T
E
L
KO
M
NI
K
A
,
V
ol
.
14,
N
o.
3,
S
ept
em
ber
20
16,
pp.
81
5~
8
23
I
S
S
N
:
1
693
-
6
930
,
ac
c
r
edi
t
ed
A
b
y
D
IK
T
I,
D
e
c
r
e
e
N
o
:
58/
D
I
K
T
I
/
K
ep/
2013
D
O
I
:
10.
12928/
T
E
LK
O
M
N
I
K
A
.
v
1
4
i
3
.
3545
81
5
R
ec
ei
v
ed
M
ar
c
h 8
,
2
01
6
;
R
ev
i
s
ed
May
1
2
,
20
1
6
;
A
c
c
e
pt
ed
J
u
ne
2
,
2
01
6
Tr
a
ns
for
m
e
r
Fa
ul
t D
i
a
gn
os
i
s
M
e
tho
d B
a
s
e
d
on
D
y
n
a
mi
c
W
e
i
g
h
te
d C
om
bi
na
ti
o
n M
od
e
l
H
o
n
g
l
i
Y
u
n
1
,
2
,
R
u
n
L
i
u
1
,
2
,
L
i
n
j
i
a
n
S
h
a
n
g
g
u
a
n
*
1
,
2
1
N
or
t
h C
hi
n
a U
ni
v
er
s
i
t
y
of
W
a
t
er
R
es
o
ur
c
e
s
a
nd E
l
e
ct
r
i
c
P
o
w
er
,
Z
hengz
hou,
C
hi
na
2
C
ol
l
a
bor
at
i
v
e I
nn
ov
at
i
on
C
ent
er
of
W
a
t
er
R
e
s
our
c
e
s
E
f
f
i
c
i
en
t
U
t
i
l
i
z
at
i
on
and G
u
ar
ant
ee E
n
gi
neer
i
ng
,
Z
hengz
ho
u,
C
hi
na
*
C
or
r
es
po
ndi
ng a
ut
hor
,
e
-
ma
i
l
:
y
hl
@
nc
w
u.
edu.
c
n
A
b
st
r
act
T
he pap
er
t
r
i
ed t
o i
nt
egr
at
e t
he D
G
A
dat
a w
i
t
h t
he ga
s
p
r
oduc
t
i
on r
a
t
e,
w
hi
c
h ar
e t
he
m
aj
or
i
nde
x
es
of
t
r
a
ns
f
or
m
er
f
aul
t
d
i
agno
s
i
s
.
D
uv
a
l
’
s
t
r
i
angl
e m
et
hod,
B
P
neur
al
n
et
w
or
k
an
d I
E
C
t
hr
ee
-
r
a
t
io
m
et
hod w
er
e w
ei
ght
e
d.
F
i
r
s
t
l
y
,
t
he
pa
per
r
egar
d
ed
t
he
ga
s
pr
odu
c
t
i
o
n r
a
t
e
as
t
he
i
nd
epe
ndent
v
ar
i
ab
l
e
s
,
f
i
t
t
ed
t
he
c
u
bi
c
c
ur
v
e
s
of
t
h
e
g
as
pr
o
duc
t
i
on
r
at
e
an
d
v
ar
i
an
c
e
of
ea
c
h
di
agno
s
i
s
m
et
hod
,
a
nd
t
hen
def
i
ned
t
he w
ei
g
ht
s
o
f
eac
h al
g
or
i
t
h
m
t
hr
ough t
h
e dat
a pr
oc
es
s
i
ng m
et
hod of
uneq
ual
pr
ec
i
s
i
on.
A
t
l
a
s
t
,
t
h
e
dy
nam
i
c
w
ei
g
ht
ed
c
om
bi
nat
i
o
n
di
ag
nos
i
s
m
od
el
w
as
es
t
abl
i
s
hed
.
T
hat
i
s
,
t
he
w
ei
ght
i
s
di
f
f
er
ent
a
s
t
he
gas
pr
odu
c
t
i
o
n
r
at
e
c
hang
es
a
l
t
ho
ugh
t
he
m
et
ho
d
i
s
i
dent
i
c
a
l
.
T
he
r
es
u
l
t
s
of
d
i
ag
nos
i
s
e
x
am
pl
es
s
h
ow
t
hat
t
h
e
ac
c
ur
ac
y
r
a
t
e of
t
he
w
ei
g
ht
ed
c
om
bi
n
at
i
o
n m
ode
l
i
s
hi
gher
t
han a
ny
s
i
ngl
e
al
gor
i
t
hm
,
an
d i
t
h
as
c
er
t
ai
n
st
a
b
ilit
y
a
s
w
e
l
l.
K
eyw
o
r
d
s:
t
r
ans
f
or
m
er
f
a
ul
t
,
w
e
i
ght
ed
c
o
m
bi
nat
i
o
n
m
ode
l
,
du
v
al
’
s
t
r
i
an
gl
e
,
bp
n
eur
al
n
et
w
or
k
,
t
hr
ee
-
r
at
i
o
m
et
hod
C
o
p
y
r
i
g
h
t
©
20
16 U
n
i
ver
si
t
a
s A
h
mad
D
ah
l
an
.
A
l
l
r
i
g
h
t
s r
eser
ved
.
1
.
I
n
tr
o
d
u
c
ti
o
n
P
o
w
er
t
r
a
ns
f
or
m
er
i
s
t
he
k
e
y
equ
i
pm
ent
of
t
he
po
w
er
s
upp
l
y
s
y
s
t
em
,
w
hi
c
h
has
i
m
por
t
ant
pr
ac
t
i
c
a
l
v
a
l
ue f
or
t
he r
e
al
-
t
i
m
e
m
oni
t
or
i
n
g an
d f
aul
t
di
agn
os
i
s
of
t
he s
t
at
e of
t
he
t
r
ans
f
or
m
er
[
1]
.
A
t
pr
es
ent
,
t
he
di
s
s
ol
v
e
g
as
anal
y
s
i
s
(
D
G
A
)
m
et
hod
of
t
r
ans
f
or
m
er
oi
l
i
s
w
i
del
y
us
ed
t
o es
t
i
m
at
e t
he
i
nt
er
n
al
f
aul
t
pr
op
er
t
i
es
of
t
he
t
r
ans
f
or
m
er
[
2]
,
and
di
ag
nos
i
s
m
et
hod l
i
k
e
t
he D
uv
al
’
s
t
r
i
ang
l
e
m
et
hod
,
t
he
t
hr
e
e
-
r
at
i
o m
et
hod,
I
E
C
t
hr
e
e r
at
i
o
m
et
hod,
e
t
c
.
a
r
e c
ons
i
d
er
ed
c
l
as
s
i
c
al
.
W
i
t
h t
he de
v
e
l
o
p
m
ent
of
ar
t
i
f
i
c
i
al
i
nt
el
l
i
genc
e t
ec
h
nol
og
y
,
neur
a
l
net
w
o
r
k
,
f
uz
z
y t
h
e
o
r
y,
ex
per
t
s
y
s
t
em
,
gene
t
i
c
a
l
gor
i
t
h
m
and
ot
her
m
et
ho
ds
ar
e ap
pl
i
e
d t
o
t
he f
au
l
t
d
i
ag
nos
i
s
of
t
r
ans
f
or
m
er
s
[
3
-
6]
.
T
her
e
i
s
a
c
om
pl
ex
r
el
at
i
o
ns
hi
p
bet
w
een
t
he
app
ear
anc
e
of
t
he
t
r
a
ns
f
or
m
er
f
aul
t
s
an
d
t
he f
aul
t
m
ec
hani
s
m
,
as
a r
es
ul
t
a s
i
n
gl
e
di
ag
nos
t
i
c
m
et
hod us
ua
l
l
y
h
av
e i
t
s
l
i
m
i
t
at
i
ons
and
t
her
e
w
i
l
l
be
a
di
ag
nos
t
i
c
b
l
i
nd
ar
ea
t
o
pr
od
uc
e
f
al
s
e
p
os
i
t
i
v
es
or
f
al
s
e
n
ega
t
i
v
es
,
gener
al
l
y
,
t
he
di
a
gnos
i
s
ac
c
ur
ac
y
r
at
e
i
s
bet
w
e
en
70%
an
d 8
0%
.
