In
terna
t
io
n
a
l
J
o
u
rn
a
l
o
f
Pow
e
r El
ectro
n
ics
a
nd Dri
v
e
Sy
s
t
em
(
IJ
PE
D
S
)
V
o
l.
11,
N
o.
1,
M
ar
c
h
2
02
0,
pp.
34~
4
4
I
S
S
N
:
2088-
86
94,
D
O
I
:
10.
11
591
/i
j
p
ed
s.
v1
1
.
i1.
pp3
4-
4
4
3
4
Jou
rn
al h
o
me
pa
g
e
:
htt
p
:
/
/ij
p
e
ds.
i
ae
sco
r
e.
com
DTC-ANN-2-level hyb
r
i
d
b
y
neuronal h
ysteresis with
mechanical sensorless in
ductio
n m
o
t
o
r
dr
i
v
e
us
i
n
g
K
U
B
O
TA
observ
er
Dri
s
A
hmed
1
, B
endj
e
b
ba
r Mok
h
ta
r
2
,
B
e
la
i
d
i Aek
3
1,
3
El
ectri
cal E
ngineeri
n
g
Dep
a
rtm
e
nt
, L
AAS Lab
orato
r
y, Ecol
e
Nat
io
nal
e
P
o
l
yt
e
c
hni
qu
e
d'
Oran
M
auri
ce
Au
din,
A
lg
eria
2
Elect
rical En
g
in
eerin
g D
e
partme
n
t
,
U
ST
U
nivers
it
y,
A
l
g
eria
Ar
ticle
Info
AB
S
T
RA
CT
A
r
t
i
c
l
e hi
sto
r
y:
Rec
e
i
ve
d
Ja
n
1
,
2019
R
e
v
i
sed
M
a
r1
,
2
019
A
c
ce
pt
e
d
N
ov
9,
201
9
In
t
his
pap
e
r,
D
T
C
i
s
applied
f
o
r
t
w
o-l
e
vel
in
ve
rt
e
r
f
ed
I
M
dri
v
e
s
based
on
neuro
n
al
hys
tere
s
i
s
co
m
p
arato
r
s
an
d
Th
e
Direct
T
o
r
q
u
e
Co
ntro
l
(
DTC
)
i
s
kno
wn
t
o
p
r
od
uce
qu
ick
an
d
rob
u
s
t
r
esp
onse
i
n
A
C
dri
v
e
s
y
stem
.
However
,
d
u
r
in
g
ste
a
d
y
s
t
a
t
e
,
t
orq
u
e
,
flu
x
a
nd
c
urre
n
t
r
ip
ple
.
A
n
imp
r
ov
em
ent
o
f
e
l
ectri
c
dri
v
e
syst
em
can
b
e
o
b
tain
ed
u
sin
g
a
D
TC
m
eth
o
d
bas
e
d
on
A
NN
s
whi
c
h red
u
ces
t
h
e
t
orqu
e
an
d
f
l
u
x
r
ip
pl
es,
th
e
estim
a
t
ed
t
he
r
o
tor
speed
u
si
ng
the
KU
BOTA
o
bs
e
r
v
e
r
m
e
th
od
b
ased
o
n
m
easu
r
em
ents
o
f
elect
rical
quantiti
es
o
f
the
mot
o
r
. The val
id
i
t
y
of the
p
roposed me
t
hods
i
s
c
o
nfirme
d
b
y
the
s
i
m
u
lati
on
r
esul
ts
.Th
e
T
HD
(Total
H
armon
i
c
Dist
ortion
)
o
f
s
ta
t
o
r
c
u
rre
n
t
,
torq
ue
r
ip
ple
a
n
d
s
t
a
t
o
r
f
lu
x
ripp
le
a
re
d
e
t
e
r
mine
d
a
n
d
com
p
ared
wi
th
c
on
ve
n
tion
a
l
D
T
C
c
o
n
t
ro
l s
c
he
me
u
sing
M
a
t
la
b
/
S
i
mu
l
i
n
k
e
n
vi
ronmen
t
.
Key
w
o
rds:
I
nduc
t
i
on
mot
o
r
dr
ive
KUBOTA
o
b
s
erv
e
r
N
e
ur
ona
l
h
y
ste
r
esis
To
tal
harm
o
n
ic
dis
tort
ion
(T
H
D
)
Tw
o-
l
e
ve
l
D
T
C-
A
N
N
T
h
is
is an
o
p
en access
a
r
ti
cle u
n
d
e
r t
h
e
CC
BY-S
A
li
cens
e
.
C
o
rres
pon
d
i
n
g
A
u
th
or:
Dr
i
s
A
h
m
ed
,
Elec
tr
ica
l
E
ngi
ne
er
i
ng
D
e
par
t
m
e
nt,
LA
AS
L
abor
a
t
or
y,
Ec
o
l
e
N
a
t
i
ona
l
e
P
o
lytec
h
n
i
q
u
e
d
'O
r
a
n
Ma
ur
i
c
e
A
udi
n.
A
lger
ia
.
Em
ail
:
d
r
i
sa
hm
ed8
2
@ya
h
oo.
c
o
m
1.
INTRODUCTI
O
N
The
dir
e
ct
t
or
que
c
o
n
tr
ol
m
e
t
ho
ds
o
f
a
s
yn
c
h
r
o
no
us
m
a
c
hi
nes
a
p
pe
are
d
i
n
t
h
e
sec
o
n
d
ha
l
f
o
f
the
19
8
0
s a
s
c
om
p
e
ti
t
i
v
e
w
i
t
h c
onve
n
t
i
o
nal
me
tho
d
s,
base
d
o
n
pu
lse
w
i
dt
h
mo
dul
a
tio
n
(P
WM
) po
we
r su
ppl
y
and
on
a
spl
i
tti
ng
of
f
l
ux
a
n
d
m
o
tor
t
o
r
que
b
y
m
a
gnet
i
c
f
i
e
l
d
or
ie
n
t
a
ti
on,
I
nde
e
d
,
t
h
e
D
T
C
com
m
a
nd
fr
om
e
x
t
e
r
n
al
r
e
f
er
e
n
ce
s,
s
u
c
h
as
t
or
q
u
e
a
nd
f
l
u
x
,
does
not
s
ea
r
c
h,
a
s
i
n
c
o
nve
nt
i
ona
l
c
o
m
m
a
nds
(
ve
c
t
or
o
r
scala
r
)
t
h
e
v
o
lt
a
g
e
s
t
o
b
e
a
ppl
i
e
d
to
t
h
e
m
ac
hin
e
,
bu
t
se
arc
h
"
th
e
b
e
st
"
s
t
a
t
e
o
f
s
w
i
t
c
h
i
n
g
o
f
t
h
e
i
n
v
e
r
t
e
r
t
o
m
e
e
t
t
h
e
r
e
qu
irem
ents
o
f
the user
[
1]
.
The
des
i
gn
o
f
a
r
t
ific
i
a
l
in
te
l
l
i
g
enc
e
w
as
d
e
v
elo
p
e
d
i
n
t
h
e
ea
rly
1
9
6
0
s
It
i
n
c
lu
de
s
m
e
tho
d
s,
t
oo
ls
a
nd
s
y
s
t
em
s
to
s
o
l
ve
t
he
p
r
o
blem
s
tha
t
n
or
m
a
lly
re
q
u
ire
the
in
te
ll
ige
n
ce
o
f
ma
n
[2].
The
term
in
te
ll
ige
n
ce
i
s
a
lways
de
fi
ne
d
as
t
he
a
bi
l
i
t
y
t
o
lea
r
n
in
a
n
e
f
f
e
c
tiv
e
wa
y
,
t
o
re
ac
t
i
n
a
n
a
d
a
p
ti
v
e
w
a
y
,
to
m
ak
e
t
h
e
ri
g
h
t
d
eci
si
on
s
in
a
s
o
phi
st
i
cat
ed
w
ay
a
n
d
t
o
und
e
r
sta
n
d
ph
e
n
om
en
a
[3].
