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.
8
, No
.
6
,
Decem
ber
201
8
, p
p.
4673
~
4683
IS
S
N: 20
88
-
8708
,
DOI: 10
.11
591/
ijece
.
v8
i
6
.
pp
4673
-
46
83
4673
Journ
al
h
om
e
page
:
http:
//
ia
es
core
.c
om/
journa
ls
/i
ndex.
ph
p/IJECE
Hybrid
Gen
etic
Algorith
m for
Op
timizati
on of F
ood
Compositi
on on
Hyperte
nsive Pa
tient
Ap
ri
li
a Nur
F
au
z
iya
h
,
W
ay
an
Fir
daus
M
ah
mu
dy
Facul
t
y
of
Com
pute
r
Sc
ie
nc
e, Br
awij
a
y
a
Univ
ersi
t
y
,
Indon
esia
Art
ic
le
In
f
o
ABSTR
A
CT
Art
ic
le
history:
Re
cei
ved
Dec
20
, 201
7
Re
vised
Ju
l
16
,
201
8
Accepte
d
J
ul
3
1
, 2
01
8
The
he
alth
y
foo
d
with
a
tt
en
ti
on
of
salt
d
egr
e
e
is
one
of
th
e
ef
fort
s
for
hea
l
t
h
y
li
ving
of
h
y
p
er
te
nsive
pa
ti
en
t.
The
eff
ort
is
important
for
red
uci
ng
th
e
proba
bil
i
t
y
of
h
yper
te
nsion
cha
n
ge
to
b
e
d
ange
ro
us
disea
se.
In
th
i
s
stud
y
,
th
e
food
compos
it
ion
is
bui
ld
with
at
t
ent
ion
nutr
it
i
on
amount,
salt
degr
ee,
an
d
m
ini
m
um
cost.
The
proposed
m
et
hod
is
h
y
brid
m
et
hod
of
Gene
t
ic
Algorit
hm
(GA
)
and
Vari
abl
e
Neighbor
ho
od
Sear
ch
(VN
S).
The
three
sce
nar
ios
of
h
y
brid
GA
-
VN
S
t
y
pes
had
b
ee
n
deve
lop
ed
in
t
his
stud
y
.
Altho
ugh
h
y
br
id
GA
and
V
NS
ta
ke
m
ore
ti
m
e
tha
n
pure
GA
or
pure
VN
S
but
t
he
proposed
m
et
hod
give
better
qualit
y
of
soluti
on.
VN
S
succ
essfull
y
h
el
p
GA
avoi
ds
pre
m
at
ure
conv
erg
ence
and
im
prove
s
better
so
lut
ion.
Th
e
shor
tc
om
ings
on
GA
in
loc
al
ex
ploi
tation
and
p
remat
ur
e
conv
er
genc
e
is
solved
b
y
VN
S,
where
as
the
sho
rtc
om
ing
on
VNS
tha
t
le
ss
ca
pa
bil
ity
in
glob
al
expl
ora
ti
on
ca
n
b
e
solv
ed
b
y
use
GA
th
at ha
s
adva
n
ta
g
e
in
gl
obal
expl
ora
ti
on
.
Ke
yw
or
d:
Food com
po
sit
ion
Hybr
i
d ga
-
vns
Hype
rtensio
n
Copyright
©
201
8
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
:
Wayan
Fir
dau
s
Mahm
ud
y
,
Faculty
of Com
pu
te
r
Scie
nc
e
,
Bra
wij
ay
a
U
niv
e
rsity
,
Jal
an Vet
eran
8,
Mal
a
ng, In
done
sia
.
Em
a
il
:
wayanfm
@u
b.
ac.id
1.
INTROD
U
CTION
The
non
-
com
m
un
ic
able
dis
ease
on
deca
de
of
this
ye
ar
beco
m
e
the
le
adin
g
of
disea
se
that
ca
us
e
death
in
I
ndon
esi
a.
As
e
xam
ple
one
of
the
non
-
c
omm
un
ic
a
ble
disease
tha
t
has
m
any
pati
ents
is
hype
rte
ns
io
n.
No
t
only
as
non
-
c
omm
un
ic
able
disease
bu
t
al
so
hype
rtensi
on
i
s
m
ai
n
fact
or
of
cr
onic
disease
app
ea
re
d,
su
c
h
as
car
diovasc
ul
ar,
ca
ncer
,
dia
betes,
a
nd
re
spi
rati
on
disease
[1
]
.
This
is
cl
ose
ly
relat
ed
to
the
patte
r
n
of
hum
an
li
fe d
ai
ly
su
ch a
s f
oo
d
co
nsu
m
ed,
nutrit
ion
a
l st
at
us
, d
ie
t an
d
ot
her
facto
rs suc
h
as b
l
ood pr
es
s
ure,
ob
e
sit
y, an
d
cho
le
ste
r
ol a
nd
ins
ulin r
e
sist
ance
[2
]
.
The
foo
d
co
nsum
ed
as
one
of
the
facto
rs
th
at
aff
ect
blood
pr
e
ssure
on
hy
per
te
ns
i
ve
pa
ti
ent
if
not
at
te
ntion
can
c
ause
de
vel
op
i
ng
da
nger
ouse
di
sease.
O
ne
of
the
bloo
d
press
ur
e
c
on
t
ro
l
on
hype
rtensiv
e
pa
ti
ent
can b
e
done by con
s
um
ed
foo
d
with less
of
s
al
t [3
]
. Th
is i
s caus
e
d
co
nsum
ed
of salt
is
m
a
in f
act
or of inc
rease
blood
press
ur
e
an
d
a
pp
ea
r
c
ard
i
ov
asc
ular
disease
[
2].
T
her
e
fore,
co
nsum
e
fo
od
with
ap
propriat
e
nutrit
io
n
need an
d
le
ss
of salt
d
e
gree i
s
need
e
d by
hype
rtensive
p
at
ie
nt for co
ntr
olli
ng b
l
ood p
ress
ur
e
.
The
cal
culat
io
n
on
f
ood
co
m
po
sit
ion
can
be
done
m
anu
al
ly
or
by
software
.
Ma
nua
ll
y,
it
m
or
e
diff
ic
ult
to
cal
culat
e
f
ood
c
om
po
sit
ion
beca
us
e
nu
t
riti
ent
am
ou
nt
m
us
t
be
ap
pro
pr
ia
te
wi
th
nutrit
ie
nt
ne
ed
on
patie
nt.
B
ut
w
it
h
software
,
it
can
be
done
easi
er
with
opt
i
m
iz
e
on
searc
h
of
foo
d
c
ompo
sit
io
n
in
ord
er
t
o
cl
os
er
with
nutrit
ie
nt n
ee
d on
hype
rtensive
pat
ie
nt [
4].
Seve
ral
te
ch
niq
ue
s
ca
n
be
use
d
for
s
olv
i
ng
f
ood
com
posit
ion
prob
le
m
.
I
wuji,
et
al
[
5]
im
ple
m
ent
li
near
pro
gr
am
m
ing
for
s
olv
i
ng
foo
d
com
posit
ion
pro
blem
on
hype
rtensi
ve
patie
nt.
That
researc
h
ser
ves
diet
plan
that
fu
lf
il
l
tolerance
l
i
m
i
t
in
D
ASH
diet
nutrie
nt
a
nd
obat
ai
n
c
os
t
m
ini
m
um
.
