Int
ern
at
i
onal
Journ
al of Ele
ctrical
an
d
Co
mput
e
r
En
gin
eeri
ng
(IJ
E
C
E)
Vo
l.
10
,
No.
3
,
June
2020
,
pp. 3
035~
3046
IS
S
N: 20
88
-
8708
,
DOI: 10
.11
591/
ijece
.
v10
i
3
.
pp3035
-
30
46
3035
Journ
al h
om
e
page
:
http:
//
ij
ece.i
aesc
or
e.c
om/i
nd
ex
.ph
p/IJ
ECE
A robust
diagn
osis meth
od
for speed s
ensor fault b
ased on
stator
currents in
the RF
OC indu
ctio
n m
oto
r drive
Cuon
g Dinh T
ran
1
,
Pavel Br
an
d
s
tetter
2
,
Mi
nh
C
hau H
uu
Ngu
yen
3
,
Sa
n
g D
ang H
o
4
, B
ach Ho
ang
Dinh
5
, P
hu
ong
Nhat
Ph
am
6
1,
4,
6
Facul
t
y
of El
ec
tr
ic
a
l and El
e
c
troni
cs
Engi
ne
er
ing,
Ton
Duc
Th
ang
Univer
si
t
y
,
Viet
nam
5
Pow
er
S
y
stem
Optimiza
ti
o
n
Re
sea
rch
Group
,
F
ac
ul
t
y
of
Elec
tr
i
ca
l
and
Elec
t
roni
cs
Engi
n
ee
rin
g,
Ton
Duc Thang
Univer
sit
y
,
Vie
t
nam
1,
2
,3,4,6
Facul
t
y
of Electrical E
ng
in
ee
ring
and
Com
pute
r
Sc
ie
nc
e, V
SB
-
T
ec
hni
ca
l
U
nive
rsit
y
of
Os
tr
ava
,
Cz
ec
h
Repu
bli
c
Art
ic
le
In
f
o
ABSTR
A
CT
Art
ic
le
history:
Re
cei
ved
A
ug
27
, 201
9
Re
vised
N
ov
2
6
,
2019
Accepte
d
Dec
10, 201
9
A
val
id
d
ia
gnos
is
m
et
hod
for
th
e
spee
d
sensor
f
ai
lur
e
(SS
F)
is
a
n
essential
req
uire
m
ent
to
e
nsure
the
reliab
ilit
y
of
Fault
-
Tol
e
ran
t
Control
(FT
C)
m
odel
s
in
induction
m
otor
drive
(
IMD
)
sy
st
ems
.
Mos
t
rec
en
t
rese
arc
hes
h
av
e
foc
used
on
d
ire
c
tly
compar
ing
th
e
m
ea
sured
and
esti
m
at
ed
rotor
s
pee
d
sign
a
l
to
detec
t
the
spe
ed
sensor
fau
lt.
How
eve
r,
using
tha
t
suc
h
esti
m
ated
val
u
e
in
both
the
f
aul
t
d
ia
gnosis
and
th
e
cont
rol
le
r
r
ec
o
nfigura
t
ion
phas
es
le
ads
to
the
insuffi
cient
per
form
anc
e
o
f
FTC
m
odes.
In
t
his
pape
r
,
a
nov
el
di
agnosis
-
te
chn
ique
b
ase
d
on
th
e
sta
tor
cur
ren
t
m
odel
c
om
bine
d
with
a
conf
usion
pre
vention
c
ond
it
ion
is
propose
d
to
det
ect
the
fai
lur
e
stat
es
of
the
spee
d
sensor
in
the
IMD
sy
stems
.
It
hel
ps
the
FTC
m
ode
to
sepa
rat
e
be
twee
n
the
di
agnosis
a
nd
rec
onf
igura
t
i
on
phase
s
agai
nst
a
spee
d
se
nsor
fau
lt.
Thi
s
proposed
SS
F
dia
gnosis
m
et
hod
ca
n
al
s
o
eff
ectiv
ely
ap
pl
y
fo
r
IMs
’
appl
i
ca
t
ions
at
the
low
-
spe
ed
r
ange
wher
e
th
e
spee
d
sensor
s
igna
l
of
te
n
suffers
from
noi
se.
MA
TL
AB/S
imulink
software
has
bee
n
used
to
implemen
t
the
sim
ula
ti
ons
in
var
ious
spee
d
ran
g
es.
The
ac
h
ie
v
ed
result
s
have
demons
tra
te
d
th
e
ca
p
abi
l
it
y
and
eff
ective
n
ess
of
the
proposed
SS
F
m
et
hod
aga
inst
spee
d
se
nsor fa
ult
s
.
Ke
yw
or
d
s
:
Diag
nosis
Fault
-
toler
ant
-
con
t
ro
l
(F
TC
)
Ind
uction m
oto
r
(
IM)
RFOC
Sp
ee
d
se
nsor
Copyright
©
202
0
Instit
ut
e
o
f Ad
vanc
ed
Engi
n
ee
r
ing
and
S
cienc
e
.
Al
l
rights re
serv
ed
.
Corres
pond
in
g
Aut
h
or
:
Ba
ch Ho
a
ng
D
inh
,
Power Sy
ste
m
Op
ti
m
iz
ation
Re
search
Grou
p,
Faculty
of Elec
tric
al
an
d El
ect
ronics E
nginee
rin
g,
To
n Du
c
Th
a
ng
Un
i
ver
sit
y,
19 Ng
uyen H
uu T
ho, Dist
rict
7,
H
o
Chi
Mi
nh Ci
ty
, V
ie
tna
m
.
Em
a
il
: dinh
hoa
ngbac
h@
t
dtu
.e
du.vn
NOM
EN
CLA
TUR
E
S
S
Stat
or
flu
x vec
tor
i
n
[α,
β] co
ordinate syst
e
m
S
R
Rotor fl
ux v
ect
or in
[α, β] c
oord
i
nate syst
em
S
S
i
Stat
or
c
urre
nt
vecto
r
in
[
α
, β
]
coor
din
at
e sys
tem
S
R
i
Rotor cu
rr
e
nt
ve
ct
or
i
n
[
α
,
β]
coor
din
at
e syst
e
m
S
S
u
Stat
or
volt
age
vecto
r
in
[
α
, β
]
coor
din
at
e sys
tem
,
SS
uu
Stat
or
volt
age
com
po
ne
nt in
[α
, β] syst
e
m
,
S
x
S
y
uu
Stat
or
volt
age
com
po
ne
nt in
[x,
y]
syst
em
Evaluation Warning : The document was created with Spire.PDF for Python.
IS
S
N
:
2088
-
8708
In
t J
Elec
&
C
om
p
En
g,
V
ol.
10
, No
.
3
,
J
une
2020
:
30
35
-
3046
3036
,,
abc
u
u
u
Stat
or
volt
age
com
po
ne
nt in
[a
, b, c] syst
em
Sx
i
Flux cu
rr
e
nt c
om
po
nen
t
Sy
i
Torq
ue
c
urren
t
co
m
po
ne
nt
m
i
Ma
gn
et
iz
in
g
c
urren
t
,
SR
RR
Stat
or
a
nd r
otor r
e
sist
ance
,
SR
LL
Stat
or
a
nd roto
r
in
duct
ance
m
L
Ma
gn
et
iz
in
g
in
du
ct
a
nce
R
T
Rotor t
i
m
e consta
nt
m
Me
chan
ic
al
a
ngula
r
s
peed
en
Me
chan
ic
al
a
ngula
r
s
peed o
f t
he
enc
oder
p
Po
le
pair n
um
ber
R
No
m
inal ro
t
or
flu
x
,
RR
Rotor fl
ux
com
pone
nt in [α, β
]
syst
e
m
Rotor fl
ux angl
e
1.
INTROD
U
C
TION
Du
e
to
t
he
a
dvanta
ges
of
ec
onom
ic
s,
ruggedn
e
ss,
sel
f
-
st
arti
ng,
a
nd
sta
ble
in
operati
on
,
t
he
t
hr
ee
-
ph
a
se
inducti
on
m
oto
r
(I
M)
has
bec
om
e
the
m
os
t
popu
la
r
el
ect
rical
m
a
chine
in
the
i
ndus
try
.
