Int
ern
at
i
onal
Journ
al of Ele
ctrical
an
d
Co
mput
er
En
gin
eeri
ng
(IJ
E
C
E)
Vo
l.
11
,
No.
1
,
Febr
uar
y
2021
, pp.
328
~
335
IS
S
N:
20
88
-
8708
,
DOI: 10
.11
591/
ijece
.
v
11
i
1
.
pp
328
-
335
328
Journ
al h
om
e
page
:
http:
//
ij
ece.i
aesc
or
e.c
om
Appl
ication of pa
rticle sw
ar
m
opti
mizati
on
with AN
FIS mod
el
for doub
le scroll
chaotic s
ystem
W. A.
Wali
Depa
rtment
o
f
C
om
pute
r
Engi
n
e
eri
ng,
Coll
ege of
Engi
n
ee
ring
,
B
a
srah
Univer
sit
y
,
Ira
q
Art
ic
le
In
f
o
ABSTR
A
CT
Art
ic
le
history:
Re
cei
ved
Ma
r 21,
2020
Re
vised
Jun
22
,
20
20
Accepte
d
J
ul
10, 2
020
The
pr
edi
c
ti
ons
for
the
origi
n
al
ch
aos
pa
tt
er
ns
ca
n
be
used
to
cor
r
ect
the
distor
te
d
ch
aos
pat
t
ern
whic
h
has
cha
ng
ed
d
ue
to
an
y
cha
ng
es
whether
from
undesire
d
disturba
nc
e
or
addi
ti
ona
l
informati
on
which
c
an
hide
under
cha
os
pa
tt
ern
.
T
his
informati
on
ca
n
b
e
r
ec
ove
re
d
when
th
e
or
ig
ina
l
cha
os
pat
t
ern
is
pre
dicted.
But
unpre
d
i
ct
a
b
il
ity
is
m
ost
fea
ture
s
of
chaos
,
and
ti
m
e
serie
s
pre
d
ic
t
ion
ca
n
b
e
used
b
ase
d
on
th
e
co
ll
e
ction
of
past
obse
rva
ti
ons
of
a
var
ia
bl
e
and
ana
l
y
sis
it
to
obta
in
the
under
l
y
ing
re
la
t
ionshi
ps
and
the
n
ext
rap
o
la
t
e
futur
e
ti
m
e
serie
s.
Th
e
addi
ti
on
al
info
rm
at
ion
ofte
n
prune
s
awa
y
b
y
seve
r
al
t
ec
h
nique
s.
Thi
s
p
ape
r
show
s
ho
w
the
cha
ot
ic
ti
m
e
serie
s
pre
diction
is
diff
ic
ult
and
distort
eve
n
if
n
eur
o
-
fu
zzy
such
as
adaptive
neur
a
l
fuz
z
y
infe
r
ence
s
y
stem
(AN
FIS
)
is
used
under
a
n
y
disturba
n
ce.
The
paper
combined
par
t
icle
sw
arm
(PS
O)
and
(AN
FIS
)
to
exa
m
the
pr
edicti
on
m
odel
and
pre
di
ct
the
origi
nal
cha
os
pat
te
rns
whic
h
comes
from
the
doubl
e
scroll
ci
r
cui
t
.
Change
s
in
the
b
ia
s
of
the
nonlinear
resistor
we
re
used
as
a
disturba
nc
e.
The
pre
di
ct
ed
cha
ot
ic
data
is
compare
d
with
dat
a
from
the
cha
ot
ic c
ir
cu
it
.
Ke
yw
or
d
s
:
ANFIS
Chaotic
ti
m
e s
eries
PSO
This
is an
open
acc
ess arti
cl
e
un
der
the
CC
B
Y
-
SA
l
ic
ense
.
Corres
pond
in
g
Aut
h
or
:
Wasa
n Wali
,
Dep
a
rtm
ent o
f
Com
pu
te
r
E
ng
i
neer
i
ng,
Ba
srah U
niv
e
r
sit
y
,
Ba
srah, Ira
q.
Em
a
il
:
wasan
.
wali
@uo
basr
a
h.
e
du.iq
1.
INTROD
U
CTION
Nonlinea
r
pr
e
dicti
on
of
c
ha
otic
tim
e
series
is
ve
ry
chall
eng
i
ng
in
t
he
pr
e
dicti
on
area
[1
]
.