Howev
e
r
li
nea
r
pro
gr
am
m
ing
m
ay
r
equ
ire
s
hi
gh
c
om
pu
ta
t
io
nal tim
e [6
,
7
]
.
Be
side
li
near
pro
gr
am
m
ing
m
et
hod,
t
he
m
eta
-
he
ur
ist
ic
m
eth
od
ca
n
be
use
d
f
or
so
l
ving
optim
iz
at
ion
pro
blem
s.
Met
a
-
he
ur
ist
ic
gi
ve
s
so
luti
on
tha
t
cl
os
er
opti
m
al
,
high
pe
rform
ance,
an
d
fl
exible
in
rea
s
on
a
ble
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.
8
, N
o.
6
,
Dece
m
ber
2
01
8
:
4673
-
4683
4674
tim
e
co
m
pu
ta
ti
on
[
8].
On
e
of
the
m
et
a
-
heu
ri
sti
c
m
et
ho
ds
is
Ge
netic
Algo
r
it
h
m
(GA)
.
H
usse
in
[9
]
us
e
G
A
to
determ
ine
i
m
p
or
ta
nt
featu
res
that
hav
e
i
nf
l
uen
ce
on
cl
ass
ifie
r
of
hea
rt
at
ta
ck.
G
A
s
uc
cessf
ully
i
m
p
rove
d
qu
al
it
y
of
t
he
resu
lt
cl
assifi
e
r
bec
om
e
87
%
,
after
t
he
prev
iou
s
m
et
ho
d
by
K
-
Me
a
ns
on
ly
reach
accu
r
acy
on
68%.
Othe
r
re
search
by
Şahm
an,
et
al
[10]
i
m
ple
m
ent
Gen
et
ic
al
gori
thm
in
cost
op
tim
iz
at
ion
for
feed
reco
m
m
end
at
ion
on
poultry
and
cat
tl
e.
T
heir
ap
proac
h
giv
e
lo
wer
c
os
t
than
us
in
g
li
near
pro
gr
a
m
m
ing
.
Ra
hm
an,
et
al
[11]
buil
t
evo
l
u
ti
on
a
ry
al
gori
thm
fo
r
so
l
ving
shrim
p
diet
fo
rm
ulati
on
pro
blem
.
Their
ap
proac
h
can
giv
e
feasi
ble
so
l
utio
n
a
nd
they
al
so
im
pe
m
ent
wh
at
-
if
sce
nar
i
o
to
pro
ve
that
e
voluti
onary
m
od
el
is
rob
us
t.
Our
previ
ou
s
r
esearch
[
4]
i
m
plem
ents
VN
S
fo
r
s
olv
in
g
f
ood
c
om
po
sit
ion
pro
blem
on
hype
rtensi
ve
patie
nt.
Although
obta
in
fea
sible
so
l
ution
and
f
ulfill
nutrie
nt
nee
d
on
hype
rtensive
pa
ti
ent
and
gi
ve
cos
t
m
ini
m
u
m
.
Ho
wev
e
r,
the
us
e
ness
of
V
NS
sti
ll
giv
e
le
ss
per
f
or
m
ance
on
s
earch
the
s
olu
t
ion
.
T
hat
was
cause
d
VNS less
of ex
pl
orat
ion capa
bili
ty
.
Seve
ral
resear
chs
ex
plain,
lo
cal
m
e
tho
d’s
s
hortcom
ing
can
be
so
l
ve
d
by
i
m
ple
m
ent
oth
er
m
et
ho
d
that
ha
ve
go
od
exp
l
or
at
io
n
ca
pab
il
it
y.
Su
c
h
the
resea
rch
by
W
i
j
ay
ani
ngr
um
,
et
al
[1
2]
pro
po
se
d
Hyb
r
id
G
A
and
Sim
ulate
d
Anneali
ng
(
SA)
f
or
s
olv
i
ng
f
ood
c
om
po
sit
io
n
pr
ob
le
m
on
poultry.
T
hat
r
esearch
pro
pos
ed
S
A
to
av
oid
local
op
ti
m
u
m
so
luti
on
on
GA.
T
he
res
ult
of
m
et
h
od
pro
pose
d
gi
ve
bette
r
s
olu
ti
on
tha
n
the
use
nes
s
cl
assic
G
A
.
Othe
r
seve
ral
r
esearch
s
[
13
]
-
[
15
]
use
G
A
an
d
V
NS
f
or
im
pro
ving
pe
rform
ance
of
sea
r
ch
so
l
ution.
The
us
ene
ss
hybr
i
d
G
A
a
nd
V
NS
had
bee
n
gi
ve
outpe
rfor
m
resu
lt
on
se
ve
r
al
researc
hs
.
T
hat
beca
us
e
G
A
ha
s
adv
a
ntage
s
on
global
exp
l
orat
ion
,
al
th
ough
the
oth
e
r
side
GA
has
sho
rtcom
ing
s
on
l
oc
al
exp
loit
at
io
n
an
d
pr
em
at
ur
e
co
nver
ge
nce
[
16
]
.
Howe
ver,
G
A’s
shortc
om
ing
can
be
so
l
ved
by
i
m
ple
m
ent
l
ocal
base
d
al
gorithm
so
betwee
n
gl
ob
al
e
xplorati
on
a
nd
local
e
xp
l
oitat
ion
ca
n
be
balance
d
[
12
]
.
V
NS
is
one
of
the
local
base
d
al
gorithm
that
has
ad
van
at
a
ge
on
searc
hing
local
so
luti
on
[
13
]
.
V
NS
e
xp
l
or
e
neig
hborh
oods
f
ro
m
current
incum
ben
t
s
olu
ti
on
[
8
]
,
[
17]
.
Most
of
the
so
luti
on
with
op
ti
m
al
value
will
be
us
ed
to
getti
ng
pro
m
isi
ng
neig
hborh
ood
so
luti
on
[17].
In
order
to
ob
t
ai
n
local
op
ti
m
a,
VNS
m
ov
e
fr
om
on
e
so
luti
on
to
oth
e
r
so
l
ution
with
m
pl
e
m
ent
local
search
im
pr
ove
m
ent.
Wh
e
n
opti
m
u
m
loc
al
is
detect
ed
,
V
NS
ca
n
jum
p
ou
t
f
rom
trap
with
chang
e
neig
hborh
ood
structu
re
an
d
f
ind
bette
r
s
olut
ion
.
T
her
e
for
e
VN
S
ca
n
preven
t
opti
m
iz
at
ion
proses
fall
into
local
op
ti
m
a
[
8].
Be
side,
no
t
li
ke
sing
le
so
l
ution
ba
se
d
m
et
hod
li
ke
S
A,
Tabu
Searc
h,
VNS
not
nee
d
m
any
par
am
et
er
an
d
ve
ry
sim
ple
[4
]
,
[
8
]
,
[
19]
.
Neig
hbor
hood
str
uctu
re
a
nd
m
ov
e
str
at
egy
bec
om
e
enou
gh
par
am
et
er
for
run
ning
V
NS.
The
a
dv
a
ntage
s
on
G
A
a
nd
VNS
can
giv
e
adv
a
ntag
es
hy
br
idiza
ti
on
G
A
a
nd
VNS
[4,
13,
18]
.Th
e
refo
re
on
this
researc
h
will
be
do
ne
hybri
dizat
ion
betwee
n
G
A
and
VNS.