In
th
e
past,
IMs
we
re
of
te
n
a
pp
li
ed
in
t
he
unco
ntr
ollabl
e
sp
ee
d
a
ppli
cat
ion
s.
N
owad
ay
s,
with
the
a
dv
a
nce
d
te
c
hnologies
in
high
-
perf
orm
ance
power
c
onve
rters
an
d
m
od
ern
co
ntr
ol
al
go
rithm
s,
IMs
can
op
e
rate
m
or
e
flexibl
y
in
bo
th
fixe
d
-
s
pee
d
a
nd
var
ia
ble
-
s
pee
d dr
i
ves [1
]
.
Ty
pical
ly
,
the
separ
at
e
co
ntr
ols
of
sp
e
ed
a
nd
to
rq
ue
are
two
pri
m
ary
req
ui
rem
ents
for
al
l
el
ect
rical
dr
i
ves
.
Var
i
ous
m
e
tho
ds
ha
ve
bee
n
pr
opose
d
to
c
ontr
ol
IM
s’
dri
ve
syst
e
m
s
and
cl
assi
fied
int
o
tw
o
m
ai
n
gro
up
s:
S
cal
a
r
C
on
tr
ol
(S
C
)
a
nd
V
ect
or
C
on
trol
(
VC).
In
SC
te
chn
iq
ues
,
t
he
tor
que
an
d
sp
ee
d
are
c
on
tr
ol
le
d
base
d
on
a
c
onsta
nt
vo
lt
s
-
pe
r
-
her
tz
rati
o
pr
inciple
.
It
is
si
m
ple,
and
no
s
ens
or
s
require
d,
bu
t
it
is
una
ble
to
con
t
ro
l
the
t
orqu
e
a
nd
s
peed
of
IM
s
at
the
sam
e
tim
e
[2
]
.
O
n
the
oth
e
r
hand,
t
he
VC
te
chn
iq
ue
in
wh
ic
h
the
r
otor
fiel
d
-
or
ie
nted
c
on
t
r
ol
(RF
OC)
is
the
m
os
t
ty
pic
al
m
et
ho
d
ca
n
overc
om
e
that
prob
le
m
of
t
he
SC
te
chn
iq
ue
.
In
the
RFOC
m
od
el
,
the
sta
tor
-
c
urren
t
vect
or
is
separ
at
ed
int
o
two
el
e
m
ents
i
Sx
and
i
Sy
,
wh
ic
h
are
perpe
nd
ic
ular
to
each
ot
her
.
As
a
res
ult,
the
torque
an
d
the
r
oto
r
flu
x
can
be
in
de
penden
tl
y
co
ntro
ll
ed
by
adjustin
g
ei
the
r
i
Sx
or
i
Sy
,
resp
ect
ively
.
It
is
si
m
il
ar
to
the
con
t
ro
l
pr
i
ncipl
es
of
sepa
rate
ly
excit
ed
DC
m
oto
rs,
wh
e
re
bot
h
th
e
tor
qu
e
a
nd
sp
ee
d
can
be
con
t
ro
ll
ed
pr
e
ci
sel
y
at
the
s
a
m
e
tim
e
in
a
wide
s
peed
r
ange
.
Th
us
,
this
RF
OC
m
od
el
can
be
ap
plied
to
m
any
com
ple
x
co
ntr
ol
ap
plica
ti
on
s
in
the
industry.
Howeve
r
,
t
o
achieve
hi
gh
perf
or
m
ance
,
RFOC
nee
ds
the
accurat
e
ly
structur
al
pa
ram
et
ers
of
IMs
as
well
a
s
the
feedbac
k
sign
a
ls
from
senso
rs
.
Thus,
the
qua
li
ty
of
IM
D
s
de
pends
on
not
on
ly
the
co
nd
it
ion
of
the
m
ac
hin
e
s
(both
elec
tric
al
and m
echan
ic
al
p
arts
) bu
t al
s
o
the
stat
es of
sens
or
s
(
reli
able or
fail
ure
)
[
3].
In
pract
ic
e,
ab
norm
al
op
erati
on
of
IMs
co
ul
d
be
ha
ppene
d
du
e
t
o
the
m
al
functi
on
s
of
m
echan
ic
al
/
el
ect
rical
par
ts
or
se
nso
r
error
s
wh
e
re
the
sp
ee
d
sens
or
f
ai
lure
is
on
e
of
the
m
os
t
sever
e
pro
blem
s
[4
-
6]
.
It
m
akes
the
loss
or
inacc
ur
acy
in
feedbac
k
sig
nals
an
d
thu
s
le
a
d
s
to
i
ncorr
ect
act
io
ns
of
t
he
co
ntr
oller
.
Fu
rt
her
m
or
e,
if
s
uch
sen
sor
fail
ur
es
can
not
be
detect
ed
a
nd
so
l
ve
d
qui
ckly
,
it
can
le
ad
to
m
or
e
se
rio
us
pro
blem
s
and
then
ca
us
e
s
to
dam
age
the
whole
IM
dri
ve
syst
e
m
[7
]
.
I
n
order
to
li
m
i
t
the
ef
fect
of
fau
lt
y
sens
or
s
i
n
the
op
e
rati
on
of
I
Ms
,
FTC
m
et
ho
ds
ha
ve
bee
n
dev
el
op
e
d
rece
ntly
to
m
ai
ntain
the
sta
ble
operati
on
of
IMs
.
Th
os
e m
et
ho
ds
are
cl
assifi
ed
into
tw
o
m
ajo
r
gro
ups
,
Acti
ve
FTC
and
Pas
sive
FT
C,
wh
e
re
the
pa
ssive
te
chn
iq
ues
are
of
te
n
desig
ned
to
work
offli
ne
again
st
so
m
e
prede
fine
d
fail
ur
es
with
ou
t
c
hangin
g
t
he
c
ontr
ol
structu
re
[8
,
9].
On
the
c
ontrar
y,
the
act
ive
s
olu
ti
ons
nee
d
on
li
ne
che
cki
ng
the
sta
tus
of
sens
ors
a
nd
r
econfig
ur
e
the
co
ntr
ol
str
uctu
re
if
fin
din
g
any
fail
ur
e
of
the
act
ual
m
easur
e
m
ent
syst
e
m
.
T
he
act
ive
FTC
sy
stem
s
com
pr
ise
three
seq
uen
ti
al
ste
ps
:
fa
ult
detect
ion
,
fa
ult
isolat
ion
,
an
d
recon
f
igurat
ion
,
w
he
r
e
the
fa
ult
dete
ct
ion
and
f
ault
isolat
ion
ca
n
al
so
b
e
com
bin
ed
in
one
ste
p
cal
le
d
t
he
fa
ult
diag
nosis.
In
t
his
pa
pe
r,
we
only
fo
c
us
on
the d
ia
gnos
is
of s
peed se
ns
or
fau
lt
based o
n
t
he
act
ive
FTC
schem
e.
The
m
os
t
com
m
on
app
r
oac
h
in
the
SSF
dia
gnos
is
is
base
d
on
the
c
om
pari
so
n
betwee
n
the
physi
cal
sens
or
si
gn
al
s
and
sta
te
va
ria
bles
of
the
est
im
at
or
(
or
obse
rv
e
r
)
.
A
ny
sp
e
ed
sen
sor
fa
ult
will
be
detect
ed
if
the
error
bet
ween
ph
ysi
cal
and
est
i
m
at
e
d
sign
al
s
exc
eeds
a
sp
eci
fi
c
threshold
[
10
-
12]
.
Accordi
ng
t
o
the
RFOC
m
e
thod,
the
el
ect
rical
torque
c
an
be
co
ntr
olled
by
ad
j
ust
in
g
the
i
Sy
el
e
m
ent
in
the
ro
t
at
ing
Evaluation Warning : The document was created with Spire.PDF for Python.
In
t J
Elec
&
C
om
p
En
g
IS
S
N: 20
88
-
8708
A rob
us
t
diagnosis
meth
od fo
r sp
ee
d
s
ens
or
fau
lt
base
d
on
stator c
ur
re
nts
in
the…
(
Cu
ong Di
nh Tr
an
)
3037
coor
din
at
e
syst
e
m
[x
,
y
]
.