Ne
ur
al
netw
orks
a
nd
relat
ed
neuro
-
f
uzzy
m
od
el
s
s
uch
as
(ANF
I
S)
hav
e
bee
n
t
he
s
ubj
ect
s
of
interest
du
e
to
thei
r
m
any
pr
act
ic
al
app
li
cat
ion
s
in
m
od
el
ing
c
om
plex
nonl
inear
syst
em
s
[2
-
4]
an
d
in
chao
ti
c
tim
e
series
pr
e
dicti
on
s
[
5
-
10
]
but
w
he
n
the
num
ber
of
obser
vatio
ns
for
trai
ning
is
lim
it
ed
or
t
he
data
does
not
h
a
ve
a
sim
il
ar
patte
r
n
t
hey
can
neither
rec
onstr
uct
the
dynam
ic
s
nor
le
a
rn
the
s
hap
e
of
at
tract
or
an
d
the
pred
ic
ti
on
beco
m
e
m
or
e
diff
ic
ult.
In
t
hi
s
pap
er
,
we
use
d
the
asy
m
m
et
rical
do
ub
l
e
scro
ll
at
tract
or
as
a
n
obsta
cl
e
fo
r
the
best
te
c
hn
i
qu
e
w
hich
is
t
he
A
NFIS
m
od
el
to
pr
e
dict
the
ori
gi
nal
patt
ern
a
nd
pr
opose
the
od
d
in
for
m
at
ion
wh
ic
h
does
not
m
a
tc
h
with
the
or
i
gin
al
pa
tt
ern
.
T
his
oddness
in
i
nfo
rm
ation
can
be
with
pur
pos
e
or
as
distor
ti
ons
.
T
he
pa
per
sho
ws
how
it
is
dif
ficult
an
d
t
he
disto
rtions
th
at
happ
e
ned
on
the
pe
rd
it
io
ns
of
chao
ti
c
at
tract
or
by
us
i
ng
t
he
ANFIS
m
od
el
unde
r
distu
r
ba
nce.
C
hao
s
i
n
el
ect
rical
ci
rcu
it
s
has
draw
n
s
tro
ng
at
te
ntion
[10
-
15]
since
t
he
ci
rcu
it
prov
i
des
a
sim
ple
veh
ic
le
for
the
ex
pe
rim
ental
ob
ser
vation
an
d
c
om
pu
te
r
si
m
ulati
on
of
c
hao
ti
c
phe
no
m
ena.
T
he
do
ub
l
e
scro
ll
ci
rcu
it
is
us
ed
f
or
nonl
inear
tim
e
seri
es
analy
sis
m
eth
od
s
for
co
nf
i
rm
ing
the
chao
ti
c
be
hav
i
or
Fi
gure
1
sho
ws
the
do
ub
le
sc
ro
ll
ci
rc
uit,
ci
rcu
it
ry,
a
nd
nonlinea
r
re
sist
or
char
act
e
risti
c. Fig
ure
2
s
hows
sym
m
et
ric d
ouble
sp
iral
att
r
act
or
.
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
Ap
plicati
on o
f
pa
rti
cl
e sw
ar
m
opti
miza
ti
on
wi
th ANFI
S m
od
el
f
or
…
(
W.
A. W
ali
)
329
Chaotic
syst
em
s
are
sensitiv
e
dep
e
ndenc
y
on
i
niti
al
conditi
ons
an
d
densel
y
dis
trib
uted
per
i
odi
c
po
i
nts
[16].
T
he
data
colle
ct
ed
al
ong
inev
it
able
no
ise
m
akes
the
predi
ct
ion
of
a
cha
otic
syst
e
m
har
d
t
o
achieve.
T
his
diff
ic
ulty
pr
op
os
es
the
m
od
el
li
ng
chao
ti
c
ti
m
e
series.
Ar
ti
fici
al
In
te
ll
igence
te
chn
i
qu
e
s
hav
e
gaine
d
si
gn
ific
ant
im
po
rtance
in
m
od
el
li
ng
because
of
t
heir
abili
ty
to
le
ar
n
f
r
om
histor
ic
al
obser
vations
dat
a
and
pr
e
dict
hi
gh
ly
nonli
near
syst
e
m
s.
AN
F
IS
is
a
ty
pe
of
syst
e
m
that
inco
r
porates
ne
ural
and
fu
zzy
log
ic
.