The
m
et
hod
pro
po
se
d wit
h t
heir
a
dv
a
ntage
s is ho
ped can
giv
e
bette
r
s
ol
ution t
ha
n resul
t our
pr
e
vious
researc
h [4
]
.
2.
RESEA
R
CH MET
HO
D
On
this
resear
ch
a
ny
hy
br
idi
zat
ion
te
ch
ni
que
s
a
re
propos
ed
betwee
n
G
A
a
nd
V
NS
,
there
are
G
A
wh
ic
h
is
im
pr
ov
ed
by
V
NS,
VNS
wh
ic
h
is
add
e
d
with
G
A
a
nd
a
no
t
her
scenari
o
of
hybr
i
dizat
ion
of
GA
V
N
S
is repr
oductio
n o
n GA i
s im
pr
ov
e
d by
VNS.
2.1.
So
luti
on Rep
r
esent
at
i
on
The
ar
rangem
ent
of
foo
d
sol
ution
is
re
pr
e
sentat
ed
with
integer
nu
m
ber
on
eve
ry
their
nu
trit
ie
nt
.
In
te
ger
num
ber
re
fer
to
in
dex
of
foo
ds
tu
ff
s
on
f
ood
data
ba
se.
T
he
s
olu
ti
on
represe
ntati
on
is
sho
wn
by
Table
1.
Table
1.
So
l
uiton Re
prese
ntati
on
Ty
p
e of
m
e
n
u
Breakf
ast
Lun
ch
Din
n
er
PK
3
1
2
S
1
3
10
N
2
12
11
H
16
6
3
B
16
18
2
6
Inform
at
ion
:
PK
=
Staple
f
ood
S
=
Ve
getable
N
=
Plant
sourc
e
H
=
Me
at
sou
rc
e
B
=
Fr
uit
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
Hyb
ri
d Ge
netic
Alg
or
it
hm fo
r Opti
miza
ti
on
of F
ood
C
omp
os
it
ion o
n Hy
pe
rte
ns
iv
e
... (
A
pr
il
ia Nu
r F
auziy
ah
)
4675
Fr
om
Table
1,
the
so
luti
on
co
ntain
five
c
ompone
nts
of
foo
d,
the
re
are
sta
ple
foo
d,
ve
get
ables,
pla
nt
so
urce,
m
eat
s
ource
am
d
fr
ui
t.
In
it
ia
l
gen
on
c
ro
m
os
om
is
gen
erate
d
r
andom
ly
with
fu
lfil
l
a
m
ou
nt
each
nu
t
riti
ent
com
pone
nt.
T
he
i
nt
eger
on
foo
d
that
ra
ndom
ly
gen
e
rated
m
us
t
be
no
m
or
e
t
han
li
m
it
ation
each
foo
d
am
ou
nt
t
hat
avail
able
on
data.
10
3
data
are
us
ed
in
t
his
resea
rc
h
th
a
t
it
sa
m
e
with
data
on
our
previo
us
researc
h [4
]
.
2.2.
Fitness
Func
tion
Qu
al
it
y
of
s
ol
ution
is
m
easur
ed
by
fitness
functi
on.
The
fitness
functi
on
that
will
be
us
e
d
on
this
stud
y
is
sam
e
with
the
fitnes
s
f
un
ct
io
n
on
our
previ
ou
s
res
earch
[
4]
so
th
at
the
res
ult
of
pure
V
NS
a
nd
res
ult
of h
y
br
idiza
ti
on
GA an
d V
NS can
be
c
om
pared. The
f
it
ness
functi
on is s
ho
wn on
(1)
=
10000
(
∗
)
+
(1)
The
obj
ect
i
ve
functi
on
on
(
1)
e
xp
la
in
that
the
cal
culat
io
n
fitness
value
of
a
s
olu
ti
on
is
do
ne
with
cal
culat
e
pen
al
ty
each
of
nutr
it
ie
nt
co
m
po
ne
nt
and
c
os
t.
10
00
is
co
nsona
nt
nu
m
ber
,
cost
is
a
m
ou
nt
of
al
l
cost
from
fo
odstuff
s.
So
t
hat
fitne
ss
value
is
obta
ined
by
div
isi
on
f
ro
m
10
00
0
and
t
otal
of
p
e
nalty
and
c
o
st
in
each
so
luti
on.
2.3.
Repr
od
uc
tio
n
Re
producti
on
on
G
A
c
onsist
ing
c
ro
ss
over
an
d
m
utati
on
.
Re
pro
duct
io
n
fa
se
ai
m
s
t
o
producti
ng
offsprin
g.
In
th
is
research
us
e
on
e
c
ut
po
i
nt
cro
ss
over
that
sel
ect
cut
po
int
rando
m
ly
and
exch
a
nge
rig
ht
side
and
le
ft
side
from
two
par
e
nt
that
ha
d
bee
n
sel
ect
ed
ra
ndom
ly
.
The
m
utati
on
that
is
us
e
d
in
t
his
resea
r
ch
is
reciprocal
e
xc
ha
ng
e
m
utati
on
wh
ic
h
is e
xc
ha
ng
e
tw
o
in
de
x t
hat h
a
d bee
n
s
el
ect
ed
in
one
par
e
nt.
2.4.
Sele
ction
Sele
ct
ion
ai
m
s
to
sel
ect
c
rom
os
o
m
that
wi
ll
be
us
ed
on
popula
ti
on
ne
xt
generati
on
de
pends
on
it
s
fitness
value.
I
n
this
researc
h,
us
es
el
et
is
m
sel
ect
ion
.
Eli
ti
sm
wo
rk
s
with
so
rti
ng
t
heir
f
it
ness
val
ue
an
d
th
e
n
sel
ect
d
epe
nds
on their
h
i
gh
es
t fit
ness value
as m
uch
as
popula
ti
on
siz
e.
2.5.
Scenari
o of h
ybri
diz
at
ion
G
A and
VNS
The
hy
br
idiza
t
ion
sce
nar
io
G
A
V
NS
w
hich
is
i
m
ple
m
ent
ed
in
this
rese
arch
will
be
exp
la
ine
d
on
sever
al
s
ub
c
ha
pters
.
Brie
fly
,
thre
e
sce
nar
i
o
of
hybri
diza
ti
on
propose
d
on
this
st
ud
y,
there
a
re
GA
VNS
wh
ic
h
is
the
th
e
final
so
luti
on
of
G
A
will
be
i
m
pr
oved
by
V
NS
,
furthe
rm
or
e
on
this
stu
dy
will
be
wr
it
te
n
GA
VNS
I
.
T
he
se
cond
sce
na
rio
is
GA
V
NS
w
hich
is
t
he
re
pro
duct
ion
on
GA
will
be
c
om
bin
ed
with
VNS
on
m
ul
ti
ples
of
50th
ge
ner
at
io
n,
f
ur
t
her
m
or
e
will
be
wr
it
te
n
GA
V
NS
I
I.
A
nd
the
la
st
scenari
o
is
V
NS
GA
wh
ic
h
is
In
it
ia
l
popu
la
ti
on
of
GA
a
re
obta
in
ed
by
V
NS
.
V
NS
will
be
r
un
as
m
uch
as
popu
la
ti
on
siz
e
so
that
giv
e a
ble s
olu
ti
on
s
as m
uch
a
s
populat
ion si
z
e f
or
i
niti
al
p
opulati
on
on GA.
2.5.1.