A
rtificia
l
i
n
te
l
lige
n
ce-
base
d
spee
d c
o
nt
r
o
l
s
(
N
e
ur
al ne
t
w
o
r
k
an
d
f
u
z
z
y
l
o
g
ic)
t
h
a
t
d
o
not
r
eq
uir
e
k
now
l
e
dge
o
f
a
ma
t
h
e
m
a
t
i
c
a
l
m
ode
l
ha
ve
r
e
c
e
n
t
l
y
b
e
e
n
p
r
opo
s
e
d
.
[
4].Fu
z
z
y
l
ogi
c
co
nt
r
o
l
l
e
r
s
a
r
e
i
d
eal
c
an
di
dat
e
s
fo
r
co
nt
ro
lli
ng
s
u
c
h
sy
st
e
m
s,
un
f
o
r
t
u
n
a
t
e
l
y
t
h
er
e
ar
e
no
p
r
e
c
i
se
m
eth
o
d
s
for
de
te
r
m
i
n
in
g
t
h
e
t
uni
n
g
s
t
r
a
t
eg
y.
T
he
l
a
tter
m
u
st
b
e
buil
t
b
y
tr
ia
l an
d er
r
o
r
us
in
g
t
h
e
tes
t
s
on t
h
e sys
t
e
m
t
o
be
a
d
j
us
te
d [
5
]
.
O
n t
h
e
ot
he
r
hand,
t
hese
a
p
p
r
o
ache
s
ha
v
e
go
o
d
r
obust
n
ess to
p
ar
a
m
e
t
r
i
c
var
i
a
t
i
o
ns
a
n
d
m
ea
sur
e
m
e
nt
n
o
i
ses
,
t
he
ir
c
om
pu
ti
ng
c
o
nd
i
t
i
o
ns,
the
e
l
a
bor
a
t
i
o
n
tim
e
a
nd
t
h
e
nee
d
f
or
e
xper
t
k
n
o
w
l
ed
ge
o
f
the
s
y
ste
m
,
lim
it
cur
r
e
nt
ap
p
l
i
c
at
i
ons
t
o
a
lim
ite
d
r
a
nge
a
n
d
s
ome
t
im
es
v
e
ry
sp
ecific [
6
].
Evaluation Warning : The document was created with Spire.PDF for Python.
I
n
t
J
P
o
w
Elec
&
D
r
i S
y
st
I
S
S
N
:
2088-
86
94
D
T
C-
ANN
-
2-
l
e
ve
l
hyb
r
i
d
by n
e
ur
on
a
l
hyste
r
esis
w
i
t
h
m
e
c
h
anic
a
l
se
nso
r
l
e
ss …
(
D
ris
Ahm
e
d)
35
In
r
ecen
t
y
ear
s
ANNs
h
av
e
gain
ed
a
w
id
e
a
t
te
nt
ion
in
c
ontrol
ap
p
lica
t
i
o
ns.
F
o
r
tha
t
,
we
d
e
v
el
o
p
ed
a
n
i
nte
l
lige
n
t
t
e
c
h
n
i
que
t
o
im
pr
o
v
e
the
dy
na
m
i
c
pe
r
f
or
ma
n
ces
o
f
t
h
e
D
TC
.
This
m
e
t
ho
d
co
nsi
s
t
s
i
n
r
e
plac
i
n
g
th
e
trad
iti
on
a
l
S
T
ap
p
l
ied
to
t
h
e
IM
-
D
TC b
y
an
ANNs
[
7
]
.
Ma
jor
d
i
sa
d
v
a
n
ta
ge
o
f
D
T
C
is
t
he
r
ip
p
l
e
on
the
c
o
up
le
a
nd
t
he
f
lu
x
a
nd
t
o
r
em
edy
t
h
is
l
a
s
t
pr
ob
lem
one
i
m
p
r
oves
the
co
n
t
r
o
l
D
T
C
b
y
s
e
v
er
a
l
t
ec
hn
ique
s
a
m
ong
t
hese
m
e
th
o
d
s
a
r
e
m
o
dif
i
c
a
t
i
on
t
h
e
t
a
ble
s
o
f
se
l
e
c
t
i
on,
t
he
a
rtific
ia
l
in
te
ll
ige
n
ce
s
w
h
ic
h
is
i
nte
r
es
ted
in
t
h
is
a
rt
ic
le
a
n
d
t
he
f
lu
x
is
e
s
t
i
m
ate
d
b
y
t
h
e
KUBOTA
o
b
s
erv
e
r.
In
t
hi
s
w
o
rk
,
o
u
r
m
ai
n
obj
ec
t
i
v
e
i
s
t
o
e
xp
loi
t
arti
f
i
ci
a
l
i
nt
e
l
l
i
ge
nce
t
o
o
l
s
nam
e
l
y
,
netw
or
ks
o
f
a
r
ti
f
i
c
i
a
l
n
e
u
ro
ns
o
n
the
DT
C
con
t
ro
l,
a
rtific
ial
ne
ur
al
n
e
t
w
o
r
k
s
o
n
th
e
D
T
C
co
ntrol
DTC-ANN
,
we
u
se
t
h
e
a
d
ap
t
i
ve
o
b
s
e
r
ver
of
K
U
B
O
T
A
to
e
st
i
m
ate
t
h
e
f
l
u
x
,
and
w
e
e
xpr
e
s
s
the
e
s
t
i
m
a
t
i
on
error
t
h
e
n
T
HD
of
s
ta
tor
c
u
r
r
e
nt
i
s
eva
l
ua
t
e
d.
The
m
o
del
l
i
n
g
is
p
r
e
se
nte
d
i
n
Mat
l
a
b
/S
im
u
lin
k
m
ode
ls
i
n
or
de
r
t
o
stu
d
y
t
he
p
er
for
m
anc
e
o
f
th
e
dr
ive
syste
m
u
n
d
e
r
s
tea
d
y
st
a
t
e
a
nd
dy
na
mic
con
d
i
tio
ns
d
ur
i
n
g
s
t
a
r
t
i
n
g,
a
nd
spee
d
r
e
ver
s
a
l
a
n
d
l
oa
d
pe
r
t
ur
bat
i
o
n
s
.
The
s
i
mu
lat
i
o
n
r
e
su
l
t
s
s
how
t
h
a
t
the
pr
op
ose
d
c
o
nt
rol
met
h
od
can
a
ch
i
e
v
e
v
e
r
y
r
o
bu
st
a
nd
sa
tisfa
c
t
or
y
pe
r
f
or
m
a
nc
e.
2.
CONVENTIONAL D
T
C
D
e
pe
n
b
r
o
c
k
an
d
I
.
Taka
h
a
s
h
i
pr
o
pose
d
D
T
C
con
tr
o
l
of
t
h
e
asy
n
chr
ono
us mac
h
i
n
e
i
n
t
h
e
mi
d
-
19
80
s,
it
h
a
s
b
e
c
o
me
i
nc
r
e
a
s
i
n
g
l
y
po
pu
lar
.
T
he
D
T
C
c
on
tr
o
l
m
ak
e
s
i
t
po
ssi
b
le
t
o
c
a
l
cu
la
te
t
he
con
tro
l
q
ua
nt
it
i
e
s
th
a
t
a
r
e
t
h
e
s
t
at
or
f
lu
x
a
n
d
the
e
l
e
c
tr
om
agne
t
i
c
tor
q
ue
f
r
o
m
the
o
n
l
y
q
u
an
t
i
t
ie
s
rela
te
d
to
t
he
s
ta
t
o
r
a
n
d
this
w
i
t
h
o
u
t
t
he
i
n
t
er
ve
n
tio
n
of
m
ec
han
i
ca
l
sens
or
s
[
8
]
.
Th
e
p
r
in
c
i
pl
e
o
f
c
ont
ro
l
is
t
o
ma
in
t
a
in
t
h
e
s
t
a
to
r
fl
ux
i
n
a
ra
nge
.
T
h
e
b
l
oc
k
dia
g
r
a
m
of
t
he
D
T
C
co
nt
r
o
l
i
s
s
ho
w
n
i
n
F
i
gu
r
e
1
.
F
i
gur
e
1.
S
tr
uc
t
u
r
e
o
f
co
n
v
e
n
tio
na
l
D
T
C
.