I
n t
r
ans
f
or
m
er
f
aul
t
d
i
a
gnos
i
s
,
m
an
y
s
c
ho
l
ar
s
adop
t
a c
om
bi
nat
i
o
n f
or
m
w
i
t
h a v
ar
i
e
t
y
of
c
hec
k
i
ng m
et
hods
[7
-
12]
,
w
h
i
c
h i
nc
r
eas
es
t
he
ac
c
ur
ac
y
r
at
e
of
f
aul
t
di
ag
nos
i
s
t
o
80
%
-
9
0%
.
T
he
d
i
agnos
i
s
c
om
bi
nat
i
on
m
odel
i
s
es
t
a
bl
i
s
he
d
f
or
c
ons
t
ant
,
t
hat
i
s
,
t
he m
odel
p
ar
am
et
er
s
ar
e i
nv
ar
i
ab
l
e i
n a
n
y
d
et
ec
t
i
on.
T
hi
s
pa
per
i
nt
r
o
duc
e
s
t
he
i
de
as
of
t
he
w
ei
g
ht
e
d c
om
bi
nat
i
o
n,
a
nd
em
phas
i
z
e
s
t
he d
y
n
am
i
c
of
t
he w
ei
g
ht
s
.
T
he
paper
ado
pt
s
t
he
m
et
ho
d
of
w
ei
ght
ed
c
om
bi
nat
i
on
d
i
ag
nos
i
s
.
F
i
r
s
t
l
y
,
D
uv
al
’
s
t
r
i
ang
l
e
m
et
hod,
B
P
ne
ur
al
net
w
or
k
and I
E
C
t
hr
e
e
-
r
at
i
o m
et
hod
ar
e
us
ed t
o
di
agn
os
e t
h
e
tr
a
ns
f
or
m
er
s
t
at
e.
T
hen
t
he
r
es
ul
t
s
of
t
he
t
hr
ee
m
et
hods
ar
e
w
e
i
gh
t
ed
a
nd
c
om
bi
ned
t
o
ge
t
t
he
f
i
nal
d
i
ag
nos
i
s
c
onc
l
us
i
o
n.
T
he
k
e
y
of
t
he r
es
ear
c
h i
s
t
o f
i
nd t
h
e
w
e
i
ght
s
.
H
er
e,
a
not
h
er
i
n
dex
of
t
r
ans
f
or
m
er
f
aul
t
d
i
a
gnos
i
s
-
t
he t
o
t
al
h
y
dr
oc
ar
b
on
gas
pr
oduc
t
i
on
r
at
e
[
13
]
as
t
h
e i
n
de
pen
dent
v
ar
i
ab
l
es
i
s
i
nt
r
od
uc
ed.
S
el
ec
t
i
ng t
h
e s
ui
t
a
bl
e s
am
pl
i
ng poi
nt
s
,
f
i
t
t
i
ng o
ut
t
he c
ub
i
c
c
ur
v
es
of
t
he
t
ot
a
l
h
y
d
r
oc
ar
bo
n gas
pr
o
duc
t
i
o
n r
at
e
and
v
ar
i
anc
e
of
di
agn
os
i
s
m
et
hod,
t
he
n
us
i
ng t
he d
at
a
pr
oc
es
s
i
ng
m
et
hod
of
une
qua
l
pr
ec
i
s
io
n
[
14,
1
5]
t
o
w
or
k
out
t
he
w
e
i
gh
t
s
of
eac
h
al
gor
i
t
hm
f
o
r
di
f
f
er
ent
f
aul
t
t
y
pes
u
nder
di
f
f
er
ent
gas
pr
od
uc
t
i
o
n
r
at
es
.
T
he
d
y
n
am
i
c
m
odel
has
c
er
t
a
i
n
s
t
abi
l
i
t
y
,
a
nd
t
h
e
ac
c
ur
ac
y
r
at
e
of
eac
h
t
y
p
e
of
t
r
ans
f
or
m
er
f
aul
t
d
i
ag
nos
i
s
i
s
i
nc
r
eas
ed
t
o
m
or
e
t
han 90%
a
s
w
e
ll.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SSN
:
1
6
9
3
-
6
930
T
E
L
KO
M
NI
K
A
V
o
l.
14
,
N
o
.
3,
S
ept
em
ber
2016
:
8
1
5
–
82
3
816
2
.
W
e
i
g
h
te
d
C
o
m
b
i
n
a
ti
o
n
D
i
a
g
n
o
s
i
s
M
o
d
e
l
S
et
e
ac
h
di
agn
os
t
i
c
a
l
gor
i
t
hm
w
i
t
h
)
...
1
(
n
i
A
i
=
,
ac
c
or
di
ng
t
o
t
he
er
r
or
t
he
or
y
of
t
he
uneq
ua
l
m
eas
ur
em
ent
ac
c
ur
ac
y
and
d
at
a
pr
oc
es
s
i
n
g
m
et
hod;
gi
v
e
d
i
f
f
er
ent
di
ag
nos
i
s
m
et
hods
t
he di
f
f
er
ent
w
ei
g
ht
s
i
p
on t
he bas
i
s
of
i
t
s
r
el
i
abi
l
i
t
y
,
t
h
e hi
gher
r
el
i
ab
i
l
i
t
y
t
he gr
eat
er
w
ei
g
ht
i
n t
h
e
c
o
m
pr
ehens
i
v
e
d
i
ag
nos
i
s
.
A
c
c
or
di
ng
t
o
t
h
e
f
aul
t
t
y
p
es
of
t
he
t
r
ans
f
or
m
er
,
t
he
m
et
hod
of
det
er
m
i
ni
ng
t
he
w
e
i
gh
t
of
t
he n k
i
nds
of
d
i
ag
nos
i
s
a
l
g
or
i
t
hm
s
i
s
:
2
2
2
2
1
2
1
1
.......
:
1
:
1
...
:
:
i
i
p
p
p
σ
σ
σ
=
;
∑
=
=
n
i
i
p
1
1
(
1)
W
h
er
e:
2
i
σ
i
s
t
he
v
ar
i
anc
e of
d
i
agnos
i
s
al
gor
i
t
hm
s
i
A
.
F
or
m
ul
t
i
pl
e
s
am
pl
es
m
eas
ur
em
ent
dat
a,
us
i
ng
B
es
s
e
l
f
or
m
ul
a
t
he
c
a
l
c
ul
at
ed
v
a
r
i
anc
e
i
s
m
or
e ac
c
ur
at
e,
t
hat
i
s
:
1
2
)
(
1
2
−
−
=
=
σ
∑
N
y
ij
y
N
j
i
(
2)
W
h
er
e:
ij
y
i
s
t
he
j
t
h
di
a
gnos
t
i
c
v
al
u
e
of
t
he
i
t
h d
i
a
gnos
i
s
al
gor
i
t
hm
i
A
,
y
i
s
t
h
e t
heor
e
t
i
c
al
t
r
ue
v
a
l
ue,
N
i
s
t
he n
um
ber
of
s
am
pl
i
ng po
i
nt
s
.
I
n v
i
e
w
of
t
he s
i
x
m
ai
n f
aul
t
t
y
p
es
of
t
he t
r
ans
f
or
m
er
,
t
he c
l
as
s
i
f
i
c
at
i
on
an
al
y
s
i
s
i
s
s
ho
w
n
i
n
T
abl
e 1.
T
abl
e 1
.
F
a
u
lt
T
y
pes
o
f
P
o
w
er
T
r
ans
f
or
m
e
r
No
F
aul
t
t
y
pes
ab.