GA
-
V
NS I
GA
V
NS
I
m
et
hod
is
sta
rte
d
with
i
niti
al
i
ze
GA
pa
ram
et
er
corres
pondin
g
with
pa
r
a
m
et
er
value
wh
ic
h
had
bee
n
pre
viously
te
ste
d.
T
he
n,
G
A
fase
is
r
un
r
epeate
dly
unti
l
te
rm
inati
on
c
onditi
on
is
f
ulfill
ed.
Af
te
r
final
so
l
ution
is
f
or
m
ed
from
GA
an
d
then
nex
t
ste
p
is
runn
i
ng
V
NS
that
is
app
l
ie
d
for
i
m
pr
ovi
ng
final
s
olu
ti
on
G
A.
V
NS
us
es
f
in
al
s
olu
ti
on
GA as i
ts i
niti
al
so
luti
on. GA
VNS
I
proce
dure is s
how
n by Fi
gure
1.
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.
8
, N
o.
6
,
Dece
m
ber
2
01
8
:
4673
-
4683
4676
PROCE
DU
RE G
A
-
VNS
I
Inpu
t
:
p
o
p
_
size :
p
o
p
u
latio
n
size
g
en
eration
: gen
eration
size
Cr
:
c
ros
so
v
er
rate
v
alu
e
M
r
:
m
u
t
atio
n
r
ate valu
e
K
m
ak
s
: ne
ig
h
b
o
rh
o
o
d
Pros
es:
//
Gen
erate
initial s
o
lu
tio
n
r
an
d
o
m
l
y
a
s
m
u
ch
as po
p
_
size
G
enerat
e ini
tial p
o
pula
tion P(
t)
t
=
0
WH
IL
E
(t
<=g
ene
ra
tion)
//Rep
rod
u
ctio
n
Cro
ss
o
v
er(P
(
t)
,
Cr)
M
utatio
n
(P
(t
)
,
M
r)
C(t)
<
-
c
ros
so
v
er
o
f
f
s
p
ring
,
m
u
tat
atio
n
off
sp
ring
Ev
a
lua
tion()
Selection
(P(
t+1
))
f
ro
m
P(
t)
and
C(t)
t
=
t +
1
END W
H
IL
E
curr: GA
f
ina
l so
l
ution
//Ap
p
ly
VNS
Cu
rr
GA
f
in
al so
lu
tio
n
b
est
cu
rr
k
1
WH
IL
E
k
<=
K
m
a
k
s
DO
// sh
ak
in
g
curr
Cha
ng
e (
b
est,
k
)
bestLo
ca
l
Lo
ca
l
sea
rch (
c
urr)
//
m
o
v
e or no
t
IF F
i
tness(bes
tLo
ca
l)
> Fi
tness(b
es
t)
T
H
EN
Bes
t
b
estLo
ca
l
k
1
EL
SE
k
k
+1
END IF
END W
H
IL
E
O
utp
ut:
b
est: th
e
b
est so
lu
tio
n
f
rom
GA VN
S I
END P
RO
CEDU
RE
Figure
1. Pse
udoc
ode
of
GA
-
VNS I
2.5.2.
GA
-
VNS II
GA
V
NS
II
is
sta
rted
with
in
it
ia
li
ze
GA
para
m
et
er
li
ke
po
pu
la
ti
on
siz
e,
gen
e
rati
on
siz
e,
cro
ss
ove
r
a
m
d
m
utati
on
ra
te
that correspondin
g wit
h
pa
ram
et
er v
al
ue
f
ro
m
G
A
te
sti
ng. Next ste
p,
G
A
is r
un
with
firstly
gen
e
rates
s
olu
t
ion
ra
ndom
ly
.
Howe
ver,
on
the
repr
oductio
n
proses
,
VNS
is
ap
plied
on
eve
ry
m
ulti
pl
es
of
50
t
h
gen
e
rati
on.
V
NS
is
a
pp
li
ed
on
s
ubset
of
offs
pr
i
ng
f
or
sear
chi
ng
bet
te
r
so
l
ution
ar
ound
in
div
i
du
each
popula
ti
on
.
Sol
ution
from
of
fsprin
g
G
A
a
nd
so
l
utio
n
f
rom
VN
S
in
e
ve
r
y
offs
pr
i
ng
will
be
eval
ua
te
d
use
obj
ect
ive
f
unct
ion
.
If
G
A
so
l
ution
is
bette
r
than
s
olu
ti
on
f
ro
m
VN
S
then
so
luti
on
is
no
t
rep
la
ced
.
Othe
rw
ise
,
GA so
l
utio
n w
il
l be r
e
placed
by so
l
utio
n
f
r
om
V
NS
. GA
VNS II
pro
c
ed
ure is sh
own
b
y
Figure
2.
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
Hyb
ri
d Ge
netic
Alg
or
it
hm fo
r Opti
miza
ti
on
of F
ood
C
omp
os
it
ion o
n Hy
pe
rte
ns
iv
e
... (
A
pr
il
ia Nu
r F
auziy
ah
)
4677
PROC
EDU
RE G
A
-
VNS
II
Inpu
t
:
p
o
p
_
size :
p
o
p
u
latio
n
size
g
en
eration
: gen
eration
size
Cr
:
c
ros
so
v
er
rate
v
alu
e
M
r
:
m
u
t
atio
n
r
ate valu
e
N_
cr
:
child
nu
m
b
er
on
cr
o
ss
o
v
er
Cr * p
o
p
_
size
N_
m
r
:
ch
i
ld
nu
m
b
er
on
m
u
ta
tio
n
M
r
*
p
o
p
_
size
Pros
es:
//
Gen
erate
initial s
o
lu
tio
n
r
an
d
o
m
l
y
a
s
m
u
ch
as po
p
_
size
G
enerat
e ini
tial p
o
pula
tion P(
t)
t
=
0
WH
IL
E
(t
<=g
ene
ra
tion)
//Rep
rod
u
ctio
n
Cro
ss
o
v
er(P
(
t)
,
cr)
if
(genera
tion
% 50
)
Ch
ild
_
itr
=
0
Wh
ile(C
hild
_
itr
<= N_
cr
)
VNS
()
if(
fit
n
ess
_
cr >
fi
tness_
v
ns)
C(t)
<
-
s
o
lu
tio
n
_
cr
else
C(t)
<
-
s
o
lu
tio
n
_
v
n
s
END
Whil
e
M
utatio
n
(P
(t
)
,
m
r
)
if
(genera
tion
% 50
)
Ch
ild
_
itr
=
0
Wh
ile(C
hild
_
itr
<= N_
m
r)
VNS
()
if(
fit
n
ess
_
m
r
> f
i
tness_
v
ns)
C(t)
<
-
s
o
lu
tio
n
_
m
r
else
C(t)
<
-
s
o
lu
tio
n
_
v
n
s
END
Whil
e
C(t)
<
-
o
f
f
sp
rin
g
f
ro
m
cr
o
ss
o
v
er,
of
f
sp
ring
f
ro
m
m
u
tat
io
n
Ev
a
lua
tion()
Selection
(P(
t+1
))
f
ro
m
P(
t)
dan
C(t)
t
=
t +
1
END W
H
IL
E
O
utp
ut:
b
est: th
e
b
est so
lu
tio
n
f
rom
GA VN
S I
I
END P
RO
CEDU
RE
Figure
2. Pse
udoc
ode
of GA
-
VNS
II
2.5.3.