Neur
on
s
le
ar
n
fr
om
exp
e
rience
an
d
pe
rfo
rm
ed
by
op
ti
m
iz
at
ion
of
th
e
antece
de
nt
an
d
c
oncl
us
io
n
pa
rt
s
par
am
et
ers.
Th
is
pap
er
us
es
ANFIS
to
pr
e
dict
a
tim
e
ser
ie
s
gen
erate
d
by
a
doub
le
sc
ro
ll
ci
rcu
it
bu
t
us
in
g
asym
m
e
tric
al
a
tt
ra
ct
or
as
kind
of
the
noise
vi
a
gen
e
rated
as
ymm
et
rical
dat
a
patte
rn
a
nd
e
valuates
the
A
NF
I
S
pr
e
dicti
on in
t
hi
s case. F
i
gure
3
s
hows
the
bl
ock d
ia
gr
am
o
f
ANFIS
pr
e
dic
ti
on
.
Figure
1. D
oubl
e scro
ll
circ
uit an
d nonli
nea
r resi
stor cha
ract
erist
ic
s
Figure
2. Sym
m
et
rical
d
oubl
e stran
ge
at
trac
tor
Figure
3. Bl
oc
k diag
ram
o
f
A
NF
I
S
pr
e
dicti
on
2.
ASYM
METR
ICA
L
DOUB
LE
STRANG
E ATTR
AC
T
OR
Gen
e
rall
y,
the
ph
e
nom
eno
n
of
m
ulti
ple
at
tract
or
s
is
m
os
tly
in
sy
m
m
e
tric
dyna
m
ic
al
sys
tem
s
[17
-
18]
.
T
hese
ex
hib
it
pairs
of
m
utu
al
ly
sy
m
m
e
tric
at
tract
or
s,
as
a
par
a
m
et
er,
are
va
ried.
The
sym
metry
i
n
double
scr
oll
stran
ge
at
tract
or,
w
hich
was
obser
ve
d
from
t
he
el
ect
ronic
ci
rcu
it
s,
m
ay
no
t
be
hel
d
in
s
pecial
cases
for
var
io
us
reas
ons,
for
exam
ple,
switc
hing
cha
racte
risti
cs
of
the
di
od
es
beca
us
e
of
the
dif
fer
e
nc
es
in
bias
val
ues
or
i
m
per
fect
m
at
c
hing
tra
ns
ist
ors
or
a
ny
ot
her
reasons
t
hat
can
m
ake
asym
m
et
ry
in
the
stran
ge
at
tract
or
.
A
nal
ysi
s
of
the
no
nl
inear r
esi
st
or circuit
that
is
s
how
n
i
n
Fi
gure
4
in
dicat
es
tha
t
the b
rea
k
volt
age
of
the
v
-
i
c
harac
te
risti
c
is
bias
dep
e
ndent.
T
hi
s
can
be
relat
ed
t
o
the
s
witc
hing
c
har
act
e
risti
cs
of
the
di
od
e
s.
Th
us
it
is
e
xpe
ct
ed
that
t
he
bi
as
sou
rces
+
Vc
c
an
d
–
Vcc,
pl
ay
a
certai
n
r
ol
e
as
a
bif
urcat
ion
el
em
ent.
Figure
5
sh
ows
the e
ff
e
ct
o
f
the
bias
on a
double
sc
rol
l at
tract
or
.
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.
11
, No
.
1,
Febr
uar
y
2021
:
328
-
335
330
Figure
4. N
onli
near resist
ors
Figure
5. Asy
m
m
e
tric
al
d
ou
ble stra
ng
e
att
r
act
or
The
pr
e
dicti
on
of
a
c
ha
otic
syst
e
m
fr
om
no
isy
obser
vat
ion
s
is
ve
ry
ha
rd.
No
isy
data
co
uld
be
con
ta
m
inate
d
f
ro
m
diff
e
ren
t
t
ypes
of
s
ource
s.
N
oise
ca
n
be
pro
pagat
ed
i
nto
t
he
pr
e
dicti
on
m
od
el
a
nd
m
ake
real
pro
blem
s
i
n
m
any
of
cha
os
real
-
li
fe
ap
pl
ic
at
ion
s.
N
ois
e
on
c
ha
otic
ti
m
e
series
pr
e
di
ct
ion
ha
s
bee
n
bar
el
y
consi
der
e
d
[19
]
to
i
m
pr
ove
t
he
perform
ance
pr
e
dicti
on
in
the
prese
nce
of
noi
se.