Th
is
s
tr
ate
gy
i
s
b
ase
d
g
ene
r
a
l
l
y
o
n
t
h
e
use
of
h
yster
e
s
i
s
c
o
m
p
a
rat
o
r
s
w
ho
se
r
ol
e
i
s
t
o
co
nt
r
o
l
t
h
e
a
m
pli
t
u
d
e
s
of
t
h
e
s
ta
tor
fl
u
x
a
nd
t
h
e
e
l
ec
tr
o
m
agne
tic
t
or
q
u
e
:
[
9]
(
1
)
i
s
s
i
s
s
p
T
e
2
3
(
2
)
Evaluation Warning : The document was created with Spire.PDF for Python.
I
S
S
N
:208
8-
8
6
9
4
I
n
t
J
P
o
w
Elec
&
D
r
i
S
y
st
V
ol
.
11,
N
o.
1
,
M
a
r
2020
:
34
–
4
4
36
The
D
T
C
c
o
nt
r
o
l
m
e
th
od
a
l
l
o
w
s
d
ir
e
c
t
an
d
i
nde
pe
nde
n
t
e
lec
t
r
o
ma
g
ne
t
i
c
t
o
r
q
ue
a
nd
f
l
u
x
co
n
t
r
o
l,
sel
ectin
g
an
o
pt
imal switch
i
ng
v
ecto
r.
The
F
i
g
u
r
e
.
2
s
how
s
t
h
e
sche
m
a
tic
o
f
the
ba
si
c
f
unc
t
i
o
n
a
l
b
loc
k
s
u
se
d
t
o
i
mp
le
me
nt
t
he
D
T
C
o
f
in
d
u
ct
i
on m
o
to
r
dr
ive.
A
vol
t
a
ge
s
o
u
r
ce
i
nver
t
er
(
VS
I)
s
upp
lie
s
t
he m
ot
or
a
nd
it
is p
oss
i
ble
to
c
on
tr
o
l
d
ir
ec
tl
y
the
s
t
a
t
or
f
lux
and
t
h
e
elec
t
r
o
m
a
gnet
i
c
t
o
r
q
u
e
by
the
se
lec
t
i
o
n
o
f
o
p
t
i
mum
inver
t
e
r
s
w
itc
hi
n
g
m
odes
[1
0]
.
F
i
gur
e
2.
V
olta
ge
v
ec
tor
s
The
swi
t
c
h
in
g
tab
l
e
a
l
l
o
ws
t
o
se
lect
t
h
e
a
p
p
ropr
iate
i
n
v
erte
r
s
witch
i
n
g
s
ta
t
e
a
cc
o
r
d
i
ng
to
t
he
s
tate
o
f
hys
ter
e
s
i
s
c
o
m
p
ar
at
or
s
o
f
f
lu
x
(
c
f
l
x)
a
n
d
tor
que
(
cc
pl)
a
n
d
t
he
s
e
c
t
or
w
her
e
i
s
the
s
t
at
or
v
ect
or
f
lu
x
(
φ
s
)
i
n
t
h
e
pla
n
(
α
,
β
),
i
n
or
der
t
o
m
ain
t
a
i
n
t
h
e
m
a
gn
i
t
u
d
e
o
f
sta
t
or
f
lux
a
n
d
ele
c
t
r
o
m
a
gnet
i
c
tor
q
u
e
i
nsi
d
e
t
h
e
hy
ster
e
s
is
ba
n
d
s.
The
ab
ove
c
o
n
s
i
d
er
a
t
i
o
n
a
llo
w
s
c
onstr
u
c
t
i
on
of
t
he
s
w
itc
h
i
n
g
t
a
ble
[
1
1]
,
i
s
g
i
ve
n
i
n
T
ab
le
1
.
Tab
l
e
1. The
sele
c
t
i
o
n
o
f
elec
tric
a
l
ten
t
i
o
n
N
1
2
3
4
5
6
Cf
lx
C
cp
l
1
1
2
3
4
5
6
1
0
7
0
7
0
7
0
-
1
6
1
2
3
4
5
0
1
3
4
5
6
1
2
0
0
7
0
7
0
7
-
1
5
6
1
2
3
4
3.
DTC
W
I
TH ARTIFICIAL
NE
U
R
O
N
A
L
N
E
T
WORK
(
DTC-A
N
N
)
Con
v
e
n
ti
ona
l
D
T
C
co
ntr
o
l
has
sever
a
l
di
sad
v
an
ta
ge
s,
s
uc
h
a
s
o
b
t
ain
i
ng
a
v
ari
a
bl
e
s
w
i
t
ch
ing
f
r
e
que
n
c
y,
t
or
que
a
nd
f
lu
x
r
i
p
p
l
es,
p
o
w
e
r
fl
uc
tua
t
io
ns,
a
n
d
ha
r
m
o
ni
c
c
u
rr
ents
i
n
t
h
e
tr
ans
i
e
n
t
an
d
s
t
e
a
d
y
sta
t
e,
be
cause
of t
h
e use
of h
ys
t
e
resis
comparators and
switchi
ng
tab
l
e
s
.
F
o
r
this,
w
e
prop
ose
d
t
o
stud
y
in
t
h
i
s
pa
r
t
t
he
d
ir
ec
t
c
o
n
t
r
o
l
of
t
he
p
air
base
d
o
n
a
r
tif
ic
ia
l
neur
a
l
netw
or
ks,
t
o
i
m
p
r
ove
t
he
p
er
for
m
anc
e
o
f
th
e
D
T
C
c
o
mm
ands,
where
the
co
nve
n
t
i
o
nal
c
o
m
p
ar
ators
an
d
t
h
e
switc
hi
n
g
ta
ble
a
r
e
r
e
plac
ed
b
y
a
ne
ur
a
l
c
ont
r
o
l
l
er
,
so
t
o
dr
i
v
e
t
h
e
ou
t
p
u
t
q
ua
n
tit
ies
of
t
he
M
A
S
t
o
the
i
r
r
e
fe
r
e
nc
e
v
a
l
ue
s
fo
r
a
fi
xed
pe
r
i
o
d
o
f
time
.
N
um
e
r
ical
sim
u
la
ti
ons
a
r
e
p
r
e
sente
d
t
o
te
st
t
he
p
er
f
o
r
m
ance
s
of
t
he
p
r
o
po
sed
metho
d
s
(
DTC-ANN),
is
r
ep
res
e
n
t
ed
by
F
i
g
u
r
e
3
. [1
2
]
.
The
s
t
ruct
ure
of
t
h
e
d
i
r
ect
n
eur
a
l
co
ntr
o
l
of
t
h
e
t
o
r
q
u
e
(DTC-A
NN
-2N
)
,
of
t
he
a
syn
c
hr
on
ou
s
m
a
c
h
i
n
e
p
o
w
e
r
e
d
by
tw
o-
le
ve
l
N
P
C
inver
t
er
,
is
r
e
p
r
e
se
nt
e
d
by
F
i
g.
3.
The
u
p
d
a
t
e
o
f
the
w
e
i
g
hts
a
n
d
Bias
o
f
t
h
i
s
n
e
t
w
o
r
k
i
s
car
ried
o
u
t
b
y
a
r
e
tr
o-
pr
opa
gat
i
on
al
gor
ithm
c
a
lle
d
t
h
e
Leve
nber
g
-
M
a
r
q
u
ar
dt
(
LM)
a
l
gor
it
hm
[
13]
.
The
c
h
o
i
ce
of
n
e
u
r
a
l
ne
tw
o
r
k
a
r
chi
t
ec
t
u
r
e
i
s
ba
sed
o
n
t
he
m
ea
n
s
q
u
a
red
erro
r
(MS
E
)
o
b
tained
dur
in
g
lea
r
ni
ng
[
14]
.
The
f
o
llow
i
n
g
f
ig
ur
e
show
s
the
s
t
r
u
ct
ur
e
o
f
n
e
u
r
a
l
n
e
t
w
o
r
k
s
f
o
r
t
w
o
-
l
e
v
e
l
n
e
u
r
o
n
a
l
D
e
si
g
n
o
f
K
U
B
O
TA
obser
ve
r
.
Th
e
bl
o
c
k
diagram
o
f
t
h
e
DTC-ANN
co
n
t
rol is sho
wn
i
n
F
i
gu
re 3
.
Evaluation Warning : The document was created with Spire.PDF for Python.