VNS
-
GA
VNS
G
A
m
eth
od
is
sta
rte
d
by
V
NS
proc
edure
that
is
app
li
ed
f
or
form
ing
set
of
s
olu
ti
on.
T
he
so
luti
on
will
be
us
ed
on
init
ia
l
po
pula
ti
on
of
GA.
Be
cause
VN
S
is
sing
le
so
luti
on
based
then
VNS
pro
cedure
will
be
rep
eat
ed
un
ti
l
f
ulfill
popula
ti
o
n
si
ze.
A
fter
i
niti
al
popula
ti
on
on
G
A
is
buil
t,
G
A
proce
dure
is
rep
eat
e
dly r
un
un
ti
l f
ulfill
ter
m
inati
on
c
ondi
ti
on
. V
NS
G
A sce
nar
i
o
is s
ho
wn b
y
ps
e
udoc
od
e
on Fi
gure
3.
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.
8
, N
o.
6
,
Dece
m
ber
2
01
8
:
4673
-
4683
4678
PROCE
DU
RE V
NS
-
GA
Inpu
t
:
p
o
p
_
size :
p
o
p
u
latio
n
size
g
en
eration
: gen
eration
s
ize
Cr
:
c
ros
so
v
er
rate
v
alu
e
M
r
:
m
u
t
atio
n
r
ate valu
e
K
m
ak
s
: ne
ig
h
b
o
rh
o
o
d
Pros
es:
pop
=
0
WH
IL
E
(pop<
=p
o
p_
siz
e)
curr: Gen
erate i
n
itial so
lution ra
nd
o
m
ly
//Ap
p
ly
VNS
best
curr
k
1
WH
IL
E
k
<=
K
m
a
k
s
DO
//
sh
ak
in
g
curr
Cha
ng
e (
b
est,
k
)
bestLo
ca
l
Lo
ca
l
sea
rch (
c
urr)
//
m
o
v
e or no
t
IF F
i
tness(bes
tLo
ca
l)
> Fi
tness(b
es
t)
T
H
EN
Bes
t
b
estLo
ca
l
k
1
EL
SE
k
k
+1
END IF
END W
H
IL
E
b
est: VNS so
lu
tio
n
END W
H
IL
E
All VNS
so
lution
:
best
//
in
itialize initial
s
o
lu
tio
n
o
n
GA
=
V
NS so
lu
tio
n
Initia
l popu
la
tion
P
(
t)
=
Al
l VN
S so
l
ution
t
=
0
WH
IL
E
(t
<=g
ene
ra
tion)
//Rep
rod
u
ctio
n
Cro
ss
o
v
er(P
(
t)
,
Cr)
M
utato
n
(
P
(t
),
M
r)
C(t)
<
-
o
f
f
sp
ring
f
r
o
m
cr
o
ss
o
v
er,
of
f
sp
ring
f
ro
m
m
u
tatio
n
Ev
a
lua
tion()
Selection
(P(
t+1
))
f
ro
m
P(
t)
and
C(t)
t
=
t +
1
END W
H
IL
E
O
utp
ut:
b
est: th
e
b
est so
lu
tio
n
f
rom
VNS G
A
END P
RO
CEDU
RE
Figure
3. Pse
udoc
ode
of VN
S
-
GA
3.
RESU
LT
S
A
ND AN
ALYSIS
In
this
sect
ion,
the
resu
lt
of
this
stud
y
will
be
ex
plained
on
seve
ral
severa
l
su
b
-
c
ha
pters
,
there
are
su
b
3.1
will
exp
la
in
a
bout
a
naly
sis
on
GA,
sub
3.2
will
exp
la
in
a
bout
analy
sis
on
G
A
V
NS
I
,
sub
3.3
will
exp
la
in
abo
ut a
naly
sis o
n GA
VNS II,
and s
ub
3.4 wil
l exp
l
ai
n
ab
out an
al
ysi
s on VN
S
G
A.
3.1.
Gene
tic A
lgo
ri
t
hm
(G
A)
The
par
am
et
er
value
will
be
us
e
on
G
A
te
sti
ng
a
nd
hybi
dizat
ion
of
G
A
is
dete
rm
in
ed
ba
sed
on
te
sti
ng
that
pr
evi
ou
sly
ha
d
done.
G
A
pa
ram
et
er
value
that
will
b
e
us
ed
are
popula
ti
on
siz
e
=
100,
gen
e
rati
on
=
12
5,
a
nd
cr
os
s
ov
er
rate
=
0.6,
m
utati
on
rate
=
0.
4.
T
he
im
ple
m
entat
ion
us
i
ng
GA
on
th
is
res
earch
giv
e
prem
at
ur
e co
nv
e
r
gen
ce
, t
he patt
ern o
f
c
onve
rg
e
nce is
sh
ow
n on Fig
ure
4.
Figure
4
sho
w
GA
ca
n
im
pr
ove
but
after
53th
gen
e
rati
on,
pr
em
at
ur
e
c
onverge
nce
is
ha
pp
e
ne
d
unti
l
la
st
gen
e
rati
on
didn’t
giv
e
be
tt
er
fitness.
P
rem
at
ur
e
conv
erg
e
nce
on
G
A
can
be
s
olve
d
by
a
ddit
ion
oth
e
r
m
et
ho
d
for
av
oid
in
g
pr
em
atu
re
c
onve
rg
e
nc
e.
Fu
t
her
m
ore,
on
this
pa
pe
r
we
pro
po
se
hybr
i
dizat
ion
on
G
A
with a
no
t
her m
et
hod wh
ic
h
is
will
b
e a
pp
li
e
d o
n
this
stu
dy is VNS.
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
Hyb
ri
d Ge
netic
Alg
or
it
hm fo
r Opti
miza
ti
on
of F
ood
C
omp
os
it
ion o
n Hy
pe
rte
ns
iv
e
... (
A
pr
il
ia Nu
r F
auziy
ah
)
4679
Figure
4. G
raph
of co
nver
gence patt
er
n on G
A
3.2.
GA
-
V
NS
I
The
first
ex
pe
rim
ent
on
this
research
is
co
m
bin
ing
hybr
i
d
G
A
with
V
NS
.
T
he
proce
dure
of
GA
-
VNS
I
on
this
researc
h
ha
s
be
en
desc
ribe
d
i
n
sect
ion
2.5.1
wh
ic
h
is
firstly
GA
is
r
un,
a
nd
t
he
final
s
ol
ution
GA
is
us
ed
asi
niti
al
so
luti
on
i
n
VNS.
The
re
s
ult
f
ro
m
hybri
d
GA
-
VN
S
I
is
te
ste
d
on
10
e
xp
e
rim
ents
an
d
giv
e
gr
eat
er
fitness
than
pure
GA.
The
gr
a
ph
of
diff
e
re
nce
fitn
ess
hybri
d
GA
-
V
NS
I
an
d
pure
GA
is
s
ho
wn
i
n
Figure
5.
Figure
5. G
raph
of com
par
iso
n betwee
n pure
GA a
nd h
y
br
i
d GA V
NS
Althou
gh
wh
e
n
in
the
end
of
GA
ge
ner
at
io
n,
G
A
has
co
nver
ge
nce,
it
ca
n
sti
ll
achieve
higher
fitne
s
s
with
a
pp
ly
in
g
VNS.