Y1
par
t
i
al
di
s
c
har
ge
PD
Y2
l
ow
ener
gy
di
s
c
har
ge
D1
Y3
hi
gh ener
gy
di
s
c
har
ge
D2
Y4
heat
f
aul
t
C
t
°
<
30
0
T1
Y5
heat
f
aul
t
C
t
C
°
<
<
°
70
0
30
0
T2
Y6
heat
f
aul
t
C
t
°
>
700
T3
F
or
one
t
r
ans
f
or
m
er
f
aul
t
,
t
he d
i
ag
nos
t
i
c
r
es
u
l
t
s
of
t
he
i
A
a
lg
o
r
it
h
m
is
:
[
]
T
im
i
i
i
s
s
s
......
2
1
=
s
(
3)
W
h
er
e:
)
6
,.....
2
,
1
(
=
m
s
im
i
s
t
he
j
ud
gm
ent
of
w
het
h
er
t
he
t
r
ans
f
or
m
er
f
a
ul
t
t
y
p
e
i
s
Y
m
b
y
al
gor
i
t
hm
i
A
,
i
f
t
he
di
a
gnos
i
s
i
s
Y
m
,
s
et
1
=
im
s
,
i
f
not
,
s
et
0
=
im
s
.
T
hen
t
he
di
a
g
nos
i
s
c
onc
l
us
i
on
m
at
r
i
x
o
f
t
he n
al
gor
i
t
hm
i
s
ex
pr
es
s
ed as
:
n
m
nm
m
m
n
n
s
s
s
s
s
s
s
s
s
×
=
2
1
2
22
12
1
21
11
...
...
...
...
...
...
S
(
4)
F
or
t
he
Y
m
t
y
p
e,
t
he
w
ei
ght
s
of
eac
h di
agn
os
t
i
c
a
l
gor
i
t
hm
ar
e:
[
]
T
nm
m
m
m
p
p
p
...
2
1
=
p
(
5)
T
he w
e
i
g
ht
v
al
ue
of
t
he d
i
a
gnos
i
s
r
es
u
l
t
f
or
t
he m
f
aul
t
t
y
p
es
i
s
ex
pr
es
s
ed
as
:
Evaluation Warning : The document was created with Spire.PDF for Python.
T
E
L
KO
M
NI
K
A
I
S
S
N
:
1
693
-
6
930
T
r
ans
f
or
mer
F
a
ul
t
D
i
a
gnos
i
s
Met
ho
d B
as
ed on
D
y
n
am
i
c
W
ei
ght
ed
C
o
mbi
nat
i
on
…
(
H
o
n
g
li Y
u
n
)
817
m
n
nm
n
n
m
m
p
p
p
p
p
p
p
p
p
×
=
...
...
...
....
...
...
...
2
1
2
22
21
1
12
11
P
(
6)
I
t
c
an be s
een t
hat
,
ev
en t
h
e s
am
e
k
i
nd of
al
gor
i
t
hm
i
A
,
t
he di
agn
os
t
i
c
v
al
u
e of
di
f
f
e
r
ent
t
y
p
es
of
f
aul
t
s
w
i
l
l
be di
f
f
er
ent
.
T
he di
ag
nos
t
i
c
r
es
ul
t
s
of
t
he w
e
i
g
ht
e
d c
om
bi
nat
i
on d
i
a
gnos
t
i
c
m
odel
c
an be
get
t
i
n
g f
r
om
t
he f
or
m
ul
a (
4)
and (
6)
.
m
i
i
m
z
×
×
×
=
P
S
S
(
7)
W
h
er
e t
he m
ai
n di
ago
na
l
e
l
em
ent
s
of
z
S
ar
e:
im
N
i
im
z
mm
s
p
S
∑
=
=
1
(
8)
T
hat
i
s
t
he w
ei
g
ht
e
d s
um
v
al
ue of
n di
agn
os
t
i
c
al
gor
i
t
hm
on Y
m
t
y
pe
f
aul
t
.
{
}
)
6
,...
2
,
1
(
max
=
m
S
z
mm
i
s
t
he di
a
gnos
t
i
c
c
onc
l
us
i
o
ns
of
w
ei
gh
t
ed c
om
bi
nat
i
o
n
m
odel
,
an
d al
s
o t
he
f
aul
t
t
y
pe c
o
nc
l
us
i
on
w
i
t
h t
h
e hi
ghes
t
degr
ee of
c
onf
i
de
nc
e.
T
abl
e 2
.
D
i
a
gnos
i
s
R
es
ul
t
s
of
D
uv
a
l
’
s
T
r
ia
n
g
le
r
a
(
m
L/
d)
f
aul
t
t
y
pe
10
14
18
22
26
30
34
38
T
ot
al
f
al
s
e
num
ber
/
F
al
s
e
r
a
t
e %
PD
S
a
m
pl
e nu
m
ber
10
4
11
3
12
3
17
4
19
3
16
5
15
4
15
5
31
26.
96
F
al
s
e
nu
m
ber
D1
S
a
m
pl
e
nu
m
ber
10
3
14
3
13
3
21
5
18
5
18
4
16
3
19
4
30
23.
26
F
al
s
e
nu
m
ber
D2
S
a
m
pl
e
nu
m
ber
8
3
10
4
17
7
18
6
21
7
22
5
19
4
20
5
41
30.
37
F
al
s
e
nu
m
ber
T1
S
a
m
pl
e
nu
m
ber
12
1
17
2
20
3
19
2
16
3
15
4
12
3
8
3
21
17.
65
F
al
s
e
nu
m
ber
T2
S
a
m
pl
e nu
m
ber
11
2
14
2
15
3
15
3
14
2
16
3
13
3
13
4
22
19.
82
F
al
s
e
nu
m
ber
T3
S
a
m
pl
e nu
m
ber
13
4
15
4
15
5
18
5
23
5
22
4
17
4
13
3
34
25
F
al
s
e
nu
m
ber
T
abl
e 3.
D
i
ag
nos
i
s
R
es
ul
t
s
of
B
P
N
eur
a
l
N
et
w
or
k
r
a
(
m
L/
d)
f
aul
t
t
y
p
e
10
14
18
22
26
30
34
38
T
ot
al
f
al
s
e
num
ber
/
F
al
s
e
r
a
t
e %
PD
S
a
m
pl
e nu
m
ber
10
2
11
2
12
3
17
3
19
4
16
4
15
4
15
5
27
23.
48
F
al
s
e
nu
m
ber
D1
S
a
m
pl
e nu
m
ber
10
3
14
3
13
3
21
3
18
2
18
4
16
3
19
4
25
19.
38
F
al
s
e
nu
m
ber
D2
S
a
m
pl
e
nu
m
ber
8
3
10
3
17
4
18
4
21
3
22
3
19
2
20
3
25
18.
52
F
al
s
e
nu
m
ber
T1
S
a
m
pl
e nu
m
ber
12
2
17
3
20
2
19
2
16
2
15
3
12
2
8
3
19
15.
97
F
al
s
e
nu
m
ber
T2
S
a
m
pl
e nu
m
ber
11
2
14
1
15
3
15
2
14
2
16
4
13
2
13
3
19
17.
12
F
al
s
e
nu
m
ber
T3
S
a
m
pl
e nu
m
ber
13
4
15
4
15
3
18
4
23
6
22
5
17
4
13
3
33
24.
26
F
al
s
e
nu
m
ber
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SSN
:
1
6
9
3
-
6
930
T
E
L
KO
M
NI
K
A
V
o
l.
14
,
N
o
.
3,
S
ept
em
ber
2016
:
8
1
5
–
82
3
818
3
.