M
or
e
over,
due
to
the
relat
io
nship
betwee
n
th
e
el
ect
rical
torq
ue
a
nd
the
spe
ed
of
IM
s
,
to
e
nhan
ce
the
ef
fecti
ve
ness
a
nd
st
ab
il
ity
of
the
fa
ult
diag
nosis
al
gorithm
,
FTC
need
s
f
our
sign
al
s
includi
ng
physi
cal
and
est
i
m
a
te
d
ro
t
or
s
peed
s
as
well
as
the
i
Sy
e
lem
ent
of
the
m
easur
e
d
and
est
im
at
ed
st
at
or
current
s
[6
,
13]
.
An
ot
her
a
ppro
ac
h
f
or
diag
no
si
ng
a
sp
eed
sensor
fa
ult
based
on
a
n
a
da
ptive
r
oto
r
-
re
sist
anc
e
ob
s
er
ver
has
be
en
pr
opos
e
d
i
n
[
5].
It
was
ba
sed
on
t
he
id
ea
that
if
there
is
a
sign
al
fail
ur
e
f
r
om
the
sp
ee
d
sens
or
occurri
ng,
the
est
i
m
ated
r
oto
r
resist
ance
is
sign
ifi
cantl
y
changed
in
it
s
a
m
pli
tud
e
.
T
hu
s
,
the
auth
ors
analy
zed
t
he diffe
ren
ce
b
et
we
en
the
esti
m
a
ted
a
nd r
eal
ro
t
or
resist
ance
s
to dete
rm
ine the
SSF
case
.
In
t
his p
a
pe
r,
we
propose
a
nov
el
m
et
ho
d c
om
bin
ing
t
wo
m
et
ho
dolo
gies
base
d
on
sta
to
r
cu
r
re
nts and
base
d
on
t
he
r
efere
nce
ro
t
or
sp
ee
d
to
dia
gnose
any
fail
ure
of
t
he
s
pee
d
se
nsor
.
T
he
pro
posed
dia
gnos
is
al
gorithm
inclu
de
s
tw
o
m
ai
n
par
ts.
The
first,
we
c
om
par
e
the
m
easur
ed
a
nd
e
stim
at
ed
sta
tor
c
urren
t
s
t
oget
he
r
to
detect
the
possible
fau
lt
in
the
sp
e
ed
se
nsor
.
Ne
xt,
a
di
ff
e
ren
ce
betwe
en
the
m
easure
d
a
nd
t
he
re
f
eren
c
e
rot
or
spe
eds
is
check
e
d
to
pr
even
t
the
co
nf
us
io
n
by
the
f
ai
lure
of
c
urre
nt
sens
or
s
the
m
se
lves.
It
m
e
ans
that
we
co
uld
base
on
the
dif
fer
e
nc
e
betwee
n
the
m
easur
ed
a
nd
est
i
m
at
ed
sta
to
r
curre
nt
s
to
di
agnose
a
ny
possible
fau
lt
s
in
the
sp
ee
d
sens
or
instea
d
of
dir
e
ct
ly
check
ing
between
m
ea
su
re
d
an
d
est
im
at
ed
ro
tor
s
peeds
.
This
propose
d
SSF
diag
nosi
s
m
e
tho
d
can
eff
ect
ively
ap
ply
for
IMs’
a
pp
li
cat
io
ns
at
the
lo
w
-
s
pee
d
range
wh
e
re
t
he
s
pee
d
se
nsor
sig
nal
of
te
n
s
uffe
rs
f
ro
m
no
ise
w
hile
oth
e
r
sim
il
ar
al
gorithm
s
would
be
c
om
plica
te
d
to
ha
nd
le
i
n
th
at
su
ch
c
onditi
on.
Be
sides
,
an
est
i
m
ation
val
ue
of
r
otor
-
ti
m
e
const
ant
(RT
C)
is
al
so
a
pp
li
ed
t
o
enh
a
nce
t
he
a
ccur
acy
of
the
pro
posed
dia
gnos
is
m
et
ho
d.
The
sim
ulatio
ns
of
F
TC
in
MATL
AB/Si
m
ul
ink
hav
e
been
im
plem
ented
in
var
i
ou
s
s
p
ee
d
ranges
to
ve
rify
the
eff
ect
iveness
of
the
pr
op
os
e
d
m
e
thod.
The
res
ults
show
that
the
FT
C
us
ing
the
propose
d
dia
gnosi
s
al
go
rithm
e
ff
ect
ively
wor
ks
in
both
no
r
m
al
and
low
-
sp
ee
d
r
an
ges.
T
he
r
em
a
ining
of
this
pap
e
r
is
org
an
iz
ed
as
f
ollows:
In
Sect
io
n
2,
t
he
m
ai
n
ideas
of
the
pro
posed
f
ault
diag
no
sis
m
et
ho
d
are
de
scribe
d
in
deta
il
s;
in
Sect
ion
3,
the
perform
ance
of
the
pr
opos
e
d
appr
oach
is
ve
rified
thr
ough
si
m
ulati
on
s
of
FTC
i
n
var
i
ous
operati
ng
s
peeds;
an
d
fin
al
ly
,
con
cl
us
io
ns
a
re
giv
e
n
in
Secti
on
4
2.
FAU
LT
DI
A
GNOSIS
AL
GORIT
HM
F
OR
T
HE SPE
ED SE
NS
O
R FAILU
R
E
This
sect
io
n
i
nc
lud
es
tw
o
c
onte
nts.
The
m
at
hem
atical
m
od
el
base
d
RF
OC
of
IMs
is
pr
ese
nted
in
the f
ir
st, a
nd th
e m
a
in idea
of
the pr
opos
e
d f
ault diag
nosis
al
gorithm
o
f
F
TC i
s d
esc
ribe
d
in
d
et
ai
l l
at
er
.
2
.
1.
M
athem
ati
cal m
od
el
of
IMD ac
cor
din
g to the
RFO
C
t
echniqu
e
Accor
ding
to
the
RFOC
te
c
hn
i
qu
e
,
the
s
pace
vect
or
of
sta
tor
c
urre
nts
is
separ
at
ed
into
tw
o
perpe
nd
ic
ular
el
e
m
ents
i
Sx
a
nd
i
Sy
in
the
ro
ta
ti
ng
c
oor
din
at
e
syst
em
[x
,y]
,
as
show
n
i
n
Fig
ure
1
[
14]
wh
e
r
e
the
m
agn
et
iz
ing
cu
rr
e
nt
co
rr
e
sp
on
ding
to
th
e
ro
to
r
fl
ux
is
sta
nd
i
ng
on
t
he
axis
x.
In
thi
s
way,
the
r
otor
flu
x
can b
e
c
ontr
olled b
y
i
Sx
,
a
nd
i
Sy
is
us
ed
to
co
ntr
ol
the
el
ect
rical
tor
qu
e
o
f
I
Ms
sim
il
ar
to
t
he
i
dea
of
co
nt
ro
ll
in
g
separ
at
e
ly
e
xcite
d
D
C
m
oto
r
s
.
Figure
1. Pr
i
nc
iple o
f vect
or c
on
t
ro
l
The
th
ree
-
phas
e
sta
tor
cu
rr
e
nt
s
and
the
r
otor
sp
ee
d
sig
nal
m
easur
ed
fro
m
senso
r
s
an
d
the
volt
age
sign
al
s
cal
cula
te
d
from
swit
chin
g
signa
ls
of
the
in
ver
te
r
com
bin
ing
DC
li
nk
volt
age
are
trans
fe
rr
e
d
to
the
RF
OC
c
ontrol
lo
op,
as
s
how
n
i
n
Fig
ur
e
2.
The
Cl
ar
k’
s
f
or
m
ula
is
use
d
t
o
tra
nsfo
rm
the
qu
a
ntit
ie
s
f
ro
m
the
three
-
ph
as
e
syst
e
m
[a,
b,
c]
to
the
sta
t
ion
a
ry
co
ordin
at
e
syst
e
m
[α,
β]
as
show
n
i
n
(
1)
an
d
t
he
Par
k’
s
form
ula
is
app
l
ie
d
to
tra
nsfo
r
m
the
sta
tor
c
urre
nts
f
ro
m
the
sta
ti
on
ary
c
oo
rd
i
nate
syst
em
[α,
β]
to
t
he
r
ot
at
ing
coor
din
at
e
syst
e
m
[x
,
y]
as
sh
own
by
(
2)
.