From
the
Fig
ur
e
5
sho
ws
VNS
ca
n
i
m
pr
ov
e
fitness
of
G
A
final
s
ol
ution
,
al
thou
gh
it
m
a
y
diff
ic
ult
to
ach
ie
ve
beca
us
e
gl
ob
al
ly
GA
ha
ve
achie
ved
hi
gh
e
r
fitne
ss
s
o
if
process
on
VNS
can
’t
gi
ve
bette
r
so
luti
on
t
hen
VNS
ca
n
def
e
nd
fi
nal
so
l
ution
of
G
A
that
hav
e
hi
gh
fitne
ss,
the
exam
ples
is
on
7
th
e
xp
erim
ent
and 8
th
e
xperi
m
ent.
3.
3
.
GA
-
V
NS
II
The
seco
nd
ex
per
im
ent
on
this
research
c
om
bin
es
VN
S
on
G
A’
s
re
pro
duct
ion
a
nd
it
s
proce
dure
is
descr
i
bed
i
n
se
ct
ion
2.5
.2.
Th
e
aim
s
of
GA
-
VNS
I
I
is
rep
a
ir
div
e
rsity
on
so
luti
on
th
rou
gh
re
pro
du
ct
io
n,
they
are cross
over
a
nd m
utati
on
. T
he patt
ern o
f h
ybrid
GA
-
V
NS II’s
fitness
is s
how
n
in
Fig
ure
6
.
Figure
6
s
how
s
co
nver
gen
ce
is
ha
pp
e
ne
d
on
40
th
ge
ner
at
i
on
an
d
61
st
ge
ner
at
io
n.
T
he
conve
rg
e
nce
on
40
th
ge
ne
rati
on
can
be
so
l
ved
with
us
e
VNS.
It
can
ha
pp
e
n
bec
ause
VN
S
is
r
un
on
G
A
re
produ
ct
ion
on
50
th
ge
ne
rati
on the
refor
e it
can
giv
e a n
e
w
s
olu
ti
on, if
ne
w
so
luit
on f
r
om
VNS is b
et
te
r
t
han
pr
e
vious s
olu
ti
on
from
GA
,
V
N
S
will
be
us
ed
as
one
of
the
s
olu
ti
on
on
GA
popula
ti
on
the
n
is
hope
d
it
c
an
giv
e
m
or
e
di
ver
se
so
luti
on
on
G
A
popula
ti
on.
That
is
can
gi
ve
infl
uen
ce
on
n
e
xt
gen
e
rat
ion
an
d
the
n
c
hild
from
GA
m
or
e
var
ia
te
the
refo
r
e co
nv
e
r
gen
ce
can
be
a
vo
i
de
d.
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.
8
, N
o.
6
,
Dece
m
ber
2
01
8
:
4673
-
4683
4680
Figure
6. G
raph
of f
it
ness
p
at
te
rn
on
GA VNS
II
The
seco
nd
c
onve
r
gen
ce
is
ha
pp
e
ne
d
on
61
st
gen
e
rati
on.
The
c
onverge
nce
was
sti
ll
un
a
voida
ble
betwee
n
61
st
gen
e
rati
on
to
100
th
gen
e
rati
on
beca
us
e
th
ere
isn
’t
othe
r
facto
r
that
c
an
gi
ve
m
or
e
var
ie
d
so
luti
on.
T
he
conve
rg
e
nce
c
an
be
ove
rcom
e
in
the
10
0
th
gen
e
rati
on
du
e
to
fact
or
of
us
e
VNS
on
G
A
reprod
uction.
More
ov
e
r,
on
ever
y
offsprin
g
on
c
rosso
ve
r
and
m
utati
on
is
i
m
pr
oved
by
VN
S
i
n
orde
r
to
ac
hieve
bette
r
fitness
.
I
m
pr
ov
em
ent
with
pure
V
NS
is
done
on
50th
gen
e
rati
on
a
nd
it
s
m
ulti
ple
i
n
orde
r
to
doe
s
no
t
ta
ke
m
uch
lo
nger
c
om
pu
ta
tio
nal
ti
m
e.
I
m
p
rovem
ent
on
cro
ss
over
a
nd
m
uta
ti
on
is
show
n
on
Fig
ure
7
and
Figure
8.
Figure
7. G
raph
of f
it
ness
p
at
te
rn
on Cr
os
s
over
Figure
8. G
raph
of f
it
ness
p
at
te
rn
on Mutat
i
on
Fr
om
Figure
7
an
d
Fi
gure
8
is
sho
wn
50
th
and
100
th
ge
ner
at
io
n
on
c
ro
ss
over
a
nd
m
uta
ti
on
are
happe
ned
im
pr
ov
em
ent
of
fit
ness
be
f
or
e
an
d
after
V
NS
is
com
bin
ed.
W
i
th
new
s
olu
ti
on
from
VN
S
will
be
evaluate
d
betw
een
so
l
utio
ns
befor
e
V
NS
is
com
bin
ed
(
off
sp
ri
ng
of
G
A)
and
a
fter
VNS
is
com
bin
ed.
Wh
e
n
so
luti
on
from
VNS
is
bette
r
t
han
s
olu
ti
on
be
fore
VNS
is
c
om
bin
ed,
t
he
n
VNS
s
olu
ti
on
will
rep
la
ce
sol
ution
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
Hyb
ri
d Ge
netic
Alg
or
it
hm fo
r Opti
miza
ti
on
of F
ood
C
omp
os
it
ion o
n Hy
pe
rte
ns
iv
e
... (
A
pr
il
ia Nu
r F
auziy
ah
)
4681
from
of
fs
pr
i
ng
on GA.
From
t
he gra
ph, VN
S
can give
im
pr
ov
em
ent of f
it
ness
a
nd m
or
e
var
ie
ty
so
l
utio
n.
T
hat
is im
po
rtant
to need
whe
n
c
onverge
nce is
ha
pp
e
ne
d on pr
oc
ess of
GA.
3.
4
.
V
NS
-
GA
The
la
st
m
et
ho
d
pr
opos
e
d
on
this
resea
rch
is
hy
br
idiza
ti
on
V
NS
-
G
A.
This
m
et
ho
d
pro
posed
fi
nal
so
luti
on
of
V
N
S
is
us
ed
as
ini
ti
al
so
luti
on
of
GA
.
T
his
m
eth
od
ai
m
to
i
m
pro
ve
qual
it
y
of
VNS
final
s
ol
ution
.
Be
cause
G
A
is
popu
la
ti
on
ba
sed
an
d
V
NS
is
sing
le
so
lu
ti
on
based
the
n
VNS
will
be
ru
n
n
tim
es
ru
n
(
n
is
popula
ti
on
siz
e)
f
or
form
ing
init
ia
l
so
luti
on
of
G
A.
S
o
t
hat
the
i
dea
of
hybr
i
dizat
ion
V
N
S
G
A
on
t
his
s
essi
on
is
fo
rm
ing
init
ia
l
so
luti
on
fro
m
VN
S
with
a
nu
m
ber
of
po
pu
la
ti
on
siz
e
and
t
hen
it
s
so
l
utions
wil
l
be
us
e
d
as
GA
i
niti
al
so
lu
ti
on
.
Howe
ver
this
le
ads
to
l
onge
r
c
om
pu
ta
tio
nal
ti
m
e
since
VNS
has
to
be
processe
d
as
m
uch
as
popula
ti
on
s
iz
e
to
pro
vid
e
t
he
so
l
utio
n
nee
ded
by
G
A
as
the
init
ia
l
so
luti
on
.