Int J
P
o
w
E
l
e
c
& D
r
i
S
y
st
I
S
S
N
:
2088-
86
94
D
T
C-ANN-2-
le
v
e
l hy
bri
d
by
n
e
u
r
on
a
l
hy
ste
r
e
s
i
s
wi
th m
e
ch
an
ic
al se
nso
r
le
ss … (D
ris A
h
m
e
d)
37
F
i
gur
e .
3
D
TC-A
NN
4.
DESIGN OF
HYS
T
ERESIS NE
U
R
O
NAL
C
O
M
PARA
TO
RS
WITH
K
U
B
O
T
A
OBS
E
RVER
Neu
r
a
l
n
et
wo
rk
s
are
mat
h
emat
i
cal
m
od
el
s
in
spi
r
e
d
b
y
t
h
e
b
r
ai
n
'
s
f
unc
t
i
on
i
ng
of
t
he
hum
an
b
e
i
ng.
Th
ei
r
f
a
c
u
l
t
y
o
f
l
e
a
rn
i
n
g
,
g
en
e
r
al
i
z
at
io
n
an
d
ap
p
r
oxi
ma
ti
on
,
ma
ke
t
w
o
n
e
w
s
o
l
u
t
ion
s
f
or
t
he
m
od
e
l
l
i
n
g
,
ide
n
tif
i
c
at
i
on
a
nd
c
o
n
t
ro
l
o
f
p
roce
sse
s
by
t
h
eir
ab
i
l
i
t
y
t
o
pr
o
cess
in
p
u
t-
ou
t
put
d
ata
of
t
he
s
y
s
tem
[1
5
]
.
The
cho
i
ce
o
f
a
neu
r
al
n
e
t
work
t
o
i
m
pro
v
e
the
pe
rform
ance
o
f
t
h
e
pr
op
ose
d
D
TC
c
o
n
tro
l
i
s
o
b
ta
i
n
ed
a
fter
s
eve
r
al
si
m
u
lat
i
on te
st
s.
The
pri
n
c
i
p
l
e
of
n
e
u
ra
l
ne
tw
or
ks
D
T
C
w
it
h
K
U
B
O
T
A
obse
r
ver
is
s
i
mi
lar
to
t
radi
t
i
ona
l
D
T
C
con
t
ro
l.
T
he
d
iffer
e
nce
is
u
s
i
ng
a
ne
ura
l
n
e
t
w
o
r
k
s
c
o
n
t
ro
l
l
e
r
t
o
re
pl
ac
e
th
e
to
rq
ue
a
n
d
f
l
ux
h
yst
e
resi
s
l
o
o
p
con
t
ro
l
l
er,
and
us
i
n
g
K
U
B
O
T
A
obser
v
e
r
for
o
b
s
e
rvi
ng
spee
d
of i
n
duc
t
i
o
n
m
otor.
[16],
[
1
7
]
.
Th
e
hy
st
e
r
esi
s
c
o
m
p
a
rat
o
rs i
s
re
p
l
ac
ed
by
a
pe
rc
e
p
t
r
on
n
eu
ro
n
n
e
t
w
o
rk, co
m
prising a
1 n
e
ur
on i
npu
t
layer
,
a
f
our
n
e
u
ro
n
h
i
d
d
e
n
l
aye
r
,
a
nd
a
1
neur
on
ou
t
p
u
t
l
a
y
er.
T
he
act
i
v
at
i
on
fu
nc
ti
o
n
s
are
of
t
a
n
s
i
g
for
m
s
for
t
h
e
i
n
p
u
t
l
a
y
e
r
and
pur
eli
n
f
or
t
he
h
id
de
n
laye
r
ne
u
r
on,
a
n
d
trai
n
l
m
for
the
ou
t
p
u
t
l
a
y
e
r
n
eur
on
is
ill
us
trate
d
i
n
F
i
gure
4.
[
17].
F
i
gure
.4
A
rchitec
t
ur
e of t
he
n
eura
l
h
y
st
e
r
es
i
s
con
trol
ler
of
t
or
que
a
nd flu
x
.
5.
T
H
E OB
S
E
R
VAT
ION
The
es
t
i
ma
t
o
r
s
u
se
d
in
ope
n
l
oop,
b
ase
d
on
t
h
e
use
o
f
a
c
o
py
of
a
m
ode
l
repr
ese
n
t
a
ti
o
n
o
f
t
h
e
mac
h
i
n
e
.
T
hi
s
a
p
p
r
o
a
ch
l
ed
t
o
t
h
e
i
m
pl
emen
t
a
ti
on
o
f
si
mp
l
e
a
nd
f
a
s
t
a
l
g
o
r
i
t
h
m
s
,
b
u
t
s
e
n
s
i
t
i
v
e
t
o
m
o
d
e
l
l
i
n
g
err
o
rs a
nd par
a
m
e
ter
varia
t
i
o
n
s
duri
ng o
p
e
r
at
io
n [18]
.
I
s
a
n
esti
m
a
t
o
r
oper
a
t
i
ng
in
a
c
l
o
se
d
l
o
op
a
nd
hav
i
ng
a
n
i
nde
pe
nde
n
t
s
y
s
t
e
m
dy
na
mics
.
It
e
st
im
ate
s
a
n
i
n
t
ern
a
l
physi
c
a
l
q
u
a
nti
t
y
o
f
a
giv
e
n
sy
st
e
m
,
b
a
se
d
on
ly
o
n
i
n
f
o
rma
t
i
on
ab
ou
t
t
h
e
inp
u
t
s
a
nd
o
u
t
pu
t
s
o
f
the
Evaluation Warning : The document was created with Spire.PDF for Python.
I
S
S
N
:208
8-
8
6
9
4
I
n
t
J
P
o
w
Elec
&
D
r
i
S
y
st
V
ol
.
11,
N
o.
1
,
M
a
r
2020
:
34
–
4
4
38
ph
ys
ica
l
s
ys
t
e
m
w
ith
t
he
f
ee
dbac
k
i
n
p
u
t
o
f
the
e
r
r
o
r
be
t
w
een
e
s
t
i
m
a
te
d
ou
tpu
t
s
a
nd
ac
tua
l
o
u
t
pu
ts,
us
i
n
g
t
h
e
K
m
a
tr
ix
g
a
i
n
to
t
her
e
by
ad
j
u
s
t
t
he
d
y
n
a
m
ic
c
on
ver
g
e
n
ce
e
r
r
or
[
19]
a
s
sho
w
n
i
n
F
ig
ur
e
5.
Figure
.
5
DTC-ANN
a
ssociat
ed
w
it
h
the
K
U
B
O
TA
obser
ve
r
.
5.
1.
Repr
ese
n
ta
ti
on o
f
the
KUB
O
T
A
Obser
v
er
The
str
u
c
t
ur
e
of
t
he
K
U
B
O
T
A
A
d
apt
i
ve
O
bse
r
ver
i
s
i
l
l
u
str
a
te
d
i
n
F
i
g
u
r
e
6
.
[
1
9
]
,
[
2
0
]
,
[
2
1
]
.
W
h
e
n
the
r
o
ta
t
i
o
n
s
p
e
e
d
o
f
the
M
A
S
is
n
o
t
m
ea
sured,
i
t
i
s
c
o
n
s
i
de
red
a
s
an
u
n
know
n
par
a
m
e
ter
in
t
he
obs
er
ve
r
e
qua
t
i
o
n
s
yste
m
based
o
n
t
h
e
v
ect
or
s
tate
m
ode
l
of
t
he
i
nduc
tio
n
m
a
c
h
i
n
e
descr
i
be
d
in
t
he
s
t
a
tor
fr
a
m
e
and
ha
v
i
ng
as sta
te
v
ec
tor
.
F
i
gur
e
6.
T
he
O
bser
ve
r
of
K
U
B
O
T
A
.
5.
2.
The
Mod
e
lli
n
g
of
th
e Ob
se
rve
r
of
K
UBOT
A
5.
2.
1.
S
t
at
e
m
o
d
e
l
(3
)
Evaluation Warning : The document was created with Spire.PDF for Python.