T
hat
current
m
od
el
of
IMs
is
app
li
ed
to
cal
cu
la
te
the
m
agn
e
ti
zi
ng
current
c
orres
pondin
g
to
th
e
ro
t
or
flu
x,
flu
x
an
gula
r
sp
ee
d,
a
nd
rot
or
fl
ux
a
n
gle
[15].
W
e
ca
n
se
e
the tra
ns
f
or
m
at
ion
s
as
fo
ll
owi
ng
:
Evaluation Warning : The document was created with Spire.PDF for Python.
IS
S
N
:
2088
-
8708
In
t J
Elec
&
C
om
p
En
g,
V
ol.
10
, No
.
3
,
J
une
2020
:
30
35
-
3046
3038
10
12
33
S
Sa
S
Sb
i
i
i
i
(1)
c
o
s
s
i
n
s
i
n
c
o
s
Sx
S
Sy
S
i
i
i
i
(2)
The dynam
ic
m
od
el
o
f IMs i
n
the
stat
ion
a
r
y sy
stem
[
α
, β] is desc
ribe
d
as
the
fo
ll
owin
g form
ulas:
;0
;
S
S
S
S
S
S
S
R
S
S
S
R
R
R
R
S
S
S
S
S
S
S
S
S
m
R
R
m
S
R
R
d
d
R
R
j
d
t
d
t
L
L
L
L
u
i
i
i
i
i
i
(3)
a
nd the elect
ric
al
torqu
e
prod
uc
ed by t
he
i
nduc
ti
on
m
oto
r
is
pr
ese
nted
b
y
3
22
e
R
S
S
R
m
R
L
p
T
i
i
L
(4)
Fr
om
the
m
ath
em
atical
relation
s
hip
s
desc
ri
bed
ab
ove,
the
con
t
ro
l
blo
c
k
diag
ram
based
on
RF
OC
app
li
ed
in
this
pap
e
r
i
s
ex
pr
e
s
sed
i
n
Fi
gure
2.
T
he
c
on
t
ro
ll
e
r
e
xecu
te
s
the
con
t
ro
l
l
oops
to
ac
hieve
the
c
on
t
ro
l
var
ia
bles,
an
d
then
the
se
va
riables
are
tra
nsfo
rm
ed
fr
om
t
he
r
otati
ng
refe
ren
ce
fr
am
e
back
to
the
sta
ti
on
a
r
y
ref
e
ren
ce
f
ram
e
by
rev
ersi
ng
the
Park’s
tra
nsfo
rm
at
ion
.
Finall
y,
the
rev
er
se
Cl
ark
’s
tra
ns
form
at
ion
is
a
pp
li
ed
to
transfo
rm
the
con
tr
ol
var
ia
bles
from
the
t
wo
-
phase
syst
e
m
[α,
β]
to
a
th
ree
-
ph
a
se
syst
em
[a,
b,
c]
and
these
ref
e
ren
ce
var
i
ables
are
em
p
loye
d
to
co
ntr
ol
the
switc
hin
g
in
ver
te
r
by
the
sine
pu
ls
e
width
m
od
ul
at
ion
(S
P
WM)
m
et
h
od.
T
hus,
the
perform
ance
of
the
co
ntr
oller
base
d
on
RF
OC
is
cl
os
el
y
relat
ed
to
the
a
ccur
ac
y
of
physi
cal
sign
al
s
m
easur
e
d
f
ro
m
the
current
a
nd
spe
ed
se
ns
ors,
and
any
a
bnorm
al
con
dit
io
n
of
the m
easur
em
e
nt syst
em
co
uld
a
ff
ect
th
e
operati
on quali
ty
o
f
I
MD
s.
Figure
2. Co
ntr
ol b
l
ock d
ia
gr
a
m
b
ased on
t
he
RFOC c
om
bin
ed wit
h
F
T
C
Evaluation Warning : The document was created with Spire.PDF for Python.
In
t J
Elec
&
C
om
p
En
g
IS
S
N: 20
88
-
8708
A rob
us
t
diagnosis
meth
od fo
r sp
ee
d
s
ens
or
fau
lt
base
d
on
stator c
ur
re
nts
in
the…
(
Cu
ong Di
nh Tr
an
)
3039
In
order
to
im
pr
ove
the
IM
Ds
’
operati
on
un
der
a
bnorm
al
c
onditi
ons
w
here
the
m
easur
e
m
ent
sign
al
s
are
ins
uffic
ie
nt
,
an
FTC
m
od
ule
is
ad
de
d
on
the
c
ontr
ol
schem
es
to
di
agnose
a
nd
re
so
lve
the
fail
ur
e
of
sens
or
s
.
Its
m
e
thodo
l
og
y
is
ba
sed
on
com
par
in
g
the
physi
cal
sign
al
s
from
sensors
an
d
t
he
est
i
m
at
ed
values
from
ob
servers
(estim
at
or
s).
I
t
wo
r
ks
li
ke
a
switc
h
bet
wee
n
two
c
ontrolli
ng
m
od
es,
nor
m
al
and
fau
lt
tolerat
e
m
od
es.
At
nor
m
al
con
diti
on,
the
co
nt
r
oller
us
es
t
he
m
easur
e
d
ro
t
or
-
sp
e
ed
sig
nal
as
t
he
feedbac
k
si
gnal
to
i
m
ple
m
ent
the
con
t
ro
l
act
io
n,
bu
t
if
ther
e
is
any
m
alf
unct
io
n
ha
pp
ened
in
the
r
otor
-
s
pee
d
se
ns
or,
it
is
switc
hed
t
o
FTC
m
od
e where
t
he
physi
c
al
sign
al
will
be
rep
la
ce
d
by
t
he
est
im
a
te
d
v
al
ues.
T
hus,
w
e
need
to s
up
e
rv
ise
and
detect
an
y
po
ssible fail
ure
of that s
uc
h
s
pe
ed
se
nsor
, a
nd
that i
s the aim
o
f
this
researc
h
.
2
.
2.
The
dia
gnosis
algorit
h
m to de
tect
fai
lure
s in the s
p
eed sens
or
As
m
entioned
in
the
pre
vious
par
t,
t
he
r
otor
sp
ee
d
sig
nal
f
ro
m
the
sens
or
has
a
n
im
po
rtant
r
ole
in
determ
ining
th
e
sta
te
var
ia
bles
of
IMDs
,
li
ke
the
ro
to
r
f
lux
,
el
ect
r
om
a
gn
et
ic
tor
que,
et
c.
in
the
c
on
t
ro
l
schem
e.
Ther
e
fore,
i
f
occurri
ng
any
fail
ures
of
the
s
peed
s
ens
or
duri
ng
operati
on,
we
ne
ed
to
detect
,
iso
la
te
,
and
rep
la
ce
th
at
insu
f
fici
ent
sign
al
by
a
m
or
e
su
it
able
sig
nal
i
m
m
ediat
e
l
y.
In
the
pa
st,
m
os
t
of
the
exi
sti
ng
SSF
diag
nosis
m
e
tho
ds
[10
-
12
]
directl
y
de
te
ct
the
sp
e
ed
sens
or
fail
ure
accor
ding
t
o
t
he
dev
ia
ti
on
be
tween
the
m
easur
ed
s
peed
sig
nal
an
d
the
est
i
m
at
e
d
sig
nal
.
The
be
low
c
onditi
on
is
a
fa
ult
dete
ct
or
wh
e
re
a
f
ault
of
the s
peed
sen
s
or
can
b
e
deci
de
d
if
the e
rror
exceed
s a s
pec
ific
thr
es
hold
[
11
]
as
sho
wn
by
_
_
e
s
t
R
_
r
e
f
R
_
r
e
f
F;
0
.
0
4
5
;
2
0
0
0
.