O
n
the
te
s
ti
ng
of
VN
S
-
G
A
is
done
with
10
tim
es
ru
n
to
m
easur
e
perf
orm
ance
of
V
N
S
-
G
A.
Fig
ure
9
s
how
the
c
om
par
ison
befo
re
GA
app
li
ed
and
aft
er
GA
ap
plied
on VNS
so
l
ution.
Figure
9. G
raph
of com
par
iso
n betwee
n pure
VNS a
nd
VNS
-
GA
Fr
om
the
Figu
re
9,
with
10
tim
es
ru
n
VNS
giv
e
a
ve
rag
e
of
fitne
ss
is
0,
5115
92
an
d
after
i
m
pr
ovem
ent
i
s
done
by
G
A,
it
can
giv
e
sol
ution
s
with
av
erag
e
of
fitnes
s
is
0,
8748
28,
that
is
pr
oven
VNS
that give
opti
m
al
so
luti
on i
n
l
ocal area
can
be im
pr
ove
d on
searcin
g
in
g
l
obal
area
u
si
ng
GA.
Hybr
i
d
GA
-
V
NS
m
et
ho
d
is
pro
po
se
d
ca
n
giv
e
foo
d
c
ompo
sit
io
n
reco
m
m
end
at
ion
is
m
or
e
op
ti
m
a
l
and
a
pprop
riat
e
with
patie
nt
need
e
d.
The
com
par
ison
of
fitnes
s
and
ru
n
ti
m
e
fr
om
any
scenario
of
hybri
dizat
ion o
n GA an
d V
NS i
s sho
wn b
y
T
able 2.
Tabel
2.
T
he
Com
par
ison o
f Hybri
dizat
ion
GA VNS
, pure
GA a
nd pu
r
e
VNS
Metho
d
Av
erage of
f
itn
ess
Av
erage of
run
ti
m
e
(
m
s
)
VNS
0
,58
3
2
0
6
1
8
8
9
,2
GA
0
,81
7
1
0
6
1
8
9
0
,3
GA
-
V
NS I
0
,86
3
2
5
4
3
3
0
6
,2
GA
-
V
NS I
I
0
,88
9
7
7
6
9
8
4
9
4
,3
VNS
-
GA
0
,87
4
8
2
8
9
8
7
4
3
,3
Gen
e
rall
y,
ave
rag
e
fitness
of
GA
is
higher
t
han
a
ve
rag
e
fitness
of
VNS
on
ou
r
previ
ous
ly
researc
h
[4
]
.
From
Tab
le
2
is
sho
wn
hybri
dizat
ion
betwee
n
G
A
a
nd
V
NS
need
m
or
e
com
pu
ta
ti
on
al
ti
m
e
than
ti
m
e
require
d
on
pu
re
V
NS
or
pur
e
GA
howe
ver
hybr
i
dizat
ion
GA
a
n
d
VNS
giv
e
be
tt
er
qua
li
ty
of
so
luti
on
and
fitness.
T
his
is
because
hybri
di
zat
ion
us
e
m
o
re
than
one
al
gorithm
tho
se
ha
s
adv
a
ntages
on
local
ex
plo
i
ta
ti
on
and
gl
obal
ex
pl
or
at
io
n
s
o
t
ha
t
in
it
s
sea
rch
process
nee
d
l
onge
r
c
om
pu
ta
ti
on
al
ti
m
e
bu
t
sti
ll
in
reas
on
a
ble
tim
e.
Be
tween
the
s
cenari
o
hy
br
id
iz
at
ion
G
A
an
d
V
NS,
G
A
-
V
NS
I
I
gi
ve
hi
gher
fitness
value
tha
n
ot
he
r
scenari
o.
T
hat
is
becau
se
GA
-
V
NS
II
has
ad
va
ntage
on
re
pair
s
ol
ution
a
nd
im
pro
ve
fitness
wh
e
n
konver
ge
nce
i
s
ha
pp
e
ne
d
on
m
ulti
ples
of
50th
generati
on.
Wh
e
reas
oth
e
r
sce
nar
i
o
su
c
h
as
G
A
-
VNS
I
,
functi
on
of
V
NS
is
only
i
m
pro
ving
agai
n
final
so
l
ution
of
GA
s
o
that
GA
VNS
I
m
a
y
le
ss
con
trib
ut
e
whe
n
conve
rg
e
nce
is
occ
ur
e
d
on
pr
ocess
of
G
A.
Gen
e
rall
y,
G
A
VNS
I
,
VNS
s
uccesf
ully
i
m
pr
ove
final
s
olu
t
ion
of
GA ho
wev
e
r w
hen GA
ha
d ha
ve
s
olu
ti
on
with
high
fitness,
VNS
of
te
n
diffi
cult ac
hieve
bet
te
r
so
luti
on.
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.
8
, N
o.
6
,
Dece
m
ber
2
01
8
:
4673
-
4683
4682
Anothe
r
sce
na
rio
is
V
NS
-
G
A.
O
n
this
ses
sion,
G
A
im
prov
e
s
s
olu
ti
on
f
ro
m
final
s
olu
t
ion
of
VN
S
.
The
c
om
pu
ta
ti
on
al
ti
m
e
req
ui
red
al
m
os
t
sam
e
with
oth
e
r
hybr
i
dizat
ion
sce
nar
i
o
be
cause
VNS
m
us
t
be
r
un
as
m
uch
as
popul
at
ion
siz
e.
Tha
t
is
because
pu
re
V
NS
gi
ve
si
ng
le
s
olu
ti
on
s
o
that
i
n
or
der
to
visible
is
use
d
as
init
ia
l
so
luti
on
of
GA.
In
this
scena
rio
,
VNS
can
help
G
A
has
init
ia
l
po
pu
la
ti
on
t
hat
not
to
o
ba
d
tha
n
inti
al
so
luti
on
of GA
that is
gen
e
rat
ed ran
dom
l
y.
The
resu
lt
of
pro
posed
m
et
ho
d
is
c
om
par
ed
with
our
pr
e
vi
ou
s
resea
rc
h
[
4]
w
hich
im
ple
m
ents
pure
VNS
on
com
po
sit
io
n
f
ood.
The
pro
pose
d
m
et
ho
d
is
te
ste
d
with
im
ple
m
ent
pr
oble
m
case
to
search
com
po
sit
ion
f
ood
for
hy
pe
rte
ns
ive p
at
ie
nt.
The
pr
ob
le
m
c
ase
us
e
d
is
sam
e
with
prob
le
m
case
on
our
pr
e
vious
researc
h
[
4]
so
that
the
resu
lt
between
our
pr
e
vious
resea
rch
a
nd
m
et
ho
d
pro
posed
on
this
research
can
be
com
par
ed.
The
res
ult
with
usi
ng
pure
V
NS
on
our
pre
viou
s
resea
rch
an
d
pro
posed
m
eth
od
on
this
res
earch
are s
how
n on
Table
3.
Tabel
3.
T
he
d
i
ff
e
ren
ce
r
es
ult
betwee
n pure
VNS a
nd h
y
br
i
d GA wit
h V
N
S
Prop
o
sed
m
e
th
o
d
Dif
f
erence (
%)
So
d
iu
m
(
m
ax
.