In
t
his
w
ork,
we
focus
on
how
t
o
m
ake
a
cha
nge
in
bias
as
distu
rb
a
nce
data
am
ong
t
he
or
i
gin
al
data
to
exam
ine
t
he
pr
e
dicti
on
of
ANFIS
base
d
on
c
hao
ti
c
no
isy
obse
rv
a
ti
on
s
.
Fi
gure
6
sho
ws
the
blo
c
k
dia
gr
am
of
the
disturbanc
e
pr
e
dicti
on syst
e
m
.
Figure
6. Distu
rb
a
nce
on a c
ha
otic sy
ste
m
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
Ap
plicati
on o
f
pa
rti
cl
e sw
ar
m
opti
miza
ti
on
wi
th ANFI
S m
od
el
f
or
…
(
W.
A. W
ali
)
331
3.
COMB
INE
D
PSO
WITH
A
NFIS
TO
RE
JE
CT
THE
BIAS
DI
STU
RB
ANCE
The
idea
be
hin
d
P
SO
is
each
par
ti
cl
e
kee
ps
track
of
the
coo
r
din
at
es
i
n
the
sp
ace
of
the
pr
oble
m
wh
ic
h
is
ass
oc
ia
te
d
with
th
e
best
s
olu
ti
on
[
20
]
.
Ma
ny
stu
dies
h
a
ve
c
ombine
d
P
SO
with
f
uzzy
a
nd
A
NF
I
S
i
n
diff
e
re
nt
app
li
cat
ion
s
[
21
-
25]
.
In
m
os
t
of
them
the
PSO
trie
s
to
op
tim
iz
e
the
AN
F
IS
par
am
et
e
rs
to
giv
e
the
best
s
olu
ti
on
s
.
I
n
ou
r
w
ork,
the
PS
O
play
s
an
im
po
rtant
r
ole
in
r
e
m
ov
in
g
any
undesire
d
data
an
d
i
m
pr
ovin
g
t
he
pr
e
dicti
on
si
gnal
.
PSO
sta
rts
by
ta
king
the
t
i
m
e
-
series
data
from
the
ci
rc
uit
unde
r
cha
nge
a
s
the
init
ia
l
first
swar
m
generat
ion
.
T
he
cl
os
e
ness
of
eac
h
pa
rtic
le
to
the
be
st
so
luti
on
de
pends
on
t
he
obj
ect
iv
e
functi
on.
T
he
obj
ect
ive
is
th
e
chao
ti
c
tim
e
-
s
eries
data
from
the
or
igina
l
doub
le
sc
ro
ll
ci
rcu
it
with
out
an
y
disturba
nce.
Fi
gure
7
an
d
Fi
gure
8
s
how
the
PSO
-
A
NFIS
m
od
el
blo
ck
di
agr
am
.
Figure
9
sho
ws
the
gr
aph
ic
al
represe
ntati
on
s
.
Con
si
der
i
ng
t
he
searc
h
sp
ac
e
of
2
-
dim
ension
a
nd
(n)
pa
r
ti
cl
es
as
an
ini
t
ia
l
po
pula
ti
on
was
ta
ken
from
the
ci
rcu
it
after
t
he
cha
nge
of
bi
as.
The
re
is
a
par
t
of
t
hese
pa
rtic
le
s
that
ha
ve
a
w
r
ong
posit
ion
du
e
to
the
c
ha
nge.
Each
par
ti
cl
e
has
a
s
pecifi
ed
posit
ion
an
d
vel
ocity
that
is
ass
ociat
ed
wi
th
it
s
pa
rtic
ular
best
perform
ance
in
the
s
war
m
.
Each p
arti
cl
e
trie
s
to
m
od
ify
it
s
po
sit
io
n
with
it
s
obj
ect
iv
e
w
hi
ch
is
s
pecified
f
r
om
the ch
a
otic t
im
e
-
series
data
from
the o
ri
gin
al
double sc
ro
ll
c
ircuit
Figure
7. Distu
rb
a
nce
on a c
ha
otic sy
ste
m
Figure
8.
PS
O
-
ANFIS p
re
dicti
on
blo
c
k dia
gra
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.
11
, No
.
1,
Febr
uar
y
2021
:
328
-
335
332
Figure
9
.
PSO
gr
a
phic
al
re
pr
e
sentat
ion
s
4.