D
e
te
r
m
i
n
a
ti
o
n
o
f
V
ar
i
a
n
ce
I
n
t
h
e
f
au
l
t
di
a
gnos
i
s
of
t
h
e
t
r
ans
f
or
m
er
,
t
he
de
v
e
l
o
p
m
ent
t
r
end
of
t
he
t
r
ans
f
or
m
er
c
a
n
be j
ud
ged
b
y
de
t
ec
t
i
ng t
he
gr
o
w
t
h
r
at
e
of
t
he
di
s
s
ol
v
ed g
as
i
n
t
h
e
t
r
ans
f
or
m
er
oi
l
[
16
]
.
I
n
t
h
i
s
paper
,
t
he c
h
ar
ac
t
er
i
s
t
i
c
s
gas
c
ont
e
nt
a
nd t
ot
a
l
g
as
pr
oduc
t
i
on r
at
e f
r
om
t
he D
G
A
d
at
a
i
s
c
ons
i
der
e
d,
i
n or
d
er
t
o
i
m
pr
ov
e
t
he
r
el
i
ab
i
l
i
t
y
of
di
agn
o
s
i
s
.
I
n
t
hi
s
pap
er
,
t
he
t
ot
al
h
y
dr
oc
ar
bon
gas
pr
o
du
c
t
i
on
r
at
e
a
r
i
s
t
he
i
n
de
pen
dent
v
ar
i
ab
l
es
;
σ
i
2
i
s
t
he v
ar
i
anc
e of
eac
h f
aul
t
di
ag
nos
i
s
m
et
hod.
B
as
ed on a l
ar
ge num
ber
of
s
a
m
pl
es
,
t
he
c
hange
of
σ
i
2
w
it
h
a
r
i
s
i
n
v
es
t
i
gat
e
d.
T
hat
i
s
w
i
t
h
t
he
c
h
a
nge
of
a
r
,
w
h
et
h
er
t
he
r
el
i
ab
i
l
i
t
y
of
t
he
di
a
gnos
t
i
c
m
et
hods
w
i
l
l
b
e
di
f
f
er
ent
.
I
f
t
he
v
al
ues
of
i
p
c
hang
i
ng,
t
h
e
c
om
bi
nat
i
on
m
ode
l
w
i
l
l
b
e
di
f
f
er
ent
.
W
hen
a
r
i
s
s
et
t
o
a f
i
x
ed v
al
ue
0
r
,
st
a
t
i
st
i
cs f
o
r
l
a
r
g
e
s
am
pl
e of
t
r
ans
f
or
m
er
f
aul
t
t
y
p
es
i
s
adop
t
ed,
t
he r
ea
l
f
aul
t
t
y
pe
of
t
he t
r
ans
f
or
m
er
i
s
s
et
t
o Y
m
(
m
=
1…
…
6)
,
t
he
n t
he t
heor
et
i
c
al
t
r
u
e
v
a
l
ue
y
m
=
1.
I
f
t
he
m
et
hod
di
a
gnos
e
t
he
t
r
ans
f
or
m
er
f
aul
t
t
y
p
es
c
or
r
ec
t
l
y
,
t
he
de
t
ec
t
i
on
v
a
l
ue
of
i
A
is
y
ij
=y
m
=
1,
or
e
l
s
e
y
ij
=
0.
P
ut
t
he
y
ij
v
a
l
ue
i
n
t
o f
or
m
ul
a (
2)
,
w
h
en
0
r
r
a
=
,
t
he
v
ar
i
a
nc
e of
di
a
gnos
t
i
c
m
et
hods
f
or
Y
m
t
y
pe
i
s
2
im
σ
.
U
s
i
ng
s
am
pl
i
ng
m
et
ho
d,
c
al
c
ul
at
e
σ
i
2
t
he
v
ar
i
anc
e
of
i
A
w
i
t
h
di
f
f
er
ent
v
al
ue
s
of
a
r
,
an
d
f
i
t
t
he c
u
bi
c
c
ur
v
e of
2
i
a
r
σ
−
:
0
1
2
2
3
3
2
a
r
a
r
a
r
a
a
a
a
i
+
+
+
=
σ
(
9)
I
n f
aul
t
ana
l
y
s
i
s
,
bas
e
d o
n t
he
a
r
and
f
or
m
ul
a
(
9)
t
o
de
t
er
m
i
ne
σ
i
2
of
i
A
,
t
hus
t
h
e
w
ei
g
ht
ed
c
o
m
bi
nat
i
on
di
agn
os
i
s
m
odel
(
8)
c
an
be
us
ed f
or
c
om
p
r
ehens
i
v
e
di
agn
os
i
s
.
4.
S
am
p
l
e
T
r
a
in
in
g
I
n t
he
pa
per
,
7
45 c
as
es
w
i
t
h c
l
ear
f
aul
t
c
aus
e
of
t
r
ans
f
or
m
er
D
G
A
dat
a
an
d t
ot
a
l
h
y
dr
oc
ar
bon
gas
pr
o
duc
t
i
o
n r
at
e
a
r
i
s
c
ol
l
ec
t
ed
as
t
h
e t
es
t
and
t
r
ai
ni
n
g s
am
pl
e.
1.
T
he s
a
m
pl
es
w
er
e c
l
as
s
i
f
i
e
d ac
c
or
di
n
g t
o
T
abl
e 1,
w
hi
c
h i
nc
l
u
de 1
15 s
am
pl
es
of
P
D
f
aul
t
,
129
s
am
pl
es
of
D
1
f
aul
t
,
13
5
s
a
m
pl
es
of
D
2
f
aul
t
,
119
s
a
m
pl
es
of
T
1
f
aul
t
,
111
s
am
pl
es
o
f
T
2 f
aul
t
and
136 s
am
pl
es
of
T
3 f
aul
t
.
2.
F
or
eac
h f
aul
t
,
t
he s
am
pl
es
ar
e s
or
t
ed ac
c
or
di
ng t
o
t
he s
i
z
e of
t
he
a
r
,
8 poi
nt
s
w
i
t
h t
he
s
a
m
e
i
nt
er
v
al
ar
e
s
et
t
o
t
h
e
obs
er
v
at
i
on
p
oi
nt
.
S
et
0
r
i
s
an
obs
er
v
at
i
o
n
v
a
l
u
e,
∆
i
s
t
he
s
t
ep,
w
hen
∆
+
∆
−
∈
2
,
2
0
0
r
r
r
a
,
t
h
e c
or
r
es
po
nd
i
ng
s
am
pl
es
ar
e r
e
gar
d
ed
as
0
r
s
a
m
pl
e.
U
s
i
ng
D
uv
al
’
s
t
r
i
a
ngl
e
m
et
hod,
B
P
ne
ur
al
net
w
or
k
and
I
E
C
t
hr
ee
-
r
at
i
o
m
et
hod
f
or
di
ag
nos
i
s
,
t
h
e
s
t
a
t
is
t
ic
a
l d
a
t
a
is
s
h
o
w
n
in
T
abl
e 2
,
T
abl
e 3
and
T
abl
e
4
.
T
abl
e 4
.
D
i
ag
nos
i
s
r
es
u
l
t
s
o
f
I
E
C
t
hr
ee
-
r
at
i
o m
et
hod
r
a
(
m
L/
d)
f
aul
t
t
y
pe
10
14
18
22
26
30
34
38
T
ot
al
f
al
s
e
num
ber
/
F
al
s
e
r
a
t
e %
PD
S
a
m
pl
e
nu
m
ber
10
3
11
2
12
2
17
1
19
3
16
2
15
2
15
3
18
15.
65
F
al
s
e
nu
m
ber
D1
S
a
m
pl
e nu
m
ber
10
2
14
3
13
3
21
3
18
2
18
2
16
1
19
2
18
13.
95
F
al
s
e
nu
m
ber
D2
S
a
m
pl
e nu
m
ber
8
4
10
3
17
4
18
3
21
6
22
4
19
5
20
6
35
25.
93
F
al
s
e
nu
m
ber
T1
S
a
m
pl
e
nu
m
ber
12
2
17
2
20
3
19
2
16
1
15
1
12
2
8
1
14
1
1.