0,8
gra
m
)
Calo
ri
Carb
o
h
id
rat
Fat
Protein
VNS
(ou
r
p
revio
u
s
resear
ch
)
0
,27
1
2
5
1
6
,27
2
3
5
7
,27
2
3
7
1
6
,00
7
7
4
8
0
,17
6
2
GA VN
S I
0
,11
4
1
3
8
5
1
2
,07
3
5
9
8
1
3
,03
6
6
1
2
1
6
4
,04
6
5
5
4
2
0
,
3
3
9
1
6
6
9
GA VN
S I
I
0
,54
5
2
3
8
4
7
0
,95
8
6
9
7
5
8
1
,45
9
2
5
3
5
,07
4
6
4
4
0
,34
2
1
6
6
9
VNS G
A
0
,
0
3
5
7
5
6
7
1
,
1133
1
,
5
5
9
9
1
6
,
3
3
8
5
1
0
,
3
6
9
7
6
4
Table
3
e
xp
la
i
ns
that
the
m
e
thod
of
pure
VNS
in
our
previo
us
resea
rc
h
act
ually
has
giv
e
n
res
ults
that
no
t
ba
d
be
cause
the
di
fference
betwee
n
the
res
ults
of
optim
iz
ation
with
total
nu
tr
ie
nts
of
the
re
qu
i
re
d
patie
nt
does
not
excee
d
10
%
.
Alth
ough
the
a
m
ou
nt
of
sodium
fr
om
pu
r
e
VNS
is
le
ss
than
hybri
d
G
A
a
nd
VNS,
it
is
no
t
ver
y
influe
nti
al
becau
se
it
i
s
within
reas
onable
li
m
it
s
w
hich
is
for
thi
s
case
prob
le
m
the
m
axi
m
u
m
a
m
o
un
t
of
sodium
m
us
t
be
in
f
ood
is
0,8
gram
s.
The
res
ult
of
pro
po
se
d
m
et
ho
d
in
this
stu
dy
giv
es
m
or
e
interest
ing
res
ult,
that
is
diff
e
ren
c
e
of
nu
t
rient
am
ou
nt
between
res
ul
t
of
hybri
d
opti
m
iz
at
ion
of
G
A
an
d
VNS
with
am
ount
of
nu
t
rients
needed
by
patie
nt
le
ss
(in
per
ce
nt).
F
or
bette
r
optim
izati
on
res
ults
can
us
e
m
or
e
fo
od
dat
a
for
m
or
e
so
luti
on
var
ia
ti
on
s
and
t
he
am
o
un
t
of
nu
trie
nts
produce
d
m
or
e
cl
os
el
y
to
th
e
needs
of the
patie
nt'
s n
ut
riti
on
.
4.
CONCL
US
I
O
N
The
pro
po
s
ed
m
et
ho
d
i
n
this
stud
y
s
uccess
f
ully
so
lve
opti
m
iz
at
ion
of
f
ood
com
po
sit
io
n
prob
le
m
for
hype
rtensive
pa
ti
ent.
Althou
gh
nee
d
lo
ng
e
r
com
pu
ta
ti
onal
tim
e
ho
we
ve
r
hybri
dizat
io
n
G
A
a
nd
VNS
gi
ve
bette
r
re
su
lt
t
han
pu
re
G
A
or
pure
V
NS.
V
NS
s
ucces
f
ully
help
G
A
avo
i
ds
pr
em
atu
re
co
nver
gence
an
d
i
m
pr
oves
bette
r
s
olu
ti
on.
T
he
shortc
om
ing
s
on
GA
i
n
l
ocal
ex
plo
it
at
ion
an
d
easy
t
rappe
d
in
pre
m
at
ur
e
conve
rg
e
nce
is
su
ccess
f
ully
so
lve
d
by
V
NS
,
wherea
s
t
he
s
hortcom
ing
on
VNS
that
le
ss
capa
bili
ty
in
global
exp
l
or
at
io
n
ca
n be s
olv
e
d by
us
e
GA that
ha
s adva
ntage
i
n glo
bal expl
or
at
ion
.
REFERE
NCE
S
[1]
W
HO
,
“
Noncommunica
ble Diseas
es
in
th
e
South
-
Ea
st Asia
Reg
io
n
”
,
India:
WHO
Press,
2011.
[2]
W
CF
R
(
W
orld
Canc
er
Rese
arc
h
Fund
Inte
rna
ti
o
nal
)
&
NCD
Alli
ances,
“
The
li
n
k
bet
wee
n
food
nutri
ti
on
,
diet
an
d
non
-
comm
unic
abl
e
d
isea
ses
”
, 2
nd
edition
O
ct
obe
r
2014.
Availabl
e
a
t:
htt
p:
//
ww
w.
wcrf.
org
[3]
Am
eri
ca
n
Hea
r
t
As
socia
ti
on
,
“
Understa
nding
and
Mana
gin
g
High
Blood
Press
ure
”,
20
14.
Availabl
e
at:
htt
p://ww
w.he
art.org
.
[4]
Fauziy
ah
,
A.N
&
Mahm
ud
y
,
W
.
F,
“
Optimiza
t
ion
of
Food
Co
m
positi
on
for
Hy
per
te
nsiv
e
Pat
i
ent
using
Va
riable
Neighbor
hood
Sear
ch
”,
Indon
esi
a
Journa
l
of
Ele
ct
rical
Engi
n
ee
r
ing
and
Computer
Sci
en
ce
(
IJEECS)
,
vol.
8,
pp
.
327
-
334,
Nov 17
.
[5]
Iwuji
AC,
Nnanna
M,
Ndulue
NI
C,
“
An
Opti
ma
l
DA
SH
Diet
Model
for
People
with
H
y
per
te
n
sion
Us
ing
Li
near
Program
m
ing
A
pproa
ch”,
Open
Jurnal
of
Optimi
zation
,
vo
l. 5, pp
.
1
4
-
21
,
2016
.
[6]
Mahm
udy
W
.
F,
Maria
n
R
.
M
,
Luong
L
.
H
.
S
,
“
Rea
l
code
d
g
ene
t
ic
al
gor
it
hm
s
for
solving
fl
exi
bl
e
job
-
shop
sche
duli
ng
probl
em
–
Part
II:
opt
imiza
ti
on
”
,
Ad
va
nce
d
Ma
te
rials
Re
search
,
vol
.
7
01,
pp
.
364
-
369
,
2013.
[7]
W
ij
a
y
ani
ngrum
V
.
N
,
Mahm
udy
W
.
F
,
“
Optimization
of
Ship’
s
Route
Schedu
li
ng
Us
ing
Gene
tic
Algorit
hm
”
,
Indone
sian J
our
nal
of
Elec
tric
al
Engi
ne
ering
and
Computer
Sc
ie
n
ce
(
IJEECS)
,
vol
.
2
,
pp
.
180
-
186,
2016.
[8]
O.
Dib,
L
.
Moa
li
c
,
M.
-
A.
Mani
er,
A.
C
aminad
a,
“
An
adva
nc
e
d
GA
–
VNS
combinat
ion
for
m
ult
ic
riteri
a
rou
te
pla
nning
in
pub
l
ic
tra
nsit
ne
tworks”,
E
xpe
rt
Syst
e
ms
wit
h
App
li
ca
t
ions
,
vol
.
72
,
pp
.
67
–
82,
2017.
Evaluation Warning : The document was created with Spire.PDF for Python.