1
;
2
0
0
s
p
e
e
d
R
m
R
t
h
r
e
s
h
o
l
d
i
f
r
o
t
o
r
s
p
e
e
d
r
p
m
t
h
r
e
s
h
o
l
d
i
f
r
o
t
o
r
s
p
e
e
d
r
p
m
(5)
Con
se
quently
,
the
est
im
a
te
d
sp
ee
d
sig
nal
is
us
e
d
i
n
bo
t
h
the
dia
gnos
is
ph
a
se
a
nd
the
reconfi
gur
e
ph
a
se
of
t
he
c
ontr
ol sch
em
e in the
FTC m
ode [12].
It m
akes th
e syst
em
h
igh
ly
dep
e
ndin
g on the se
ns
it
ivit
y of
the
sp
ee
d
est
i
m
at
ion
al
gorithm
that
cou
ld
l
ead
to
i
nsuf
fici
ent
pe
rfor
m
anc
e
w
hen
t
her
e
is
any
ba
d
i
nterfer
enc
e
in
that
s
uc
h
est
i
m
at
or
’s
qu
al
it
y.
More
over
,
a
t
the
lo
w
-
s
pee
d
range,
the
s
pe
ed
se
nsor
sig
nal
oft
en
suffe
rs
f
r
om
no
ise
.
Th
us
,
t
he
res
ult
of
t
he
SSF
diag
nosis
co
uld
be
seriousl
y
aff
e
ct
ed
by
the
a
ccur
acy
of
the
sp
eed
m
easur
em
ent
s
yst
e
m
.
Ther
ef
or
e
,
it
is
n
eces
sary
to
dev
el
op
so
m
e
new
ideas
to
enh
a
nce
the
per
f
or
m
an
ce
of
the FTC m
od
e
in v
a
rio
us
sp
ee
d
z
on
e
s.
In
this
pro
pose
d
F
TC
a
ppr
oac
h,
we
sug
gest
t
he
SSF
dia
gnosi
s
al
gorithm
ba
sed
on
t
he
c
urren
t
m
od
el
wh
e
re
the
co
m
par
ison
bet
w
een
the
m
easur
ed
a
nd
est
im
at
ed
sta
tor
cu
r
ren
ts
is
ap
plied
to
detect
the
sp
eed
sens
or
fail
ur
e
.
Af
te
r
t
hat,
th
e
sp
ee
d
sig
nal
in
the
c
on
tr
ol
le
r
based
RF
OC
us
es
the
e
stim
at
ed
ro
tor
sp
ee
d
determ
ined
from
the
m
od
el
ref
er
e
nce
adap
ti
ve
syst
e
m
(MRAS)
[
15
-
17]
in
the
reconfigu
rati
on
ste
p.
Fr
om
th
e
curr
ent
m
od
el
of
I
Ms
[18],
the
a
ct
ual
sta
tor
cu
rr
e
nts
can
be
cal
culat
ed
f
rom
the
ro
to
r
spe
ed
a
nd
the roto
r flu
x
i
n
the
stat
ion
a
r
y coor
din
at
e sy
stem
as.
1
(
1
)
S
SS
RR
S
R
R
R
mm
Td
jT
L
L
d
t
i
(6)
On
the
oth
e
r
ha
nd,
the
r
otor
flu
x
var
ia
ble
S
R
i
s
cal
culat
ed
from
the
sta
tor
vo
lt
age
,
sta
tor
curren
t
,
and str
uctu
ral
par
am
et
ers
of IM
s acco
rd
i
ng to
the
volt
age
m
od
el
[
18]
2
(
(
)
[
]
)
S
S
S
S
S
R
m
R
R
S
S
S
S
mR
L
L
L
L
R
d
t
LL
u
i
i
(7)
Suppose
that
we
re
place
the
act
ual
ro
tor
s
peed
ω
R
with
t
he
m
easur
ed
sign
al
of
the
sp
e
ed
sens
or
ω
R_m
in
(6
)
,
the esti
m
a
te
d
s
ta
tor
c
urren
t ca
n be c
orres
pondin
gly cal
culat
ed
by
_
1
ˆ
(
1
)
S
SS
RR
S
R
R
m
R
mm
Td
jT
L
L
d
t
i
(8)
Fr
om
(
6) a
nd (8), the
d
if
fer
e
nc
e b
et
wee
n ac
tual
an
d
e
stim
ated
stat
or c
urre
nts ca
n be
displ
ay
ed
b
y
_
ˆ
)
(
)
S
S
S
R
S
S
R
m
R
R
m
T
j
L
(
i
i
(9)
Evaluation Warning : The document was created with Spire.PDF for Python.
IS
S
N
:
2088
-
8708
In
t J
Elec
&
C
om
p
En
g,
V
ol.
10
, No
.
3
,
J
une
2020
:
30
35
-
3046
3040
Fr
om
(9),
by
a
naly
zi
ng
the
st
at
or
c
urre
nt
s
pa
ce
vect
or
an
d
the
r
otor
flu
x
s
pace
vecto
r
i
n
the
sta
ti
on
a
ry
s
yst
e
m
[α, β]
, we ca
n r
ecei
ve
e
qu
at
io
ns
(10),
(11
),
(
12)
as
foll
ows
:
_
ˆˆ
[
(
)
(
)
]
(
)
(
)
R
S
S
S
S
R
m
R
R
R
m
T
i
i
j
i
i
j
L
(10)
2
__
2
__
ˆˆ
[
(
)
(
)
(
)
[
(
)
(
)
(
)
ˆˆ
(
)
]
(
)
(
)
(
)
]
(
)
(
)
RR
S
S
R
m
R
R
S
S
R
R
m
R
R
mm
RR
S
S
R
m
R
R
S
S
R
R
m
R
R
mm
TT
i
i
i
i
LL
TT
i
i
i
i
LL
(11)
22
_
ˆˆ
(
)
(
)
(
)
(
)
R
S
S
R
S
S
R
R
m
R
R
R
m
T
i
i
i
i
L
(
1
2)
Th
us
,
the
dif
fe
ren
ce
betwee
n
the
act
ual
an
d
m
easur
ed
r
otor
sp
ee
d
can
be
determ
ined
acc
ordin
g
to
t
he
s
ta
tor
current a
nd th
e
rotor fl
ux in
t
he
stat
ion
ary
syst
e
m
[α, β]
[
16]
as b
el
ow
.
_
22
ˆˆ
(
)
[
(
)
.
(
)
.
]
)
m
R
R
m
S
S
R
S
S
R
R
R
R
L
i
i
i
i
T
(13)
Wh
e
re,
the
ac
tual
sta
tor
currents
i
Sα
,
i
Sβ
are
m
easur
ed
from
the
cur
r
ent
sens
or
s
a
nd
tra
nsfo
rm
e
d
in
t
o
the
sta
ti
on
ary
syst
e
m
[α,
β]
by
Cl
ark
’s
for
m
ula
;
the
est
i
m
at
ed
sta
tor
currents
ˆ
S
i
,
ˆ
S
i
are
der
i
ved
f
r
o
m
fo
ll
owin
g dif
fe
ren
ti
al
equati
ons
[19]:
2
2
(
)
[
]
S
m
m
R
R
S
S
R
S
R
m
R
R
RR
S
R
m
d
i
L
L
R
L
u
R
i
i
i
L
i
d
t
L
L
L
L
L
(14)
2
2
(
)
[
]
S
m
m
R
R
S
S
R
S
m
R
R
R
RR
S
R
m
di
L
L
R
L
u
R
i
i
L
i
i
d
t
L
L
L
L
L
(15)
2
1
(
)
[
]
R
S
m
S
RR
S
R
S
R
R
R
S
S
m
m
m
S
R
d
i
L
L
R
RL
u
i
i
i
i
d
t
L
L
L
L
L
L
L
(16)
2
1
(
)
[
]
R
S
m
S
RR
R
S
S
R
R
R
S
S
m
m
m
S
R
di
L
L
R
LR
u
i
i
i
i
d
t
L
L
L
L
L
L
L
(17)
More
ov
e
r,
i
n
(
13)
th
e
r
otor
ti
m
e
coef
fici
ent
(RTC),
T
R
=
R
R
/L
R
,
is
an
im
po
r
ta
nt
par
am
et
er
that
aff
ect
s
the
accu
racy
of
the
diag
nosis
al
gorithm
.