RESU
LT
S
A
ND
ANALY
S
IS
The
sim
ulati
on w
as im
ple
m
ented
usi
ng Mat
la
b
a
nd the
res
ul
ts arra
ng
e
d
as
the foll
owin
g:
Gen
e
rate
the
chao
ti
c
tim
e
-
series
data
f
rom
the
doub
le
scro
ll
ci
rcu
it
(V
c
1
an
d
Vc
2
)
w
hich
s
how
in
Figure
10
.
Ma
ke
the
ch
a
nge in
+Vcc
1
i
n n
on
li
nea
r
resis
tor w
hich
is
show
n
i
n
Fi
gure
1
Gen
e
rate m
or
e
ch
a
otic data
from
the ch
aotic
ci
rcu
it
after
the
ch
a
ng
e
.
Use
AN
F
IS
t
o
pr
e
dict
cha
otic
data.
Fi
gure
11
sho
ws
the
com
par
ison
be
tween
th
e
sim
ulati
on
s
data
f
ro
m
the circ
uit an
d pr
e
dicat
ion
s
of the
ANFIS m
od
el
. T
he fig
ur
e
shows the
d
ist
or
ti
ons i
n
the
a
tt
ractor
The dist
ort
ion
s
b
eca
us
e
of the
addit
ion
al
c
ha
ng
e
d data.
The
sim
ulati
on w
as im
ple
m
ented
usi
ng Mat
la
b
a
nd the
res
ul
ts arra
ng
e
d
as
the foll
owin
g:
Gen
e
rate
the
chao
ti
c
tim
e
-
series
data
fro
m
the
do
uble
scro
ll
ci
rcu
it
(V
c
1
an
d
Vc
2
)
wh
ic
h
sho
w
in
Figure
10
.
Ma
ke
the
ch
a
nge in
+Vcc
1
i
n n
on
li
nea
r
resis
tor w
hich
is
show
n
i
n
F
i
gure
1
Gen
e
rate m
or
e
ch
a
otic data
from
the ch
aotic
ci
rcu
it
after
the
ch
a
ng
e
.
Use
AN
F
IS
t
o
pr
e
dict
ch
a
otic
data.
Fi
gure
1
1
sho
ws
the
com
par
ison
be
tween
th
e
sim
ulati
on
s
data
f
ro
m
the circ
uit an
d pr
e
dicat
ion
s
of the
ANFIS m
od
el
. T
he fig
ur
e
shows the
d
ist
or
t
io
ns
i
n
the
a
tt
ractor
The dist
ort
ion
s
b
eca
us
e
o
f
the
addit
ion
al
c
ha
ng
e
d data.
Applie
d
th
e
P
SO
on
t
he
A
N
FI
S
m
od
el
dat
a.
Fig
ur
es
12
,
13
,
an
d
14
sho
w
the
ANFI
S
pr
e
dicti
on
s
m
od
el
after P
SO w
hic
h
a
pp
li
ed
ac
c
ordin
g
to
the
flo
wch
a
rt in
Fig
ure
7.
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
Ap
plicati
on o
f
pa
rti
cl
e sw
ar
m
opti
miza
ti
on
wi
th ANFI
S m
od
el
f
or
…
(
W.
A. W
ali
)
333
Figure
12
s
ho
ws
the p
a
rtic
le
s
update
in
t
hei
r
posit
ions
acc
ordin
g
to
t
he
i
nd
i
vidual
nea
r
est
obj
ect
ives
t
aken
from
the circu
i
t wit
hout
disturbance
.
Figures
13
a
nd
14
s
how PS
O
-
ANFIS p
re
dicti
on
on the
att
ra
ct
or
.
PSO
-
A
NFIS
c
an
be
us
ed
to
pr
e
dict
an
d
se
pa
rate
any
kind
of
hid
de
n
pieces
of
in
form
at
i
on
f
ro
m
any
no
is
y
s
ource
si
gn
al
as
sho
wn
in
Fig
ur
e
15
a
nd
it
was
te
ste
d
unde
r
inc
om
pl
et
e
chao
ti
c
dat
a
wh
ic
h
gav
e
a
go
od
est
i
m
ation
hence,
we
can
say
it
d
eser
ved to
be u
nd
e
r
c
onsid
erati
on
.