I
n
t
J
P
o
w
Elec
&
D
r
i S
y
st
I
S
S
N
:
2088-
86
94
D
T
C-
ANN
-
2-
l
e
ve
l
hyb
r
i
d
by n
e
ur
on
a
l
hyste
r
esis
w
i
t
h
m
e
c
h
anic
a
l
se
nso
r
l
e
ss …
(
D
ris
Ahm
e
d)
39
with
:
So
t
h
e
o
b
s
er
v
a
t
o
ry
a
s
s
o
ci
a
t
ed
w
i
t
h
th
i
s
m
o
d
e
l
i
s
w
r
i
t
t
en
a
s
:
(
4
)
With:
B
y
a
ski
n
g
t
h
a
t
estima
t
i
o
n
er
r
o
r
be
t
w
een
t
he
m
odel
a
n
d
t
h
e
obse
rv
er
.
(
5
)
5.
2.
2.
A
d
ap
t
a
t
i
on
m
ec
h
a
n
i
sm
The
s
p
ee
d
ad
j
u
s
t
me
nt
m
ech
a
n
ism
i
s
d
er
i
v
e
d
from
the
a
p
p
l
icat
i
o
n
of
L
YAPUNO
V
th
eo
r
e
m
on
sy
st
e
m
st
a
bi
li
ty
. Le
t
LYA
P
U
N
OV
fu
n
c
ti
on
d
e
f
in
e
d
po
s
i
tive
[2
2
]
:
(
6
)
O
t
he
r
w
ise,
t
he
d
e
r
i
v
at
ive
of
t
h
i
s
fu
nc
ti
o
n
w
it
h
r
e
spect
t
o
ti
m
e
i
s
negat
i
ve
:
(7
)
With:
(8
)
Equ
a
tio
n
(8
)
mu
st
b
e
set
neg
a
tiv
e
acco
rding
to
t
h
e
L
YAP
U
NOV
s
t
a
bil
i
t
y
t
he
or
y.
T
her
e
f
o
r
e
,
by
c
a
re
ful
sele
c
t
ion
of
t
he
g
a
i
n
ma
trix
G
,
the
m
a
trix
Q
m
ust
be
a
ne
gat
i
v
e
de
f
i
nite
m
atr
i
x
an
d
the
a
d
a
p
t
a
ti
on
m
e
c
h
ani
s
m
for
esti
m
a
t
i
n
g
t
he
s
pee
d
w
ill
be reduce
d
b
y c
a
n
c
ell
a
t
ion
o
f
t
he
2
nd
term o
f th
e
eq
u
a
tio
n
(9
)
[
23
].
The
est
i
ma
te
o
f
the
spee
d
i
s
d
one
b
y
the
f
o
l
l
o
w
i
ng
law
:
(
9
)
To
i
mp
rov
e
the
s
p
e
e
d
o
f
dyna
mi
c
o
b
s
erv
a
tio
n
,
p
rop
o
s
e
to
u
se P
I
ins
t
ead
o
f
a
pur
e
in
te
gr
a
t
or
[
24]
:
(
10)
6.
S
I
MULAT
I
O
N
R
E
S
UL
TS
AND
A
NALYSIS
Th
e
di
rec
t
t
o
r
qu
e
cont
r
o
l
app
l
i
e
d
t
o
a
n
in
du
ct
i
o
n
ma
ch
in
e
is
s
i
m
u
late
d
un
der
the
Ma
t
l
ab
/
Sim
u
l
i
n
k
en
vi
ron
m
en
t
.
T
h
e
s
i
m
ul
ati
on
i
s
p
erfo
rme
d
u
nd
e
r
t
h
e
s
ame
co
nd
iti
ons,
Res
ul
t
s
how
n
in
F
i
gur
es
(
F
r
om
7
t
o
13)
.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
S
S
N
:208
8-
8
6
9
4
I
n
t
J
P
o
w
Elec
&
D
r
i
S
y
st
V
ol
.
11,
N
o.
1
,
M
a
r
2020
:
34
–
4
4
40
6.
1.
C
o
mpar
ati
v
e study
b
etwe
en
D
TC a
nd
D
T
C
-hy
s
teresis co
m
pa
r
a
to
rs
Th
e
DTC
and
th
e
DTC-
ANN
ap
p
l
i
e
d
to
a
n
in
d
u
ct
ion
machin
e
is
s
imu
l
a
te
d
u
nder
the
Ma
tla
b
/
S
i
m
u
l
i
nk
e
n
v
i
r
onm
e
n
t.
T
he
s
im
ul
a
t
i
o
n
i
s
p
er
for
m
ed
u
nder
th
e
sam
e
c
ond
it
io
ns.
R
e
s
u
lt
s
h
own
i
n
F
i
g
u
r
e
s (
7
to
9)
t
h
e
to
r
qu
e,
t
he sp
e
ed
and
t
h
e
fl
u
x
.
(a)
(b
)
F
i
gur
e
7.
Tor
q
u
e
r
esponses,
a)
C
l
ass
i
cal
D
TC
c
on
tr
o
l
,
b)
D
T
C
w
i
t
h
ne
ur
al
hys
ter
e
s
i
s
com
p
ar
a
t
or
s
(a)
(b
)
Fig
u
r
e
.8
S
p
e
ed
r
esp
o
n
s
es.
a)
Class
i
cal D
T
C
co
ntrol, b
)
D
TC
w
ith
ne
ur
a
l
hys
ter
e
sis
c
o
mpa
r
at
or
s
(a)
(b
)
F
i
gur
e
.
9
T
he
s
tat
o
r
fl
ux.
a
)
Cla
s
sica
l
D
T
C
c
ontr
o
l,
b
)
D
TC
w
i
t
h
n
e
u
r
a
l
hy
ste
r
esis
c
om
pa
r
a
tor
s
0
0.
5
1
1.
5
2
2.
5
-2
0
0
20
40
60
80
100
ti
m
e
(
s
)
to
r
q
u
e
t
e
m
(
N
.
m
)
0
0.
5
1
1.
5
2
2.
5
-20
0
20
40
60
80
100
120
ti
m
e
(
s)
T
e
m(
N
.
m)
0
0.
5
1
1.
5
2
2.
5
-
200
0
200
400
600
800
1
000
1
200
1
400
ti
m
e
(
s
)
s
p
ee
d
of
r
o
t
a
t
i
o
n(
t
r
/
m
i
n
)
0
0.
5
1
1.
5
2
2.
5
-
200
0
200
400
600
800
1000
1200
1400
ti
me
(
s
)
t
he
s
peed
of
r
ot
at
i
on(t
r
/
m
i
n
)
0
0.
5
1
1.
5
2
2.
5
0
0.
5
1
1.
5
2
2.
5
ti
m
e
(
s
)
th
e
fl
u
x
sta
to
r
i
q
u
e
(
W
e
b
)
0
0.
5
1
1.
5
2
2.
5
0
0.
5
1
1.
5
2
2.
5
ti
m
e
(
s
)
fl
u
x
(
W
b
)
Evaluation Warning : The document was created with Spire.PDF for Python.
I
n
t
J
P
o
w
Elec
&
D
r
i S
y
st
I
S
S
N
:
2088-
86
94
D
T
C-
ANN
-
2-
l
e
ve
l
hyb
r
i
d
by n
e
ur
on
a
l
hyste
r
esis
w
i
t
h
m
e
c
h
anic
a
l
se
nso
r
l
e
ss …
(
D
ris
Ahm
e
d)
41
6.
2.
C
o
ntro
l
DTC-
ANN
s
en
s
o
rless (KUBO
T
A
O
b
serv
er)
to
estima
t
e th
e
sp
eed
a
n
d t
h
e
flux
R
e
s
u
lt
s
h
o
wn
i
n
Fi
gure
1
0
t
o
e
s
tim
ate
the
spee
d,
e
s
t
im
a
t
e
the
f
lu
x
a
n
d
e
v
a
l
ua
t
i
o
n
o
f
the
er
r
o
r
e
s
tim
ati
on.
(a)
(
b
)
(c)
(d
)
Fig
u
r
e 10
.
DT
C-ANN
co
n
t
rol with
KU
B
OTA, a)
Th
e est
i
mation
of
t
h
e
f
l
u
x
,
b)
Est
i
ma
t
i
on
er
r
o
r
of
f
l
u
x
,
c
)
The
est
i
m
a
t
i
o
n
of
t
he
s
pe
e
d
,
d)
E
stim
atio
n
er
r
o
r
of
s
pee
d
6.