76
F
al
s
e
nu
m
ber
T2
S
a
m
pl
e nu
m
ber
11
2
14
2
15
3
15
2
14
1
16
2
13
2
13
2
16
14.
41
F
al
s
e
nu
m
ber
T3
S
a
m
pl
e nu
m
ber
13
4
15
3
15
4
18
3
23
5
22
3
17
3
13
2
27
19.
85
F
al
s
e
nu
m
ber
Evaluation Warning : The document was created with Spire.PDF for Python.
T
E
L
KO
M
NI
K
A
I
S
S
N
:
1
693
-
6
930
T
r
ans
f
or
mer
F
a
ul
t
D
i
a
gnos
i
s
Met
ho
d B
as
ed on
D
y
n
am
i
c
W
ei
ght
ed
C
o
mbi
nat
i
on
…
(
H
o
n
g
li Y
u
n
)
819
3.
U
s
i
ng
t
he
f
or
m
ul
a
(
2)
t
o
c
a
l
c
ul
a
t
e
v
ar
i
anc
e
of
obs
er
v
a
t
i
on
po
i
nt
s
w
i
t
h
eac
h
m
et
hod,
w
hi
c
h
i
s
s
ho
w
n
i
n T
ab
l
e 5
, T
abl
e
7.
4.
A
c
c
or
di
n
g
t
o
t
h
e
v
ar
i
anc
e
of
obs
er
v
a
t
i
o
n p
o
i
nt
s
i
n T
abl
e 5
,
T
abl
e
6
and
T
abl
e
7,
us
i
ng
m
at
l
ab t
o f
i
t
t
he c
ubi
c
c
ur
v
e
2
i
a
r
σ
−
of
eac
h f
aul
t
t
y
p
es
,
w
hi
c
h
i
s
s
how
n
i
n F
i
gur
e 1
,
F
i
gur
e 2
adn F
i
gur
e
3 a
nd f
or
m
ul
a (
10)
,
(
11)
.
T
abl
e 5
.
E
ac
h
O
bs
er
v
at
i
on
P
oi
nt
’
s
σ
2
of
D
u
v
al
’
s
T
r
ia
n
g
le
f
aul
t
t
y
pe
r
a
(
m
L/
d)
PD
D1
D2
T1
T2
T3
10
0.
4444
0.
3333
0.
4286
0.
0909
0.
2
0.
3333
14
0.
3
0.
2308
0.
4444
0.
125
0.
1538
0.
2857
18
0.
2727
0.
25
0.
4375
0.
1579
0.
2143
0.
3571
22
0.
25
0.
25
0.
3529
0
.
1111
0.
2143
0.
2941
26
0.
1667
0.
2941
0.
35
0.
2
0.
1538
0.
2273
30
0.
3333
0.
2352
0.
2381
0.
2857
0.
2
0.
1667
34
0.
2857
0.
2
0.
2222
0.
2727
0.
25
0.
25
38
0.
3571
0.
2222
0.
2632
0.
4286
0.
3333
0.
25
T
abl
e 6
.
E
ac
h
O
bs
er
v
at
i
on
P
oi
nt
’
s
σ
2
of
B
P
N
eur
a
l
N
e
t
w
or
k
f
aul
t
t
y
pe
r
a
(
m
L/
d)
PD
D1
D2
T1
T2
T3
10
0.
2222
0.
3333
0.
4286
0.
1818
0.
2
0.
3333
14
0.
2
0.
2308
0.
3333
0.
1875
0.
0769
0.
2857
18
0.
2727
0.
25
0.
25
0.
1053
0.
2143
0.
2143
22
0.
1875
0.
15
0.
2353
0
.
1111
0.
1429
0.
2353
26
0.
2222
0.
1
176
0.
15
0.
1333
0.
1538
0.
2727
30
0.
2667
0.
2352
0.
1429
0.
2143
0.
2667
0.
2381
34
0.
2857
0.
2
0
.
1111
0.
1818
0.
1667
0.
25
38
0.
3571
0.
3571
0.
1579
0.
4286
0.
25
0.
25
T
abl
e 7
.
E
ac
h
O
bs
er
v
at
i
on
P
oi
nt
’
s
σ
2
o
f
IE
C
T
hr
ee
-
R
at
i
o M
et
h
od
f
aul
t
t
y
pe
r
a
(
m
L/
d)
PD
D1
D2
T1
T2
T3
10
0.
3333
0.
2222
0.
5714
0.
1818
0.
2
0.
3333
14
0.
2
0.
2308
0.
3333
0.
125
0.
1538
0.
2143
18
0.
1818
0.
25
0.
25
0.
1579
0.
2143
0.
2857
22
0.
0625
0.
15
0.
1765
0
.
1111
0.
1429
0.
1765
26
0.
1667
0.
1
176
0.
1765
0.
0667
0.
0769
0.
2273
30
0.
1333
0.
1
176
0.
1905
0.
0714
0.
1333
0.
1429
34
0.
1429
0.
0667
0.
2778
0.
1818
0.
1667
0.
1875
38
0.
2143
0
.
1111
0.
3158
0.
1429
0.
1667
0.
1667
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SSN
:
1
6
9
3
-
6
930
T
E
L
KO
M
NI
K
A
V
o
l.
14
,
N
o
.
3,
S
ept
em
ber
2016
:
8
1
5
–
82
3
820
F
ig
ur
e
1.
r
a
-
σ
1j
2
C
ur
v
es
of
D
uv
al
’
s
T
r
i
angl
e
F
ig
ur
e
2
.
ra
-
σ
2j
2
C
ur
v
es
of
B
P
N
eur
a
l
N
et
w
or
k
F
i
gur
e
3
.
ra
-
σ
3j
2
C
ur
v
es
of
I
E
C
T
hr
ee
-
R
at
i
o Met
hod
−
×
+
−
×
+
−
×
+
−
×
=
σ
−
×
+
−
×
+
−
×
+
−
×
=
σ
−
×
+
−
×
+
−
×
+
−
×
=
σ
−
×
+
−
×
+
−
×
+
−
×
=
σ
+
−
×
+
−
×
+
−
×
=
σ
+
−
×
+
−
×
+
−
×
=
σ
3
10
3196
.
6
2
10
833
.
5
2
3
10
0424
.
3
3
5
10
4531
.
4
2
16
2
10
6773
.
1
2
10
0829
.
3
2
3
10
6924
.
1
3
5
10
895
.
2
2
15
2
10
4343
.
2
2
10
9074
.
1
2
4
10
9142
.
8
3
5
10
8312
.
1
2
14
3
10
05
.
9
2
10
6674
.
7
2
3
10
869
.
3
3
5
10
3571
.
5
2
13
60915
.
0
2
10
6003
.
4
2
3
10
9067
.
1
3
5
10
5742
.
2
2
12
0523
.
1
2
10
7221
.
8
2
3
10
8818
.
2
3
5
10
8168
.
2
2
11
a
r
a
r
a
r
a
r
a
r
a
r
a
r
a
r
a
r
a
r
a
r
a
r
a
r
a
r
a
r
a
r
a
r
a
r
(1
0
)
+
−
×
+
−
×
+
−
×
=
σ
+
−
×
+
−
×
+
−
×
=
σ
+
−
×
+
−
×
+
−
×
=
σ
+
−
×
+
−
×
+
−
×
=
σ
+
−
×
+
−
×
+
−
×
=
σ
+
−
×
+
−
×
+
−
×
=
σ
71562
.
0
2
10
7433
.
5
2
3
10
2099
.
2
3
5
10
6965
.
2
2
26
40237
.
0
2
10
0829654
.
3
2
3
10
59
.
1
3
5
10
9669
.
1
2
25
12165
.
0
2
10
9904
.
1
2
3
10
7164
.
1
3
5
10
6557
.