Due
to
the
in
flue
nce
of
the
en
vi
ro
nm
ent
tem
p
eratur
e
in
op
e
r
at
ion,
T
R
can
be
ch
ang
e
d,
bu
t
we
cannot
direct
ly
m
easur
e
th
is
qu
antit
y
[2
0
-
25]
.
Thus,
we
nee
d
to
est
i
m
at
e
the
ap
pro
pr
ia
te
ro
t
or
ti
m
e
coef
fici
ent b
ecau
s
e
it
cou
l
d
ca
use
an
undesi
red
error
i
n
the
dia
gnos
is
m
et
ho
d i
f
it
is
far
from
the
act
ual
value
.
F
r
om
the
m
at
he
m
at
ic
al
m
od
el
of
IMs,
we
ha
ve
a
relat
ion
s
hip
of
t
he
r
ot
or
tim
e
coef
fici
ent, m
agn
et
iz
in
g
c
urre
nt,
m
agn
et
iz
in
g
in
duct
ance
, a
nd f
l
ux co
m
ponen
t c
urre
nt as
descr
i
bed
by
m
R
m
S
x
di
T
i
i
dt
(18)
R
m
m
Li
(19)
Fr
om
two ab
ov
e eq
uations,
w
e estim
at
e the roto
r
ti
m
e coef
f
ic
ie
nt f
r
o
m
1
ˆ
()
R
m
S
x
R
R
T
L
i
d
t
(20)
Evaluation Warning : The document was created with Spire.PDF for Python.
In
t J
Elec
&
C
om
p
En
g
IS
S
N: 20
88
-
8708
A rob
us
t
diagnosis
meth
od fo
r sp
ee
d
s
ens
or
fau
lt
base
d
on
stator c
ur
re
nts
in
the…
(
Cu
ong Di
nh Tr
an
)
3041
By
su
bs
ti
tuti
ng
(
20)
i
nto
(
13),
the
di
ff
e
ren
c
e
betwee
n
t
he
act
ual
an
d
m
e
asur
e
d
ro
t
or
s
peed
can
be
rewrit
te
n
b
y
_
22
ˆˆ
F
(
)
[
(
)
.
(
)
.
]
ˆ
)
m
i
n
d
i
c
a
t
i
o
n
R
R
m
S
S
R
S
S
R
R
R
R
L
i
i
i
i
T
(21)
It
is
the
in
dicat
ion
functi
on
(
F
indication
)
to
iden
ti
fy
the
sp
ee
d
s
ens
or
fail
ur
e
a
ccordin
g
to
t
he
diff
e
re
nce
bet
ween
m
easur
ed
an
d
est
i
m
at
ed
sta
tor
c
urren
ts,
the
ro
t
or
flu
x,
an
d
the
est
im
a
te
d
r
otor
ti
m
e
coef
f
ic
ie
nt.
I
n
the
norm
al
op
e
rati
on
c
ondi
ti
on
,
t
he
est
im
at
ed
sta
to
r
c
urr
ents
s
houl
d
be
the
sam
e
with
the
m
easur
ed
s
ign
al
s
from
cu
rr
e
nt
sens
or
s
,
so
t
he
value
of
the
s
peed
er
r
or
in
dicat
ion
functi
on
in
(2
1)
ap
proxim
at
ely
equ
al
s
to
zero
.
I
f
occ
urri
ng
any
fail
ur
e
of
t
he
sp
ee
d
se
ns
or,
the
re
is
a
sign
i
ficant
de
via
ti
on
bet
ween
t
he
est
i
m
at
ed
and
m
easur
e
d
current
values
t
hat
le
ads
to
a
n
i
ncr
e
ase
in
the
F
indicat
ion
value
an
d
that
su
c
h
fail
ur
e
ca
n
be
r
ec
ognized
.
How
ever,
this
F
indication
f
un
ct
io
n
co
uld
m
ake
a
m
ist
ake
in
i
den
ti
fyi
ng
t
he
s
pee
d
se
ns
or
fa
ult
if
t
he
cu
rr
e
nt
se
nsors
fail
them
sel
ves.
To
pr
e
ven
t
t
his
m
ist
ake,
an
a
dd
it
io
nal
co
nd
it
ion
cal
le
d
F
re
f
c
om
par
ing
be
tween
t
he
m
e
asure
d
sp
ee
d
ω
R_m
an
d
the
r
e
fer
e
nce
sp
ee
d ω
R_ref
is
add
e
d
t
o
the
diagnosis al
gorithm
as sh
ow
n b
y
__
F
r
e
f
R
m
R
r
e
f
(22)
Finall
y, the p
r
opos
e
d
dia
gnosi
s algo
rithm
i
nclu
des
tw
o
pa
rts, as
desc
ribe
d
in (2
1)
,
(22
).
The
fir
st o
ne
com
par
ing
between
the
m
ea
su
re
d
an
d
est
im
at
ed
sta
tor
currents
to
dete
ct
the
po
ssible
fau
lt
in
the
sp
ee
d
sens
or
,
a
nd
the
seco
nd
co
m
par
ing
betw
een
the
m
eas
ur
e
d
an
d
the
ref
ere
nce
r
oto
r
s
peeds
to
pr
e
ve
nt
the con
fu
si
on
by the
possible
f
ai
lure
of c
urr
ent se
ns
ors t
he
m
s
e
lves. I
t i
s c
al
le
d
the F
speed
functi
on
s
how
n b
y
22
__
ˆˆ
F
[
(
)
.
(
)
.
]
_
1
ˆ
)
F
F
_
2
m
i
n
d
i
c
a
t
i
o
n
S
S
R
S
S
R
R
R
R
s
p
e
e
d
r
e
f
R
m
R
r
e
f
L
i
i
i
i
t
h
r
e
s
h
o
l
d
T
t
h
r
e
s
h
o
l
d
(23)
The
SS
F
dia
gnos
is
al
gorithm
can
surel
y
iden
ti
fy
any
sp
eed
sens
or
fa
ult
accor
ding
to
the
F
speed
wh
e
r
e
the
values
of
“
thres
ho
l
d_1”
a
nd
“t
hr
es
hold
_2”
are
pre
def
i
ne
d
de
pe
nd
i
ng
on
th
e
accu
rac
y
le
vel
of
the
a
ct
ually
instal
le
d
se
ns
ors.
In
this
pa
pe
r,
we
pro
pose
t
he
set
ti
ng
valu
es
ab
out
10%
of
the
rated
sta
tor
c
urre
nt,
a
nd
10%
of
the
re
fer
e
nc
e
ro
to
r
sp
ee
d
f
or
th
res
ho
l
d_1
and
th
res
ho
l
d_2,
res
pecti
vely
.
Thu
s
,
the
pro
po
s
ed
SS
F
dia
gnos
i
s
al
gorithm
can
be descri
bed as
the
flo
wch
a
rt
i
n
Fi
gure
3
.
Figure
3. Flo
w
char
t
of the
sp
e
ed
se
nsor
f
ai
lu
re
diag
no
sis
algorit
hm
Fu
rt
her
m
or
e,
t
his
pro
posed
SSF
dia
gnos
is
m
et
ho
d
can
e
ff
ect
ively
ap
ply
fo
r
IMs
’
ap
plica
ti
on
s
at
the
low
-
sp
ee
d
range
w
her
e
t
he
sp
ee
d
sen
s
or
sig
nal
oft
en
su
ffers
f
ro
m
no
ise
wh
il
e
ot
her
sim
i
la
r
al
go
rithm
s
would
be
com
plica
te
d
to
ha
ndle
in
that
su
ch
conditi
on
.
T
ha
t
adv
anta
ge
co
m
es
fr
om
the
i
dea
of
ind
e
pe
ndently
us
in
g
t
he
sta
tor
cu
rr
e
nt
m
od
el
to
detect
t
he
s
pee
d
se
nsor
fa
ult
beca
u
se
the
c
urre
nt
sens
or
sig
nal
is
le
ss
aff
ect
ed
by
the
no
ise
i
n
va
rio
us
s
peed
zo
nes
.
It
m
eans
that
any
fail
ure
co
nd
it
io
n
of
the
sp
ee
d
sen
sor
c
an
be
Evaluation Warning : The document was created with Spire.PDF for Python.
IS
S
N
:
2088
-
8708
In
t J
Elec
&
C
om
p
En
g,
V
ol.
10
, No
.