Figure
10
. D
ouble
scr
oll
ci
rc
ui
t
Figure
11
. D
ouble
scr
oll
ci
rc
ui
t
Figure
12
. Part
ic
le
s u
pdat
ed
in
thei
r po
sit
io
ns
accor
ding t
o
th
e ind
i
vidual
ne
arest o
bject
ive
s
Figure
13
. PS
O
-
ANFI
S
pr
e
di
ct
ion
Figure
14
. PS
O
-
ANFI
S
pr
e
di
ct
ion
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.
11
, No
.
1,
Febr
uar
y
2021
:
328
-
335
334
Figure
15
.
PS
O
-
ANFI
S m
odel
u
se
d
f
or
pr
e
dicti
on
a
nd s
pe
rati
on
s
5.
CONCL
US
I
O
N
Howe
ver,
PSO
-
A
NFIS
sho
ws
the
diff
ic
ultl
y
to
pr
e
dict
the
or
i
gin
al
double
scro
ll
at
tract
or
wh
e
n
it
is
distor
te
d,
sti
ll
,
it
is
a
go
od
m
et
ho
d
to
rec
ov
e
r
pa
rtic
le
s
wh
ic
h
we
re
ab
sent
due
to
an
y
chan
ge
a
nd
recove
r
the
or
igi
nal
pa
tt
ern
.
This
m
eth
od
al
so
can
be
us
ed
to
rec
over
the
hi
dd
e
n
inform
ation
unde
r
the
at
tract
or
by
recog
nizing t
he
odd pa
rtic
le
s
after they
we
re
co
m
par
ed
w
it
h or
i
gin
al
att
ra
ct
or
data.
REFERE
NCE
S
[1]
J.
SID
orowic
h,
"M
odel
ing
of
chaotic
ti
m
e
seri
es
for
pre
dic
t
ion,
in
t
erp
olation,
and
sm
oothi
ng
,
"
IEEE
Inte
rnationa
l
Confe
renc
e
on
A
cousti
cs,
Spe
ec
h
,
and
Signa
l Proc
essing,
vol
.
4
,
pp
.
121
-
124
,
1992
.
[2]
F.
Prado,
e
t
al
.
,
"F
ore
ca
st
ing
base
d
on
an
en
sem
ble
Autore
g
ressive
Mo
ving
Avera
ge
-
Adap
t
ive
neur
o
-
Fuz
zy
infe
ren
c
e
s
y
st
em
-
Neura
l
n
et
work
-
Gen
et
i
c
Algor
i
thm Fram
ework
,
"
Ene
rgy
,
vol
.
1
97,
pp
.
117
-
159
,
2020
.
[3]
S.
Kar,
et
al
.
,
"
Applic
a
ti
ons
of
neur
o
fuz
z
y
s
y
st
ems
:
A
brie
f
rev
ie
w
and
future
o
utl
ine
,"
Appl
i
ed
Soft
Computing
,
vol.
15
,
pp
.
243
-
259,
2014
,
[4]
M.
Male
ki
za
de
h,
et
a
l
.
,
"S
hort
-
te
rm
loa
d
fo
rec
ast
using
en
sem
ble
neur
o
-
f
uzzy
m
odel
,"
Ene
rgy,
vo
l.
1
96,
pp.
117
-
127
,
20
20.
[5]
S.
Ganje
far
,
“
Optimiza
ti
o
n
of
quant
um
-
inspire
d
neur
al
ne
twork
using
m
emeti
c
al
gor
i
thm
for
func
ti
on
appr
oximati
on
a
nd
cha
o
ti
c
t
ime
serie
s pr
edicti
on
,
"
Neurocomputing
,
vol
.
291
,
pp
.
175
-
186,
2018
.
[6]
H.
Rad
,
et
al
.
,
"
Predic
ti
on
of
ro
ck
m
ass
rat
ing
s
y
stem
base
d
on
cont
inuous
func
t
ions
using
Chaos
–
AN
FI
S
m
odel
,"
Inte
rnational
Jo
urnal
of Roc
k
M
ec
hani
cs
and
M
i
ning
Sc
ie
nc
es,
v
ol.
73
,
pp
.
1
-
9,
2
015.
[7]
M.
Abdolla
hz
ad
e
,
e
t
al
.
,
"A
new
h
y
brid
enha
n
ced
loc
a
l
li
n
ea
r
n
e
uro
-
fuz
z
y
m
odel
base
d
on
the
op
ti
m
iz
ed
singul
ar
spec
trum
ana
l
y
s
is
and
it
s
appl
i
ca
t
ion
for
nonlinear
and
cha
o
tic
ti
m
e
serie
s
fore
ca
st
ing
,"
Infor
mation
Sci
en
ce
s
,
vol.