3
2
24
64446
.
0
2
10
2133
.
2
2
5
10
2367
.
9
3
6
10
747
.
8
2
23
47983
.
0
2
10
183
.
1
2
4
10
6047
.
5
3
5
10
0545
.
2
2
22
12862
.
0
2
10
6811
.
1
2
4
10
6373
.
9
3
5
10
7902
.
1
2
21
a
r
a
r
a
r
a
r
a
r
a
r
a
r
a
r
a
r
a
r
a
r
a
r
a
r
a
r
a
r
a
r
a
r
a
r
(
11)
σ
3
2
Evaluation Warning : The document was created with Spire.PDF for Python.
T
E
L
KO
M
NI
K
A
I
S
S
N
:
1
693
-
6
930
T
r
ans
f
or
mer
F
a
ul
t
D
i
a
gnos
i
s
Met
ho
d B
as
ed on
D
y
n
am
i
c
W
ei
ght
ed
C
o
mbi
nat
i
on
…
(
H
o
n
g
li Y
u
n
)
821
+
−
×
+
−
×
+
−
×
=
σ
+
−
×
+
−
×
+
−
×
=
σ
+
−
×
+
−
×
+
−
×
=
σ
+
−
×
+
−
×
+
−
×
=
σ
−
×
+
−
×
+
−
×
+
−
×
=
σ
+
−
×
+
−
×
+
−
×
=
σ
52087
.
0
2
10
7187
.
2
2
4
10
4767
.
7
3
6
10
2996
.
7
2
36
13797
.
0
2
10
4403
.
1
2
3
10
0505
.
1
3
5
10
8442
.
1
2
35
26032
.
0
3
10
4056
.
7
2
4
10
4512
.
1
3
6
10
197
.
7
2
34
6181
.
1
1
10
5336
.
1
2
3
10
3237
.
5
3
6
10
7832
.
5
2
33
2
10
4544
.
9
2
10
7082
.
5
2
3
10
9601
.
2
3
5
10
2093
.
4
2
32
8548
.
0
2
10
3136
.
7
2
3
10
2546
.
2
3
5
10
5041
.
2
2
31
a
r
a
r
a
r
a
r
a
r
a
r
a
r
a
r
a
r
a
r
a
r
a
r
a
r
a
r
a
r
a
r
a
r
a
r
(
12)
O
bv
i
ous
l
y
,
w
he
n t
he
t
r
ai
ni
n
g s
am
pl
es
ar
e
not
t
h
e s
am
e,
t
he
2
i
a
r
σ
−
c
ur
v
e
w
i
l
l
b
e
di
f
f
er
ent
.
I
n
t
he
f
aul
t
di
agn
os
i
s
,
put
t
he
a
r
i
nt
o
f
or
m
ul
a
(
10)
-
(
12)
,
t
he
v
ar
i
anc
e
c
an
b
e
c
al
c
ul
at
ed.
U
s
i
ng
f
or
m
ul
a
(
1)
t
o
det
er
m
i
ne
t
he
w
ei
g
ht
s
of
t
hr
e
e
d
i
ag
nos
t
i
c
m
et
hods
f
or
t
h
e
d
i
ag
nos
i
s
of
t
h
e
s
a
m
e t
y
pe
of
f
aul
t
Y
m
,
t
hat
i
s
:
2
3
2
2
2
1
3
2
1
1
:
1
:
1
:
:
m
m
m
m
m
m
p
p
p
σ
σ
σ
=
,
and
∑
=
=
n
i
im
p
1
1
(
13)
P
ut
t
he f
or
m
ul
a (
13)
i
nt
o
t
he c
om
pr
ehens
i
v
e
di
a
gn
o
s
i
s
m
odel
(
8)
,
t
he
f
i
nal
d
i
a
gnos
i
s
r
es
ul
t
s
w
i
l
l
be det
er
m
i
ned.
5.
D
i
ag
n
o
s
i
s
E
xam
p
l
e
U
s
i
ng
w
e
i
g
ht
ed
c
om
bi
nat
i
o
n
m
odel
t
o di
a
gnos
e
T
1
t
y
pe
t
r
ans
f
or
m
er
f
aul
t
,
T
he
s
am
pl
e
dat
a
i
s
c
aus
ed b
y
l
oc
a
l
ov
er
heat
i
n
g of
t
w
o
-
ph
as
e w
i
ndi
ng l
e
ad t
er
m
i
nal
,
w
h
i
c
h
bel
on
gs
t
o t
he
l
o
w
t
em
per
at
ur
e o
v
er
h
eat
f
aul
t
,
t
h
e D
G
A
d
at
a of
t
he s
am
pl
e
ar
e s
ho
w
n i
n T
abl
e 8.
T
abl
e 8
.
C
om
ponent
s
of
D
G
A
D
at
a
gas
H2
CH4
C2
H2
C2
H6
C2
H2
v
ol
um
e
/
pp
m
43.
7
30.
2
46.
6
3.
7
19.
4
T
he t
ot
al
h
y
dr
oc
ar
bo
n ga
s
pr
oduc
t
i
o
n r
at
e i
s
23.
2
m
L/
d,
put
i
t
i
nt
o f
or
m
ul
a (
2)
,
t
he
c
al
c
ul
a
t
ed
v
ar
i
anc
e
of
eac
h
di
a
gnos
i
s
al
gor
i
t
hm
i
n
t
hi
s
s
t
at
e ar
e
as
f
ol
l
o
w
s
:
1)
D
uv
al
’
s
t
r
i
a
ng
l
e m
et
hod
[
]
T
26
545
.
0
18
256
.
0
16
704
.
0
35
626
.
0
24
669
.
0
22
814
.
0
2
1
=
σ
2)
B
P
ne
ur
al
net
w
or
k
[
]
T
23592
.
0
16481
.
0
11607
.
0
19049
.
0
16025
.
0
22346
.
0
2
2
=
σ
3)
T
he i
m
pr
ov
ed t
hr
e
e
-
r
at
i
o m
et
hod
[
]
T
2014
.
0
13701
.
0
10027
.
0
20347
.
0
16215
.
0
11504
.
0
2
3
=
σ
A
c
c
or
di
n
g t
o f
or
m
ul
a (
1)
,
(
5
)
t
o get
t
he
w
e
i
ght
m
at
r
i
x
:
=
3828
.
0
3873
.
0
4058
.
0
3789
.
0
3746
.
0
4953
.
0
3268
.
0
3220
.
0
3506
.
0
4047
.
0
3791
.
0
2550
.
0
2904
.
0
2907
.
0
2436
.
0
2164
.
0
2463
.
0
2497
.
0
P
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SSN
:
1
6
9
3
-
6
930
T
E
L
KO
M
NI
K
A
V
o
l.
14
,
N
o
.
3,
S
ept
em
ber
2016
:
8
1
5
–
82
3
822
U
s
i
ng t
he
abo
v
e
t
hr
ee
al
gor
i
t
hm
s
t
o di
agn
os
e,
t
h
e
r
es
ul
t
i
s
(
P
D
,
D
1,
T
1)
,
t
hen t
he
c
or
r
es
pondi
ng
di
agn
os
i
s
m
at
r
i
x
i
s
:
T
=
0
0
1
0
0
0
0
0
0
0
1
0
0
0
0
0
0
1
S
P
ut
i
t
i
nt
o
f
or
m
ul
a (
7)
,
w
h
i
c
h i
s
t
h
e
w
e
i
g
ht
ed
c
om
bi
nat
i
on d
i
ag
nos
i
s
m
odel
,
c
a
n g
e
t
:
=
0
0
0
0
0
0
0
0
0
0
0
0
0.3828
0.3873
0.4058
0.3789
0.3736
0.4953
0
0
0
0
0
0
0.3268
0.322
0.3506
0.4047
0.3791
0.255
0.2904
0.2907
0.2436
0.2164
0.2463
0.2497
z
S
R
es
ul
t
s
ho
w
s
t
hat
t
he f
our
t
h v
al
ue
i
s
t
he
l
ar
ges
t
i
n
di
a
gona
l
el
em
ent
s
,
w
h
i
c
h
i
s
0.