3
,
J
une
2020
:
30
35
-
3046
3042
cl
early
disti
nguish
e
d
acc
ording
t
o
the
diff
e
ren
ce
betwee
n
the
real
an
d
es
tim
a
te
d
sta
tor
currents
e
ve
n
thou
gh
t
he
no
ise
af
fe
ct
ing
the
sp
ee
d
sens
or
si
gn
a
l.
Thu
s
,
the
pro
posed
a
ppr
oa
ch
can
be
e
f
fecti
vely
app
li
ed
to
diag
nose the
f
a
il
ur
e
of
sp
ee
d
s
ens
or
s i
n vari
ous s
pee
d ran
ge
s.
3.
SIMULATI
O
N
RESU
LT
S
In
t
his
pa
rt,
t
he
pro
posed
SS
F
dia
gnos
is
m
et
hod
has
been
ver
ifie
d
th
r
ough
var
io
us
oper
at
ion
s
pee
ds
of
the
IM
i
n
th
e
case
of
total
ly
disco
nnect
in
g
t
he
s
pee
d
se
ns
or
sig
nal.
T
he
sim
ulatio
ns
hav
e
im
ple
m
e
nted
i
n
the MAT
LAB/
SI
MUL
I
NK en
vir
on
m
ent w
it
h t
he param
et
ers
of the
I
M s
ho
wn in T
able
1
.
Table
1.
T
he
param
et
ers
of
t
he
induc
ti
on m
oto
r
Descripti
o
n
Sy
m
b
o
l
Un
it
Valu
e
Rated
Power
P
n
kW
2
.2
Rated
Voltag
e
U
n
V
400
Rated
T
o
rqu
e
T
n
Nm
1
4
.8
Rated
sp
eed
ω
n
rp
m
1420
Rated
stato
r
cu
r
rent
I
S
A
4
.85
Stato
r
resistan
ce
R
S
Ω
3
.17
9
Ro
to
r
resistan
ce
R
R
Ω
2
.11
8
Mutu
al ind
u
ctan
ce
L
m
H
0
.1
92
Mo
m
en
t of
inertia
J
Kg
m
^2
0
.04
7
Po
le pair nu
m
b
er
p
-
2
Stato
r
in
d
u
ctan
ce
L
S
H
0
.20
9
Ro
to
r
in
d
u
ctan
ce
L
R
H
0
.20
9
Rated
Ro
to
r
f
lu
x
Ψ
Sn
Wb
0
.75
7
3.1.
N
orm
al
-
s
peed r
ange
In
t
he
first
cas
e,
the
IM
is
operated
at
t
he
norm
al
sp
eed
ra
ng
e
with
a
c
onsta
n
t
loa
d
to
rque
of
5
N.
m
from
t=
0.
5s
ec
.
As
sho
wn
in
Fi
gure
4
belo
w,
the
ref
e
re
nce
spe
ed
ste
ps
u
p
50%
of
t
he
rate
d
sp
ee
d
(
710
r
pm
)
at
t=
0.
5
sec
a
nd
keeps
to
t=
3
se
c,
the
n
re
duces
accor
ding
t
o
a
ram
p
li
ne
fro
m
3.
0
sec
to
3.5
sec
bac
k
to
25%
of
the
rat
e
d
valu
e
(
355rpm
).
S
uppose
at
t=
2.0
sec,
t
her
e
is
a
fa
ult
of
the
s
peed
se
nsor
occ
urrin
g,
an
d
the
value
of
t
he
s
pee
d
si
gn
a
l
is
dow
n
to
z
ero
im
m
ediat
e
ly
.
Fig
ur
e
5
de
picte
d
t
he
i
nst
abili
ty
and
i
nc
orrect
op
e
rati
on
of
t
he
I
M sy
ste
m
w
it
ho
ut t
he
F
TC
m
od
ule
.
Figu
re
4.
R
otor
sp
ee
d
a
nd th
re
e
-
phase c
urre
nt
of I
M
in the heal
thy s
ens
or
s
-
co
ndit
ion at
no
rm
al
-
s
peed
Figure
5. The
r
otor
s
peed
of
I
M i
n
the
SS
F
-
c
onditi
on
at
n
orm
al
-
sp
eed
without F
TC
m
et
ho
d
Figure
6
s
how
s
the
sp
eed
de
viati
on
bet
wee
n
act
ual
an
d
est
i
m
at
ed
sp
eed
sign
al
s
deter
m
ined
from
the
sta
tor
c
urr
ents
acco
rd
i
ng
to
the
F
indication
wh
er
e
it
ch
ang
e
s
f
ro
m
zero
c
orres
ponding
t
o
the
nor
m
al
conditi
on
t
o
a
sign
ific
a
nt
val
ue
afte
r
t=
2
se
c
du
e
to
lo
sin
g
the
sen
sor
sig
nal
.
Ac
co
rd
i
ng
to
(
23),
beca
use
this
sp
ee
d
de
viati
on
is
higher
t
ha
n
T
hr
es
hold
1
-
val
ue,
it
m
akes
the
in
dicat
ion
flag
of
th
e
sp
ee
d
se
nsor
fail
ur
e
go
i
ng
up
im
mediat
el
y
at
that
tim
e
.
T
he
perform
ance
of
th
e
pro
po
se
d
F
T
C
m
et
ho
d
a
gainst
the
s
pee
d
-
s
en
s
or
fau
lt
has
bee
n
dem
on
strat
ed
in
Fig
ur
e
7.
He
re,
w
hen
lo
si
ng
the
feedbac
k
sp
eed
sig
nal,
the
ro
t
or
osc
il
la
te
s
in
a
sh
ort
tim
e,
and
t
hen
t
he
FTC
m
od
ule
qu
ic
kly
isolat
es
the
physi
cal
sp
eed
in
put
and
rep
la
ces
it
by
the
est
im
a
te
d
ro
t
or
s
pee
d
from
the
MR
AS
est
i
m
at
or
to
m
a
intai
n
the
sta
ble
ope
rati
on
of
the
IM
dr
i
ve.
The
res
ult
sho
ws
t
hat
FTC
i
m
ple
m
ents
the
co
ntr
ol
act
ion
ver
y
well
beca
us
e
t
he
act
ual
con
t
ro
ll
ed
spe
e
d
keeps
cl
os
el
y t
rack
i
ng the
r
e
fer
e
nce
sp
ee
d
.
Evaluation Warning : The document was created with Spire.PDF for Python.
In
t J
Elec
&
C
om
p
En
g
IS
S
N: 20
88
-
8708
A rob
us
t
diagnosis
meth
od fo
r sp
ee
d
s
ens
or
fau
lt
base
d
on
stator c
ur
re
nts
in
the…
(
Cu
ong Di
nh Tr
an
)
3043
Figure
6.
The
roto
r
s
peed di
fference
of
diag
no
sis
al
gorithm
an
d
t
he
in
dicat
io
n
fl
ag of
SSF
at
norm
al
-
sp
eed
with
the
FTC
m
et
ho
d
Figure
7.
R
otor
sp
ee
d
a
nd th
re
e
-
phase c
urre
nt of I
M
in the
SSF
-
c
onditi
on
at
norm
al
-
sp
ee
d wit
h
FTC m
et
ho
d
3.2.
Low
-
s
pee
d range
In
this
dem
on
s
trat
ion
,
we
us
e
t
he
sam
e
con
trol
syst
em
and
SSF
dia
gnos
is
al
go
rithm
as
t
he
previ
o
us
case
.
The
ref
e
r
ence
sp
ee
d
ste
ps
up
10%
of
t
he
rated
s
pee
d
(142
rp
m
)
at
t
=
0.5
sec
an
d
ke
eps
to
t
=
3
sec
,
and
then
reduces
a
ccordin
g
to
a
ram
p
li
ne
fr
om
3.
0
sec
t
o
3.5
sec
back
t
o
5%
of
the
r
at
ed
sp
ee
d
(71
rp
m
)
.
Wh
e
n
occurri
ng
a
s
peed
se
nsor
fau
lt
at
t
=
2
sec,
the
se
nsor
sign
al
go
es
do
wn
t
o
zer
o,
a
nd
it
m
akes
the
sp
ee
d
dev
ia
ti
on
cal
c
ulate
d
from
the
propose
d
SSF
diag
nosis
al
gorithm
cha
ngin
g
ra
pid
ly
at
that
tim
e.