295
,
pp
.
107
-
125,
2015
.
[8]
Y.Bod
y
anski
y
,
et
al
.
,
“
H
y
brid
ada
pti
v
e
wave
le
t
-
n
eur
o
-
fuz
z
y
s
y
stem
for
chaotic
ti
m
e
seri
es
ide
nti
fi
cation
,
"
Information
Sc
ience
s
,
vo
l.
220,
p
p.
170
-
179
,
201
3.
[9]
A.
Pano
-
Azuc
en
a,
et
al
.
,
"P
red
i
c
ti
on
of
cha
ot
ic
tim
e
serie
s
b
y
usi
ng
AN
Ns
,
AN
FI
S
and
SV
Ms
,
"
7th
Inte
rnat
ional
C
onfe
renc
e
on
Mode
rn Cir
cuits and
Syst
ems Te
chnol
ogi
es
,
pp
.
1
-
4,
2018
.
[10]
M.
Nhaba
ngue
,
et
al
.
,
"Chao
tic
ti
m
e
serie
s
pre
dic
t
ion
wi
th
func
ti
on
al
l
ink
ext
rem
e
l
ea
rning
AN
FIS
(FL
-
EL
AN
FIS
),
"
Inte
rnationa
l Confe
renc
e
on
P
ower,
Instrum
en
tat
ion, Cont
rol
a
nd
Computing,
p
p.
1
-
6
,
2018
.
[11]
A.
Davie
s,
W
.
Schwarz
,
"N
onli
nea
r
D
y
n
amics
of
El
ectroni
c
S
y
st
ems
,"
P
roce
edi
n
gs
Of
The
Work
shop
Ndes
World
Sci
en
ti
fic,
1993.
[12]
W
.
Marsza
l
ek
,
el
al
.
,
"2D
Bif
u
rca
t
ions
and
Ch
aos
in
Nonl
ine
a
r
Circ
u
it
s:
a
Pa
ral
l
el
Com
putati
onal
Approa
ch,
"
15
th
Inte
rnation
al
Confe
ren
ce
on
Synt
hesis,
M
odel
ing
,
Analysi
s
and
Simulat
io
n
Me
tho
ds
and
Appl
i
cat
ions
t
o
Circui
t
D
esign
(
SMACD)
,
pp.
1
-
300,
2018
.
[13]
B.
Sam
ard
zi
c,
e
t
al
.
,
"A
naly
sis
of
spati
al
cha
os
appe
ara
n
ce
in
ca
sca
d
e
connect
ed
nonli
near
el
e
ct
ri
ca
l
c
irc
ui
ts
,
"
Chaos,
Solitons
&
F
ractal
s,
vol.
95,
pp
.
14
-
20
,
2
017.
[14]
C.
W
ang,
et
a
l
,
"Capt
uring
and
shunting
ene
rg
y
in
cha
o
tic
Chu
a
ci
r
cui
t
,”
Chao
s,
Soli
tons
&
Fr
act
als,
vo
l.
134
,
2020.
[15]
G.
Le
u
tc
ho
,
et
al
.
,
"D
y
nami
c
al
ana
l
y
sis
of
a
nove
l
au
tono
m
ous
4
-
D
h
y
pe
rje
rk
ci
r
cui
t
wit
h
h
y
p
erb
ol
ic
sin
e
nonli
ne
ari
t
y
:
Ch
aos,
antim
onotoni
ci
t
y
and
a
pl
ethora
of
coe
x
isti
n
g
at
tracto
rs
,"
Ch
aos,
Soli
tons
&
Fract
als,
vol
.
10
7,
pp.
67
-
87
,
2018
.
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
Ap
plicati
on o
f
pa
rti
cl
e sw
ar
m
opti
miza
ti
on
wi
th ANFI
S m
od
el
f
or
…
(
W.
A. W
ali
)
335
[16]
S.
Chung,
et
al
,
"Regul
ar
sensiti
vity
computatio
n
avoi
ding
cha
o
ti
c
eff
ec
ts
in
pa
rt
ic
l
e
-
in
-
ce
l
l
plas
m
a
m
et
hods
,
"
Journal
of
Computati
onal
Ph
ysics
,
vol. 400,
2020
.
[17]
V.