3.
Test
o
f
t
he ro
b
u
s
tnes
s
in lo
w
s
pe
ed
o
f the KUBOTA ob
s
er
v
e
r
R
e
s
u
lt
sh
ow
n
i
n
F
i
gur
e
1
1
t
e
s
t
of
t
he
l
ow
s
pee
d
.
F
i
gur
e
.
11
Te
st
in
g
i
n
l
ow
s
pee
d
(
≈10Ra
d
/s)
0
0.
5
1
1.
5
2
2.
5
3
3.
5
0
10
20
30
40
50
60
ti
m
e
(
s
)
t
h
e
s
p
ee
d
(
r
a
d
/
s
)
Wr
We
s
t
Evaluation Warning : The document was created with Spire.PDF for Python.
I
S
S
N
:208
8-
8
6
9
4
I
n
t
J
P
o
w
Elec
&
D
r
i
S
y
st
V
ol
.
11,
N
o.
1
,
M
a
r
2020
:
34
–
4
4
42
6.
4.
THD of
th
e
c
u
r
ren
t
st
at
or
(D
T
C
, DTC-
A
NN
)
R
e
s
u
lt
sh
ow
n
i
n
F
i
gur
e
s
(
12
to
13)
t
he
T
H
D
of
t
he
c
ur
r
e
nt
s
tat
o
r
(DTC an
d
DTC-AN
N).
F
i
gur
e
.
12
TH
D
value
o
f
c
las
s
ica
l
D
TC
Fig
u
r
e .
1
3
THD
v
a
lu
e
o
f
DT
C
-
ANN
6.
5.
A
n
a
l
y
s
i
s
a
nd discu
ssi
o
n
The
figur
es
s
how
t
ha
t
t
h
e
si
m
u
la
t
i
o
n
r
esu
lts
u
s
i
n
g
a
rtific
ia
l
i
nte
l
lige
n
ce
t
e
ch
n
i
que
s
(
n
e
u
r
a
l
hys
ter
e
s
i
s)
a
n
d
D
TC-
A
N
N
s
how
t
ha
t
t
h
e
tr
a
c
k
i
n
g
o
f
t
h
e
se
t
p
o
in
t
i
s
per
f
ec
t
.
W
e
n
o
te
t
ha
t
the
r
i
pple
o
f
e
l
e
c
tr
om
agne
t
i
c
t
or
que
a
n
d
s
t
a
tor
f
l
ux
r
e
d
uces
p
er
f
ect
l
y
c
om
p
a
r
ed
t
o
co
nv
en
ti
o
n
a
l
DT
C
wi
th
out
n
eu
ral
hys
ter
e
s
i
s
c
o
mpa
r
at
or
I
t
is
m
or
e
a
ppar
e
nt
t
hr
o
u
g
h
t
he
t
r
a
jec
t
or
y
o
f
t
h
e
s
ta
tor
fl
ux
I
n
a
d
d
it
io
n
t
o
a
l
ar
ge
de
cr
e
a
se
i
n
T
H
D
a
s
s
how
n
in
t
he
t
a
b
le
a
b
ove
,
We
w
e
r
e
a
b
l
e
t
o
c
o
ncl
ude
t
hat
the
D
T
C
c
o
n
t
r
o
l
b
y
ne
ur
al
hys
ter
e
s
i
s
sho
w
e
d
g
o
od
per
f
o
r
m
a
n
ce
tha
n
t
he
c
lass
ica
l
D
TC
c
o
n
tr
o
l
but th
e
DTC-ANN
is
m
o
s
t
ex
cellen
t
.
These
r
e
s
u
l
t
s
pr
o
v
e
t
h
a
t
our
s
e
n
s
o
r
l
ess
c
o
nt
r
o
l
w
i
t
h
a
dap
t
ati
o
n
o
f
is
i
nse
n
s
iti
ve
t
o
t
h
e
v
a
r
i
at
io
ns
o
f
the
sta
t
or
r
esis
t
a
nce
s
.
I
t
i
s
also
n
ot
i
c
e
d
t
hat
the
obse
r
ver
co
r
r
e
c
t
s
w
e
l
l
th
e
ro
t
o
r
f
l
u
x
(
t
h
e
s
q
u
a
re
o
f
t
h
e
r
o
t
o
r
fl
ux
)
a
n
d
th
e
sp
ee
d
of
r
ot
a
t
i
o
n
,
s
in
c
e
t
h
e
e
sti
m
at
ed
q
u
a
n
titi
es
f
o
llo
w
a
n
acce
p
t
abl
e
w
a
y
t
h
e
act
u
a
l
magn
i
t
ud
e
s
of
t
he
m
a
c
hi
ne
,
h
enc
e
a
tr
a
c
k
i
ng e
r
r
o
r
is a
lm
os
t
ze
r
o
be
t
w
e
en
t
he
two s
ize
s
, Thi
s
i
m
p
lies a
sta
b
le
obse
r
v
a
ti
o
n
.
B
u
t
w
e
h
ave
a
pr
o
b
lem
of
t
he
r
ipple
s
,
e
s
pec
i
all
y
f
o
r
t
he
o
bse
r
v
er
o
f
KUBOTA.
S
i
m
u
l
a
t
i
o
n
r
esul
ts
s
how
t
ha
t
usin
g
the
o
b
s
e
r
v
e
r
i
s
im
por
tan
t
i
n
t
he
c
o
n
tr
ol
o
f
the
ma
chi
n
e,
t
he
e
s
tim
ati
on
er
ro
r
a
s
z
e
r
o
i
n
t
h
e
s
t
ea
dy
s
ta
te
,
The
ma
j
o
r
a
d
van
t
a
g
e
for
K
U
B
O
TA
o
bser
vat
i
on
t
e
c
h
n
i
q
u
e
it’
s
in
sens
i
t
i
v
i
t
y
t
o
t
he
m
ac
h
i
n
e
s
etti
ng
s
an
d
h
i
s
r
e
spo
n
d
i
n
g
t
o
low
s
p
e
e
d
s
t
h
a
t
a
r
e
n
e
a
r
t
o
1
0
R
a
d
/
s
,
T
h
i
s
p
r
o
v
e
s
its robustness.
To
r
ed
uce
t
h
e
bur
de
n
of
P
I
tu
n
i
n
g
a
n
d
t
o
en
ha
nc
e
the
dr
i
v
e
pe
r
f
or
m
a
nc
e
at
l
ow
s
peed
r
e
g
i
o
n
a
d
ap
t
i
ve
P
I
contr
o
l
l
er
i
n
MR
A
S
i
s
r
e
plac
ed
by
a
r
t
i
f
i
c
i
al
i
nte
l
l
i
g
e
n
t
N
e
u
r
o
fuzz
y
c
o
n
t
r
o
lle
r
.
A
n
exha
us
ti
ve
a
n
al
ys
i
s
i
s
c
a
r
r
i
ed
o
u
t
w
i
t
h
MRA
S
obse
r
ver
w
i
t
h
r
ot
or
f
l
u
x
and
rea
c
t
i
ve
p
ower
s
che
m
e
s
w
ith
P
I
and
NFC
as
a
d
ap
t
i
ve
c
ontr
o
l
l
er
s
and
sim
u
lat
i
on
r
e
su
l
t
s
ar
e
c
o
mpa
r
ed
a
nd
sh
ow
n
in
p
a
p
e
r
s
[
24,
13]
.
Tab
l
e
2
s
h
ow
s
t
h
a
t
t
he
s
i
m
ul
ati
o
n
r
e
su
lts
u
sing
a
r
t
i
f
i
c
i
a
l
i
nte
ll
ig
en
ce
t
e
c
h
n
i
q
u
e
s
(n
e
u
ral
hy
st
ere
s
i
s
)
show
t
ha
t
the
t
r
a
c
ki
ng
o
f
t
h
e
set
p
o
i
n
t
is
p
e
r
fe
ct.