4058
,
and i
t
c
or
r
es
po
nd t
o t
he f
our
t
h f
aul
t
t
y
pes
T
1,
t
he
c
onc
l
us
i
o
n i
s
c
ons
i
s
t
e
nt
w
i
t
h t
h
e ac
t
ua
l
s
i
t
uat
i
on
.
I
n
t
hi
s
pap
er
,
4
77 s
am
pl
es
of
t
he t
r
a
i
n
i
n
g m
odel
i
s
s
el
ec
t
ed r
a
nd
om
l
y
,
us
i
ng
t
he
w
ei
g
ht
e
d c
om
bi
nat
i
on m
odel
t
o d
i
ag
nos
e,
t
he c
o
nc
l
us
i
ons
ar
e s
ho
w
n
i
n
T
abl
e 9
.
T
abl
e 9
.
D
i
ag
nos
i
s
R
es
ul
t
s
of
W
ei
ght
ed C
om
bi
nat
i
on Mode
l
F
aul
t
t
y
pe
(
s
a
m
pl
e s
i
z
e)
F
al
s
e
num
ber
F
al
s
e
r
at
e%
P
D
(
63)
6
9.
52
D
1
(
75)
8
10.
67
D
2
(
82)
8
9.
76
T
1
(
77)
6
7.
82
T
2
(
85)
8
9.
41
T
3
(
95)
8
8.
42
T
he
T
abl
e 9 s
ho
w
s
t
hat
t
he
m
ax
i
m
u
m
di
agn
os
t
i
c
f
al
s
e
r
at
e of
t
he
w
ei
ght
ed c
om
bi
nat
i
on
di
a
gnos
i
s
m
odel
i
s
10.
67%
,
and
t
h
e
m
i
ni
m
u
m
i
s
7.
82%
.
T
he
y
ar
e
bet
t
er
t
han
a
n
y
of
t
he
abo
v
e
di
a
gnos
i
s
a
l
gor
i
t
hm
.
O
bv
i
o
us
l
y
,
w
i
t
h t
he c
han
ge of
gas
pr
oduc
t
i
on r
at
e t
h
e d
y
n
am
i
c
w
ei
gh
t
ed
v
a
l
ues
c
onf
or
m
s
t
o t
he obj
ec
t
i
v
e c
ondi
t
i
ons
bet
t
er
,
t
h
e c
or
r
ec
t
r
at
e of
t
hi
s
di
agn
os
i
s
m
odel
i
s
m
u
c
h bet
t
er
t
han
t
ha
t
s
i
n
gl
e di
agn
os
i
s
m
et
hod.
6
.
C
o
n
c
l
u
s
i
o
n
I
n t
hi
s
pa
per
,
t
h
e D
G
A
da
t
a
and gas
pr
od
uc
t
i
on r
at
e,
t
he t
w
o i
ndex
es
of
t
he t
r
ans
f
or
m
e
r
f
aul
t
d
i
ag
nos
i
s
,
ar
e c
om
bi
ned.
T
he
di
agn
os
i
s
v
ar
i
a
n
c
e of
D
uv
al
’
s
t
r
i
a
ng
l
e,
B
P
neur
a
l
n
et
w
or
k
and t
he i
m
pr
ov
e
d t
hr
ee r
a
t
i
o m
et
hod i
s
f
i
t
t
ed w
i
t
h t
h
e t
ot
al
h
y
dr
oc
ar
b
on gas
p
r
oduc
t
i
o
n r
at
e
s
epar
at
e
l
y
t
o
pr
od
uc
e t
he
c
ubi
c
c
ur
v
es
.
A
d
y
nam
i
c
w
ei
g
ht
e
d c
om
bi
nat
i
on
di
ag
nos
i
s
m
odel
i
s
es
t
abl
i
s
he
d,
t
hat
i
s
,
t
he
w
ei
g
ht
i
s
d
i
f
f
er
ent
as
t
h
e
g
as
pr
od
uc
t
i
o
n r
at
e c
h
ang
e
s
al
t
hou
gh
t
he
m
et
hod i
s
i
den
t
i
c
a
l
.
T
he r
es
ul
t
s
of
di
ag
nos
i
s
ex
am
pl
es
s
ho
w
t
h
at
t
he
ac
c
ur
ac
y
r
at
e of
t
he
w
ei
g
ht
e
d
c
om
bi
nat
i
o
n
m
od
el
i
s
hi
g
her
t
ha
n
an
y
s
i
ng
l
e
al
gor
i
t
hm
,
and
i
t
i
nc
r
eas
ed
t
o
m
or
e
t
han
90%
i
n m
ai
n t
r
ans
f
or
m
er
f
a
ul
t
di
a
gnos
i
s
,
an
d i
t
has
c
er
t
ai
n
s
t
ab
i
l
i
t
y
as
w
e
l
l
.
T
he m
ode
l
has
bee
n
app
l
i
e
d t
o
t
he
f
aul
t
di
a
gn
os
i
s
s
y
s
t
em
of
a 110
K
V
t
r
ans
f
or
m
er
.
R
ef
er
en
ces
[
1
]
L
i
n
W
M
,
L
i
n
CH,
T
a
s
y
M
X
.
T
r
ans
f
or
m
e
r
-
f
au
l
t
di
a
gn
o
s
i
s
by
i
nt
e
gr
a
t
i
ng
f
i
el
d d
at
a
a
n
d s
t
a
nda
r
d
c
o
de
s
w
i
t
h
t
r
ai
ni
ng
en
ha
nc
i
bl
e
a
da
pt
i
v
e p
r
ob
ab
i
l
i
s
t
i
c
n
et
w
or
k
.
G
e
ner
at
i
on
T
r
an
s
m
i
s
s
i
o
n an
d
D
i
s
tr
i
b
u
ti
o
n
.
2
00
5;
1
52(
3)
:
335
-
3
41.
[
2
]
G
B
/
T
725
2
-
20
01
.
T
h
e
a
na
l
y
s
i
s
a
nd
j
ud
gm
ent
g
ui
de
l
i
ne
s
o
f
t
he
di
s
s
ol
v
ed
ga
s
of
t
r
a
ns
f
or
m
er
o
i
l
.
Evaluation Warning : The document was created with Spire.PDF for Python.
T
E
L
KO
M
NI
K
A
I
S
S
N
:
1
693
-
6
930
T
r
ans
f
or
mer
F
a
ul
t
D
i
a
gnos
i
s
Met
ho
d B
as
ed on
D
y
n
am
i
c
W
ei
ght
ed
C
o
mbi
nat
i
on
…
(
H
o
n
g
li Y
u
n
)
823
[
3
]
S
han
gY
ong
,
Y
an
C
h
un
j
i
a
ng
,
Y
an
Z
ha
ng,
et
al
.
S
y
nt
h
et
i
c
i
ns
ul
at
i
on
f
au
l
t
di
ag
no
s
t
i
c
m
o
del
of
oi
l
-
i
m
m
er
s
ed
p
ow
er
t
r
an
s
f
or
m
er
s
ut
i
l
i
z
i
ng
i
n
f
or
m
a
t
i
on
f
us
i
o
n
.
P
r
oc
ee
di
ng
s
of
t
h
e
C
SEE
.
2
002
;
22(
7)
:
115
-
11
8.
[
4
]
M
o J
uan,
W
a
n
g X
ue,
D
o
ng
M
i
ng,
et
a
l
.
D
i
ag
no
s
t
i
c
m
ode
l
of
i
n
s
ul
at
i
on f
au
l
t
s
i
n pow
er
equ
i
pm
e
n
t
ba
s
ed
o
n
r
ou
gh
s
e
t
t
he
or
y
.