Vait
hia
n
at
h
a
n
and
J.
Veij
un,
“
Coexi
stenc
e
o
f
four
diffe
ren
t
at
tr
ac
tor
s
in
a
fu
ndamenta
l
powe
r
sy
st
em
m
odel
,
”
IEE
E
Tr
ansacti
o
ns on
Circuits a
nd
Syste
ms
I
,
vo
l
.
46
,
pp
.
405
-
40
9,
1999
.
[18]
J.
Kengne
,
“
Coexi
sten
ce
of
c
haos
with
h
y
pe
rch
aos,
per
iod
-
3
doubli
ng
bifurc
ation,
and
tr
ansie
nt
ch
aos
in
the
h
y
per
cha
o
ti
c
oscil
l
at
or
wi
th
g
y
ra
tors,
”
Inte
rn
ati
onal
Journal
of
Bifurcati
on
a
nd
Chaos
in
App
li
ed
S
ci
en
ce
s
an
d
Engi
ne
ering,
vol
.
25
,
2015
.
[19]
D.
Karuna
singha
,
et
al
.
,
"Enh
ancem
ent
of
ch
aot
i
c
h
y
drolog
ical
t
i
m
e
serie
s
pre
di
ction
with
re
al
-
t
ime
noise
red
u
ct
io
n
using E
xt
ende
d
Kalman
Filt
er,
”
Journal
of
Hydrology,
vol
.
565
,
pp.
737
-
746
,
20
18.
[20]
M.
AkliKac
imi,
et
al
.
,
"N
ew
m
ixe
d
-
codi
ng
PS
O
al
gorit
hm
for
a
self
-
ada
p
ti
ve
an
d
aut
om
at
ic
l
ea
r
ning
of
Mam
dani
fuz
z
y
ru
le
s
,"
En
gine
ering
Applic
ati
ons of
Artifici
al
Int
el
l
ige
nc
e,
v
ol.
89
,
2020
.
[21]
P.
A.
Adede
ji
,
e
t
al
.
,
"W
ind
turbine
power
output
ver
y
short
-
t
erm
fore
ca
st:
A
compara
t
ive
stud
y
o
f
dat
a
cl
ust
eri
ng
tech
n
ique
s in
a
PS
O
-
ANFIS m
od
el
,"
Journal
o
f
C
le
aner
Product
io
n,
vol
.
254
,
2020
.
[22]
M.
Rezaka
z
emi,
et
al
.
,
"H
2
-
sel
ective
m
ixe
d
m
at
ri
x
m
embrane
s
mode
li
ng
using
AN
FIS
,
PSO
-
ANFIS,
GA
-
ANFIS
,
"
Inte
rnational
Jo
urnal
of
Hydrog
en
En
ergy
,
vol
.
42,
no
.
22
,
pp
.
1
5211
-
15225,
20
17.
[23]
P.
A.
Aded
ej
i
,
e
t
al
,
"W
ind
turbine
power
ou
tput
ver
y
short
-
te
rm
fore
c
ast:
A
co
m
par
at
ive
stud
y
of
data
c
luste
r
in
g
te
chn
ique
s in
a
PS
O
-
ANFIS m
od
el
,"
Journal
o
f
C
le
aner
Product
io
n,
vol
.
254
,
2020
.
[24]
M.
Ali,
Muhl
asi
n,
e
t
al
.
,
"Com
bine
d
AN
FIS
m
et
hod
with
FA
,
PS
O,
and
ICA
as
Stee
ring
Con
trol
Optimiza
t
ion
o
n
El
e
ct
ri
c
Car
,
"
El
e
ct
rica
l
Pow
e
r,
Elec
troni
cs,
Comm
unic
ati
ons,
Controls,
and
Informatic
s
S
e
minar
(
EE
CCIS)
,
pp.
299
-
304
,
20
18.
[25]
Y.
K.
Sem
ero
,
e
t
al
.
,
"P
V
power
fore
ca
sting
usin
g
an
int
egr
at
ed
GA
-
PSO
-
ANFI
S
appr
oac
h
and
Gauss
ia
n
proc
ess
reg
ression
base
d
fea
ture
sel
ec
t
ion
strat
eg
y
,
"
CSEE
Journal
of
Powe
r
and
Ene
rgy
Syste
ms
,
vol.
4,
no.
2
,
pp.
210
-
218
,
20
18.
Evaluation Warning : The document was created with Spire.PDF for Python.