In
sta
ntane
ously
,
the
FTC
f
un
ct
ion
is
im
ple
m
ented
agai
ns
t
that
su
c
h
fail
ur
e
,
an
d
the
i
ndic
at
ion
flag
s
urge
to
a
high
le
vel,
as
sh
ow
n
in
Fi
gure
8
.
The
n,
t
he
sp
ee
d
est
i
m
ato
r
will
prov
i
de
an
al
te
rn
at
ive
sp
ee
d
sig
nal
f
or
the
RFOC
lo
op
of
the
IM
dr
i
ve.
As
a
res
ult,
t
he
act
ual
s
pee
d
trajecto
ry
sti
ll
perfect
ly
kee
ps
the
sp
ee
d
refe
ren
c
e
and
m
ai
ntains
a
s
m
oo
th
oper
at
ion
at
the
lo
w
-
s
pee
d
ra
nge
,
eve
n
in
the
f
aulty
sensor
si
tuati
on
,
a
s
sho
wn
i
n
Figure
9
.
Figure
8.
The
roto
r
s
peed di
fference
of
diag
no
sis
al
gorithm
an
d
t
he
in
dicat
io
n
fl
ag of
SSF
at
low
-
sp
ee
d wit
h
the
F
TC m
et
ho
d
Figure
9.
R
otor
sp
ee
d
a
nd th
re
e
-
phase c
urre
nt of I
M
in the
SSF
-
c
onditi
on
at l
ow
-
s
peed with
FTC m
et
ho
d
Fu
rt
her
m
or
e,
we
ha
ve
al
so
im
ple
m
ented
the
tradit
ion
al
S
SF
diag
nosis
m
et
ho
ds,
wh
ic
h
is
based
on
the
com
par
iso
n
betwee
n
the
m
easur
ed
an
d
the
est
im
a
te
d
sign
al
s
[10
-
12]
.
Thes
e
sim
ulatio
ns
we
re
perf
or
m
ed
by
the
par
am
et
ers
si
m
il
ar
t
o
the
two
cas
es
above.
Due
to
the
si
m
plici
t
y
of
the
diag
nosis
al
gorithm
,
the
SS
F
-
detect
ion
ti
m
e
of
t
he
tradit
io
nal
m
e
thods
is
sho
rter
tha
n
t
he
pro
pose
d
m
et
ho
d,
and
the
fa
ulty
sp
ee
d
sign
al
is
quic
kl
y
rep
la
ced
by
the
est
i
m
at
ed
value.
As
a
re
sul
t,
the
sp
eed
c
ha
racteri
sti
c
appl
yi
ng
the
tradit
ion
al
diag
nosis
m
eth
od
ac
hieve
s
s
ta
bili
ty
faster
than
that
of
us
i
ng
the
pro
po
s
e
d
m
et
ho
d,
as
s
how
n
in
Fig
ure
10.
Howe
ver,
in
pract
ic
e,
due
t
o
the
no
ise
af
fe
ct
ed
on
t
he
m
easur
e
d
sp
ee
d
sign
al
i
n
the
low
-
s
peed
ope
r
at
ion
,
the
tradit
io
nal
SSF
diag
nos
is
al
go
rit
hm
i
s
not
surel
y
recog
nized
th
e
fail
ur
e
quic
kly
as
presen
te
d
in
the the
or
et
ic
al
si
m
ulati
on
s
.
Be
sides,
a
dis
adv
a
ntage
of
t
he
tra
diti
on
al
FTC
m
et
ho
ds
is
ver
y
se
ns
it
ive
to
t
he
est
im
at
ed
ro
to
r
sp
ee
d,
wh
ic
h
cou
l
d
be
af
fec
te
d
by
the
qua
li
ty
of
the
sp
e
ed
est
im
a
ti
on
m
et
ho
ds
or
cu
rr
e
nt
sens
or
fa
il
ur
es.
Wh
e
n
occ
urrin
g
a
current
-
sen
so
r
fa
ult,
the
est
i
m
at
ed
sp
eed
value
determ
i
ned
by
the
est
im
at
or
is
inaccur
at
e
;
thu
s
,
it
co
uld
be
m
ist
aken
with
the
sp
ee
d
se
ns
or
fail
ure
an
d
re
place
the
healt
hy
s
pe
ed
se
nsor
si
gnal
by
a wro
ng esti
m
a
te
d
val
ue.
The
r
efore, t
his m
is
diag
nosis can
lead to
a
br
ea
kd
own of t
he
wh
ole I
M
D
syst
e
m
.
Evaluation Warning : The document was created with Spire.PDF for Python.
IS
S
N
:
2088
-
8708
In
t J
Elec
&
C
om
p
En
g,
V
ol.
10
, No
.
3
,
J
une
2020
:
30
35
-
3046
3044
To
ve
rify
the
robu
st
ness
an
d
reli
abili
ty
of
the
pro
pose
d
FTC
m
et
ho
d,
s
om
e
s
i
m
ul
at
ion
s
un
der
the
conf
us
io
na
l
sta
te
with
a
cur
re
nt
sensor
fail
ure
ha
ve
bee
n
im
p
lem
ented
by
bo
t
h
tradit
io
na
l
an
d
the
pro
pose
d
m
et
ho
d.
He
re,
the
m
oto
r
ha
s
bee
n
ope
rated
at
norm
al
and
low
-
spe
ed
ra
nges
wh
il
e
occ
urrin
g
fail
ur
e
of
the
current
se
ns
o
r
in
on
e
phase
.
Figure
11
presents
the
pe
rfor
m
ance
of
bo
t
h
tradit
io
na
l
and
pro
po
se
d
FTC
m
e
tho
ds
at
a
ref
e
ren
ce
of
50%
of
the
rate
d
s
pee
d
wh
il
e
occurri
ng
a
c
urre
nt
se
ns
or
fa
il
ur
e
in
ph
a
s
e
A
at
t
=
2.5
sec.
As
s
how
n
in
Fi
gure
11
(
b),
the
perform
ance
of
the
tr
a
diti
onal
FTC
appr
oa
ch
is
un
s
ucces
sf
ul
from
the
m
o
m
en
t
of
occ
urri
ng
t
he
cu
rr
e
nt
se
nsor
fail
ur
e
beca
us
e
the
S
SF
diagnosis
is
m
istak
en
,
and
it
s
co
ntr
oller
is
switc
he
d
to
us
i
ng
the
inaccu
rate
est
i
m
at
ed
sp
eed
value.
I
n
c
ontrast
,
the
IM
s
yst
e
m
op
e
rated
unde
r
th
e
pro
po
s
ed
FTC
m
e
tho
d
st
il
l
keep
s
sta
bly
in
the
sam
e
si
tuati
on
becaus
e
the
SSF
diag
no
sis
has
rec
ognize
d
exactl
y
the
tro
ub
le
in
t
he
current
se
nsor
t
hat
is
not
t
he
fail
ure
of
the
s
pee
d
s
ens
or.
It is prese
nted
i
n
Fi
gure
11(c
).
(a)
(b)
Figure
10. T
he
r
oto
r
s
peed
of
IM
D
i
n
the
SS
F
-
c
onditi
on w
i
th
tra
diti
on
al
F
TC m
e
tho
d
:
(a)
norm
al
-
sp
e
ed ran
ge, (
b)
low
-
s
pee
d ran
g
e
(a)
(b)
Figure
11. C
om
par
ison
betw
een tra
diti
on
al
and pr
opos
e
d FTC
m
et
ho
ds
at
n
orm
al
sp
eed
range:
(a)
Th
ree
-
phas
e stat
or
c
urre
n
t
s (
c
urren
t
fail
ure in
P
hase
A),
(b)
Pe
rfo
rm
ance of th
e
trad
it
io
nal FTC m
et
ho
d wit
h
a c
urre
nt
sen
s
or f
ai
lu
re
,
(c)
Per
form
ance of th
e
prop
ose
d
F
TC m
et
ho
d wit
h
a c
urre
nt
sen
s
or f
ai
lu
re
Evaluation Warning : The document was created with Spire.PDF for Python.