increas
ing
↑↑
↑↑
↑↑
Decreas
ing
↓↓
↓↓
↓↓
G
1
Increas
ing
↓
↓
↑↑
Decreas
ing
↑
↑
↓↓
G
2
Increas
ing
↑↑
↑↑
↑↑
Decreas
ing
↓↓
↓↓
↓↓
Evaluation Warning : The document was created with Spire.PDF for Python.
I
S
SN
:
2
088
-86
94
I
J
PED
S
Vo
l. 5
,
No
. 2
,
O
c
tob
e
r
201
4
:
2
68–
2
82
27
6
Tabl
e
4.
Param
e
t
e
r use
d
i
n
si
m
u
l
a
t
i
on
Induction
Motor
parameters
Cascaded
H-Br
idge Multilev
e
l I
nverter
par
a
meters
Nominal Power
1.5 kW
V
DC
for each
cell
120 V
V
line
-
line
(rms)
400Vrms 50 Hz
Power
electron
ic device
IGBT
Sta
t
or Re
sista
n
ce
3
Ω
Rotor Resistance
3.793
Ω
Referen
c
e
valu
e
Stator Indu
ctan
ce
0.322188 H
Torque r
e
fer
e
nce
9 N.m
Rotor Inductan
c
e
0.330832 H
Flux refer
e
nce
0.9 Wb
Mutual Indu
ctan
ce
0.3049 H
Inertia 0.02799
kg
.m
2
Friction
Factors
0.01025 N.m.s
Pole 2
pole
5.
2
Torque
and
flux rip
p
le
Fi
gu
re
13
an
d
Fi
g
u
re
1
4
s
h
ow
t
h
e
fl
u
x
a
n
d
t
o
r
q
ue
pea
k
-t
o-
peak
ri
pp
l
e
s resp
ect
i
v
el
y
fo
r
bot
h
pr
o
pose
d
c
ont
r
o
l
l
e
r an
d
hy
st
eresi
s
co
nt
r
o
l
l
e
r
.
These
fi
g
u
re
s
cl
earl
y
i
ndi
cat
e t
h
e re
duct
i
o
n
of t
o
r
que a
n
d
fl
u
x
ripple as m
u
ch
as 10% i
n
the
propose
d
c
o
ntro
ller co
m
p
ared to
th
e h
y
steresis con
t
ro
ller.
Using
th
e
p
r
op
o
s
ed
con
t
ro
ll
er in
th
e syst
e
m
,
the
designated fuzzy ru
les fo
r
t
o
rq
ue an
d
f
l
ux
co
n
t
ro
ller are
p
r
op
erly m
o
n
ito
red
an
d corrected
th
e lev
e
l
o
f
e
r
r
o
r
s
t
o
e
n
s
u
r
e
t
h
a
t
the
output fuzzy controller
sig
n
a
ls are
with
in
th
e app
r
op
riate triang
le lev
e
l h
e
n
ce
appropriate voltage vector
e
ither to i
n
crea
se or
decrease
torque or fl
ux is sele
cted and
bee
n
applie
d con
s
ecu
tiv
ely with
i
n
a triang
le wav
e
form
p
e
riod
.
Ho
we
ver by
u
s
i
ng t
h
e hy
st
ere
s
i
s
cont
r
o
l
l
e
r i
n
t
h
e sy
st
em
, the v
o
l
t
a
ge vect
or i
s
ch
osen
ba
sed o
n
t
h
e
com
p
arison of raw error si
gnal w
ith
th
e h
y
steresis b
a
nd
wh
ich
th
e
vo
ltag
e
v
ector for
torq
u
e
and
fl
ux
increasing or decreasing
a
r
e
a
pplie
d
f
o
r
ent
i
r
e swi
t
c
hi
ng
pe
r
i
od.
Figu
re 1
1
. Stator
flu
x
step res
p
o
n
se
(a
)
P
r
o
p
o
se
d c
ont
r
o
l
l
e
r
(
b
)
Hy
st
eresi
s
cont
rol
l
e
r
Evaluation Warning : The document was created with Spire.PDF for Python.
I
J
PED
S
I
S
SN
:
208
8-8
6
9
4
Title o
f
ma
nu
scrip
t
is sh
o
r
t
an
d clea
r, imp
lies resea
r
ch resu
lts (First Au
tho
r
)
27
7
Fi
gu
re
1
2
. T
o
r
que
st
ep
res
p
o
n
se
(a)
Pr
o
p
o
s
ed c
ont
rol
l
e
r
(
b
)
Hy
st
eresi
s
c
ont
rol
l
e
r
Fig
u
r
e
13
. Stato
r
f
l
ux
r
i
pp
le
(a)
Pr
opo
sed
C
o
n
t
ro
ller (b
) Hysteresis
Con
t
ro
ller
Evaluation Warning : The document was created with Spire.PDF for Python.