P
r
o
c
e
e
di
ngs
of
t
h
e
C
S
E
E
.
20
04
;
2
4(
7)
:
16
2
-
16
7.
[
5
]
L
i
ao
R
ui
-
j
i
n,
Li
ao
Y
u
-
x
i
ang,
Y
ang
Li
-
j
u
n
,
e
t
a
l
.
St
u
d
y
on
s
y
nt
h
et
i
c
s
i
ag
no
s
i
s
m
et
ho
d
of
t
r
an
s
f
o
r
m
e
r
f
au
l
t
u
s
i
ng m
ul
t
i
-
neu
r
a
l
net
w
o
r
k
an
d
ev
i
d
en
c
t
he
or
y
.
P
r
o
c
e
edi
ng
s
of
t
h
e C
S
E
E
.
20
06
;
2
6(
3)
:
119
-
123
.
[
6
]
C
H
E
N
X
i
aoq
i
ng
,
LI
U
J
u
e
m
i
n,
H
U
A
N
G
Y
i
ngw
ei
,
et
al
.
T
r
a
n
s
f
o
r
m
er
f
au
l
t
d
i
a
gn
os
i
s
us
i
n
g
i
m
pr
ov
e
d
a
r
t
if
ic
ia
l
f
i
s
h
s
w
a
r
m
w
i
t
h
r
oug
h
s
et
al
go
r
i
t
h
m
.
H
i
g
h
V
ol
t
ag
e E
ng
i
ne
er
i
ng
.
201
2;
3
8(
6)
:
1
40
3
-
1
409
.
[
7
]
Y
A
N
G
T
i
ngf
a
ng,
Z
H
O
U
L
i
x
i
ng,
LI
J
i
n
gl
u,
et
a
l
.
P
ow
er
t
r
a
n
s
f
o
r
m
er
f
a
ul
t
di
agn
o
s
i
s
ba
s
e
d
on opt
i
m
a
l
w
ei
ght
s
c
o
m
b
i
n
ed
m
od
el
.
P
o
w
er
s
y
s
t
em
t
ec
hn
ol
o
g
y
.
20
13;
37(
1)
:
19
0
-
19
4.
[
8
]
C
hen S
hi
-
l
on
g
,
R
on
g
J
un
-
x
i
ang,
B
i
G
u
i
-
hon
g,
e
t
al
.
S
p
ec
t
r
u
m
c
om
pa
r
at
i
v
e
s
t
ud
y
o
f
c
om
m
u
t
a
t
i
on
f
ai
l
ur
e
a
nd s
h
or
t
-
c
i
r
c
ui
t
f
aul
t
i
n
U
H
V
D
C
t
r
an
s
m
i
s
s
i
on
s
y
s
t
em
.
T
EL
KO
M
N
I
KA
T
e
l
ec
o
m
m
u
n
i
c
at
i
on
C
om
put
i
n
g
E
l
e
c
t
r
on
i
c
s
an
d
C
ont
r
ol
.
20
14
;
12
(
4)
:
75
3
-
7
62
.
[
9
]
Y
A
N
G
T
i
ngf
an
g,
LI
J
i
n
gl
u
,
Z
e
ng
X
i
a
ng
j
u
n,
et
al
.
L
ar
g
e
pow
er
t
r
an
s
f
or
m
er
f
a
ul
t
di
ag
no
s
i
s
b
as
ed
on
a di
a
gn
os
i
s
m
o
de
l
c
om
bi
ne
d
w
i
t
h s
ev
er
a
l
m
et
h
od
s
.
A
u
t
o
m
at
i
o
n of
e
l
e
c
t
r
i
c
pow
er
s
y
s
t
em
s
.
20
09
;
33(
20)
:
92
-
96
.
[
1
0]
Z
h
ao
W
enq
i
ng
,
L
i
Q
i
ng
l
i
an
g
,
W
ang D
e
w
e
n.
C
om
bi
na
t
i
o
na
l
d
i
a
g
nos
i
s
f
o
r
t
r
a
ns
f
o
r
m
e
r
f
au
l
t
s
bas
ed
on m
u
l
t
i
-
m
od
el
s
.
H
i
gh v
ol
t
ag
e
en
gi
ne
er
i
ng
.
2
01
3;
39(
2)
:
302
-
3
09.
[
1
1]
N
i
h
at
P
am
uk
,
Y
i
l
m
az
U
y
ar
o
g
l
u.
A
n
E
x
per
t
S
y
s
t
em
f
or
P
o
w
er
T
r
ans
f
or
m
er
F
a
ul
t
D
i
a
gn
os
i
s
U
s
i
n
g
A
dv
an
c
e
d
G
e
ner
al
i
z
ed
S
t
oc
h
as
t
i
c
P
et
r
i
N
e
t
.
P
r
z
egl
a
d
E
l
e
k
t
r
ot
e
c
hni
c
z
ny
.
20
12
;
8
8(
8)
:
2
33
-
2
41.
[
1
2]
F
a
ng
-
y
u
R
en
,
S
hu
-
bi
nS
i
,
Z
hi
-
qi
ang
C
ai
,
et
al
.
T
r
a
ns
f
or
m
e
r
f
a
ul
t
an
al
y
s
i
s
ba
s
e
d
on
B
a
y
e
s
i
an
net
w
or
k
s
and i
m
p
or
t
a
n
c
e m
e
as
ur
e
s
.
J
o
ur
n
al
of
S
han
gh
ai
J
i
aot
on
g U
n
i
v
er
s
i
t
y
.
20
15;
2
0(
3)
:
3
53
-
357
.
[
1
3]
Z
HA
O
Ch
e
n
g
-
g
ong
.
R
ea
s
o
ni
ng
on
t
h
e i
nt
er
n
al
f
a
ul
t
s
t
a
t
e
of
t
r
a
ns
f
or
m
er
b
as
ed
on
f
uz
z
y
P
et
r
i
n
et
s
and
g
as
gr
ow
t
h
r
a
t
e
.
P
ow
er
s
y
s
t
em
pr
o
t
e
c
t
i
on
a
nd c
o
nt
r
ol
.
201
3;
41
(
9)
:
12
1
-
12
4.
[
1
4]
Y
e
t
a
i
,
Fe
i
.
E
r
r
o
r
t
he
or
y
a
nd
d
at
a
pr
oc
es
s
i
ng
.
B
ei
j
i
ng:
M
ec
h
ani
c
al
I
n
du
s
t
r
y
P
r
e
s
s
.
20
08
[
1
5]
Z
i
Y
ang
Q
i
a
ng
,
F
e
ng
P
i
ng
W
u
,
J
i
a
R
ui
D
o
ng
,
et
a
l
.
O
pt
i
m
i
z
a
t
i
o
n
of
p
ow
er
s
y
s
t
e
m
s
c
he
du
l
i
ng
ba
s
e
d
on
SC
EM
-
U
A
a
l
go
r
i
t
hm
.
T
E
L
KO
M
N
I
KA
T
el
e
c
om
m
u
ni
c
a
t
i
on
C
om
p
ut
i
n
g
E
l
ec
t
r
o
ni
c
s
a
nd
C
o
nt
r
ol
.
201
5;
13
(
2)
:
41
3
-
4
20
.
[
1
6]
Li
n
C
E
,
Li
ng J
M
,
H
ua
ng
CL
.
A
n
ex
per
t
s
y
s
t
em
f
or
t
r
an
s
f
or
m
er
f
au
l
t
di
ag
no
s
i
s
an
d m
ai
n
t
e
na
nc
e
u
s
i
ng
di
s
s
o
l
v
ed
gas
an
al
y
s
i
s
.
I
EEE
T
r
a
n
s
.
Po
w
e
r
D
e
l
.
20
06
;
2
1(
2)
:
67
3
-
68
0.
Evaluation Warning : The document was created with Spire.PDF for Python.