We
n
o
t
e
t
h
a
t
t
h
e
r
i
p
p
l
e
o
f
e
lectr
o
m
a
gne
t
i
c
t
o
r
q
ue
a
nd
sta
t
o
r
fl
ux
r
edu
c
e
s
p
e
r
f
e
ctl
y
c
o
m
p
a
r
e
d
to
c
onve
n
tio
n
a
l
D
T
C
wi
tho
u
t
n
eu
r
a
l
hyste
r
e
s
i
s
c
o
m
p
a
r
a
t
or
I
t
i
s
m
or
e
a
ppar
e
nt
t
hr
o
u
gh
the
tr
a
j
e
c
t
o
r
y
o
f
t
h
e
s
t
at
or
f
l
u
x
I
n
a
dd
iti
o
n
t
o
a
l
a
r
g
e
d
e
c
r
e
a
s
e
i
n
T
H
D
a
s
s
h
o
w
n
i
n
t
h
e
t
a
b
l
e
a
b
o
v
e,
W
e
w
e
r
e
a
ble
to
c
o
n
c
l
u
d
e
tha
t
t
he
D
TC
c
on
tr
o
l
by
neur
a
l
hy
st
ere
s
i
s
s
h
o
w
ed
good
p
e
r
fo
rma
n
ce
t
h
an
the
cla
ssic
a
l
D
TC
c
o
n
tr
ol,
co
m
p
ar
e
d
t
o
t
h
e
pape
r
s
[
22,
25]
.
D
e
t
ail
i
s
s
ho
w
n
i
n
Ta
b
l
e
2.
Tab
l
e
2.
The
pe
r
f
or
m
a
nce
s
o
f
K
U
B
O
T
A
obse
r
ver
P
r
e
c
i
s
i
on
swiftne
ss
Osc
ill
a
tion
Low spe
e
d
D
T
C
-
A
N
N
w
ith
KUB
O
T
A
obse
r
ve
r
More
a
cc
ur
a
t
e
F
a
st
b
e
l
ow
Mi
ssing
E
x
c
e
l
l
e
nt
≈
10
Ra
d/s
Evaluation Warning : The document was created with Spire.PDF for Python.
Int J
P
o
w
E
l
e
c
& D
r
i
S
y
st
I
S
S
N
:
2088-
86
94
D
T
C-ANN-2-
le
v
e
l hy
bri
d
by
n
e
u
r
on
a
l
hy
ste
r
e
s
i
s
wi
th m
e
ch
an
ic
al se
nso
r
le
ss … (D
ris A
h
m
e
d)
43
Ta
b
l
e
3
s
how
s
tha
t
t
he
s
im
u
l
a
t
i
o
n
re
su
l
t
s
u
s
i
ng
art
i
fic
i
a
l
i
nte
l
lige
n
ce
te
c
h
n
i
que
s
(neur
a
l
h
y
s
t
e
r
esis)
show
t
hat
t
h
e
t
r
ac
king
o
f the
set po
i
n
t is per
fec
t
. Co
n
v
en
t
i
on
a
l D
T
C w
i
t
h
ou
t
ne
ura
l
h
yste
resis c
o
mpa
r
ator
it
i
s
mor
e
a
ppar
e
n
t
t
hr
ou
g
h
t
he
t
ra
ject
ory
of
t
he
s
tat
o
r
flu
x
i
n
ad
di
t
i
o
n
t
o
a
lar
g
e
decr
ease
i
n
T
HD
as
s
hown
i
n
t
h
e
tab
l
e
a
b
o
v
e
,
w
e
w
e
r
e
a
bl
e
t
o
c
onc
l
ude
t
hat
t
h
e
D
T
C
c
o
nt
r
o
l
b
y
n
eu
ral
hy
st
e
r
esi
s
s
ho
we
d
g
ood
p
erfo
rma
n
ce
tha
n
t
he
c
lass
ic
al
D
TC
c
on
tro
l
.
Tab
l
e.
3 Com
p
a
r
iso
n
b
etw
e
e
n
the
P
er
form
anc
e
s of
C
on
ve
ntio
na
l
D
T
C
a
n
d
D
TC
-
A
N
N
.
M
i
ni
m
i
z
a
tions
ripple
s
o
f
t
h
e
t
orque
M
ini
m
iza
tions
r
ipple
s
o
f
the
flux
Ia
s TH
D (%)
C
onve
nt
iona
l
D
T
C
E
x
ist
E
x
ist
27.
77
DTC
–
AN
N
Few
F
e
w
12.
28
7.
CONCL
U
S
IONS
I
n
t
hi
s
w
o
rk,
w
e
ma
i
n
ly
p
rese
nte
d
t
he
e
s
tim
ati
o
n
of
t
he
r
o
t
or
f
l
ux
b
y
th
e
KUB
OTA
ad
ap
ti
v
e
s
t
a
t
e
obs
erve
r,
t
hen
w
e
e
va
l
u
a
t
e
d
t
he
e
stim
at
ion
e
r
ror
of
t
he
f
lu
x
,
w
e
also
d
e
v
o
t
ed
t
o
im
pro
v
e
th
e
perfor
ma
nc
es
o
f
the
direc
t
c
on
t
r
ol
o
f
th
e
tor
q
ue
o
f
the
a
s
ync
hro
n
o
u
s
tw
o-leve
l
U
P
S
po
w
e
red
ma
chin
e
base
d
o
n
a
rt
ific
ial
in
t
e
l
l
i
g
e
n
ce
t
e
c
h
n
i
que
s
b
y
D
TC-A
N
N
,
The
s
i
m
u
la
t
i
on
r
esul
t
s
s
how
t
ha
t
the
use
of
b
ot
h
es
ti
ma
t
o
r
s
i
s
impor
ta
nt
i
n
t
h
e
con
t
rol
of
t
h
e
i
n
duc
t
i
o
n
m
ac
hine
,
the
tra
n
sie
n
t
an
d
ve
r
y
s
hor
t
r
e
gim
e
a
nd
the
e
rror
be
t
w
e
e
n
the
f
l
ux
e
s
t
i
m
ate
d
a
n
d
m
ea
sur
e
d
t
o
zer
o
i
n
t
he
s
te
ad
y
st
ate,
t
h
e
r
o
b
u
s
t
n
e
s
s
t
e
s
t
s
o
f
t
h
e
e
s
t
i
m
a
t
o
r
a
r
e
a
l
s
o
v
e
ri
fi
e
d
.t
he
obs
erv
e
r
o
f
K
U
B
OTA
al
so
p
l
a
y
i
t
s
r
ol
e
,
a
nd
g
iv
e
go
od
re
s
u
lt.
w
e
c
a
n
s
ay
t
he
u
se
o
f
the
e
s
ti
m
a
tor
bri
ngs
a
c
lea
r
i
m
p
rovem
e
n
t
t
o
the
lo
o
p
ed
s
truc
tur
e
.
N
o
te
t
h
a
t
t
he
r
esea
rch
w
o
r
k
i
s
ve
ry
f
e
w
,
espe
cial
ly
w
i
t
h
rega
rd to
the
o
b
ser
v
e
r
o
f K
U
B
O
T
A
so I w
a
n
t t
o
e
x
p
an
d f
u
rther
an
d us
i
n
g
differe
nt
c
on
tr
ol
s.
I
n
o
r
d
er
t
o
im
p
r
ove
t
he
p
er
for
m
ance
o
f
t
h
e
D
T
C
(t
orq
u
e
r
i
pp
le
r
e
duc
t
i
o
n
s,
f
l
ux,
a
nd
t
he
T
H
D
va
lue
of
t
he
s
tat
o
r
curre
nt
)
,
s
im
ul
a
t
i
o
n
tes
t
s
o
f
t
he
c
o
n
tr
ol
b
y
vari
a
t
i
o
n
an
d
i
nverse
l
y
of
t
h
e
l
oa
d
t
o
rq
ue,
w
e
r
e
prese
n
t
e
d,
t
he
r
e
s
ults
o
b
t
a
i
ne
d
sh
ow
t
h
i
s
s
t
r
a
teg
i
es
p
ro
po
s
e
d
w
i
t
h
the
t
e
c
h
n
i
que
s
o
f
t
he
a
rtific
ia
l
i
n
tel
l
i
g
e
n
c
e
(DTC-ANN an
d
DTC
-
hy
s
t
eresi
s
n
eu
r
o
n
a
l
)
are
v
ery
p
o
w
erfu
l
and
ro
b
usts.
REFE
RENCES
[1]
F
,
M
oran
d,
"
Te
chn
i
q
u
e
d’O
b
serv
ati
on
san
s
C
a
p
t
e
urs
de
V
itesse
e
n
vue
d
e
la
C
om
m
a
nd
e
d
e
s
M
achin
e
s
S
y
nch
r
one",
T
hès
e
d
e
D
o
ct
orat,
I
n
st
itue
Na
tio
na
l
de
s Sc
ie
n
c
e
s
Ap
p
l
iq
ué
e
s
L
y
on
Fran
c
e
20
05
.
[2]
A
m
eur,"Comm
a
nd
e
sans
C
ap
te
u
r
d
e
V
i
t
e
sse
p
ar
D
T
C
d
’un
e
M
ach
ine
S
yn
chro
ne
à
A
im
ant
s
d
o
t
é
d’u
n
O
b
s
e
rvateu
r d’o
r
dre Co
m
p
le
t
à SM
C", M
ém
o
i
re de M
a
g
i
ster en
El
ec
trot
echn
i
q
u
e,
U
n
i
vers
i
té Ba
t
na
20
03
.
[3]
Al-Rou
h,
"
Co
ntr
i
bu
tion
à
la
C
o
mma
n
d
e
sa
ns
Ca
p
t
e
u
r
d
e
l
a
MAS"
,
Thès
e
de
D
o
c
to
rat,
Un
iv Henri
P
o
i
n
car
é
,
Na
ncy -1
, Ju
ill
e
t
200
4
.
[4]
V
.
B
o
s
t
a
n
,
M
.
C
u
i
b
u
s
,
C
.
I
l
a
s
,
G
.
G
r
i
v
a
,
F
.
P
r
o
f
u
m
o
,
R
.
B
o
j
o
i
,
"
Ge
ne
ra
l
Ad
a
p
ta
tio
n
l
a
w
f
o
r
M
RAS
high
P
e
rf
o
r
m
a
n
ce
S
e
ns
orl
e
ss
i
n
duc
t
i
on
M
o
to
r
dri
v
es
"
,
P
E
SC,
Va
n
c
ou
v
e
r, Ca
na
da
,
Ju
n
e
1
7-22
.
[5]
D
.
Y
acin
e
,
"Con
trô
l
e
d
e
l
a
F
r
éq
uenc
e
de
C
o
m
m
u
t
a
tio
n
des
Hy
st
éré
s
i
s
uti
l
i
s
és
d
a
n
s
les
Comm
andes
d’une
M
ach
in
e à Ind
u
ct
ion
"
, M
é
m
o
i
r
e de M
ag
ister,
Universi
té de
Batna,
20
07.
[6]
D
.
A
h
m
ed,
"
Etu
de
d
es
D
iff
é
rent
es
S
t
r
at
égi
e
s
d
e
C
omm
a
nd
e
no
n
Li
néai
re
d
e
l
a
M
achin
e
A
s
yn
chron
e
a
ve
c
E
s
t
i
m
a
tion
du Flux
et d
e
la
V
i
t
e
s
s
e"
,
M
é
mo
ire d
e
M
agi
s
t
e
r,
Ecole
Na
ti
on
a
l
e Polytech
ni
qu
e d’O
r
an,
20
15
.
[7]
A
.
Hamm
o
u
m
i
,
A
.
M
a
ss
ou
m,
A
.M
erou
f
e
l,
P
.W
ira,
"
A
ppl
icatio
n
des
Rés
eaux
d
e
N
e
uron
es
p
o
u
r
la
C
om
m
a
n
d
e
de
l
a
M
achi
n
e
As
y
n
chro
ne
s
an
s
Cap
t
eur
M
écan
iqu
e
"
,
A
rti
c
le,
Med
i
m
i
ra science
Publ
isher,
vol
u
m
e
5
3
,
N
u
m
b
er
2
,
20
12
.
[8]
L
.
Yo
ub,
A
.
C
racu
nes
c
u, "E
t
ud
e
Com
p
arat
iv
e
e
n
tre
l
a
C
o
m
m
a
n
d
e
V
éct
o
r
iell
e
a
F
l
ux
O
ri
ent
e
e
t
la
C
o
m
m
a
n
d
e
D
i
rect
e du
Cou
p
l
e
de l
a
M
achi
n
e A
s
yn
chro
ne",
U.P
.
B.SCI.Bull
,
series C,
Vol.
69, N
O
.2,
2007.
[9]
D
r
aou
i
.
A
bd
e
l
ghani
1
,
All
a
oua
B
ou
med
i
èn
e
2
,
“D
irect
T
o
r
q
u
e
Control
of
T
w
o
I
nducti
on
M
ot
ors
U
s
in
g
th
e
Ni
ne
-
S
w
it
ch
I
nv
erter",
In
ter
n
a
t
i
ona
l Jou
r
n
a
l
of Power El
ectr
oni
cs
a
nd D
r
ive S
y
stem
(IJP
E
D
S
)
,
Vol.
9
,
No.
4
,
D
ecem
ber 2
018
, p
p.
155
2
~
1
564
.
[1
0]
S
a
rra
M
ass
o
u
m
,
Abd
e
lk
ader
M
erou
f
e
l,
A
b
d
errah
i
m
Bent
a
a
llah
,
F
ati
m
a
Z
oh
ra
B
el
aim
eche,
A
h
m
ed
M
as
so
u
m
,
"S
ens
o
rl
ess
F
u
zzy
S
li
d
i
ng
M
o
d
e
S
peed
C
o
n
t
r
oller
f
o
r
Ind
u
ct
io
n
M
otor
w
ith
D
TC
b
ased
o
n
Artifici
a
l
Neural
N
e
tw
orks
",
Ws
eas
Tra
n
s
a
cti
o
n
s
on
Power Sys
t
e
m
s,
Vo
l
u
m
e
11
,
20
16.
[1
1]
D
j
amila
C
herifi
,
Y
a
hi
a
M
i
l
o
u
d
,
"
Ro
bus
t
S
p
eed
-
S
en
so
rless
V
ecto
r
C
o
n
tro
l
o
f
Doubl
y
F
e
d
Ind
u
ction
M
o
tor
Dri
v
e
U
s
i
ng
Sl
iding
Mode
R
ot
or
F
lu
x
Ob
serv
e
r
",
Int
e
rn
ati
o
n
a
l
Jo
ur
nal
of
A
p
p
l
ied
Po
wer
E
n
g
i
n
eerin
g (IJAP
E),
Vol.
7,
N
o
.3
, D
ecem
ber 20
18
, p
p
. 2
48~
2
6
3.
[1
2]
Ch
aym
ae
F
a
has
s
a,
M
oh
amed
A
kherraz
,
Abd
e
rra
him
Benn
ass
a
r,
"
S
en
s
o
rl
ess
DTC
of
a
n
In
duction
M
o
t
o
r
bas
e
d
on
I
nte
llig
e
n
t
D
u
a
l
Ob
se
rv
e
r
a
nd
A
NN
ba
se
d
Se
le
c
t
or
T
a
b
le
"
,
W
s
eas
T
r
a
n
s
a
c
t
ions on syst
ems
an
d
co
nt
rol,
v
o
l
u
m
e
1
3
,
201
8
.
[1
3]
M
i
n
i
R
,
Sh
aban
a
Ba
cker
P
,
B.Hari
ram
S
a
th
eesh
,
D
i
n
es
h
M
.
N
,
"
L
o
w
S
p
eed
E
stim
ati
on
of
S
en
so
rles
s
D
T
C
In
du
cti
on
M
o
t
o
r
D
r
iv
e
Usin
g
M
RA
S
w
i
th
N
euro
F
uzzy
A
d
a
pti
v
e
Cont
roller",
In
ter
nat
io
nal
Jou
r
n
a
l
of
E
l
ectr
i
cal
an
d Co
mp
u
t
e
r
E
n
g
i
ne
e
r
in
g
(IJ
E
CE
),
Vol.
8
,
N
o
.
5,
p
p
.
2
6
91~2
702.
,
Oct
ober
201
8
Evaluation Warning : The document was created with Spire.PDF for Python.