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.
10
,
No.
4
,
A
ugus
t
2020
,
pp.
3358
~
33
66
IS
S
N: 20
88
-
8708
,
DOI: 10
.11
591/
ijece
.
v10
i
4
.
pp3358
-
33
66
3358
Journ
al h
om
e
page
:
http:
//
ij
ece.i
aesc
or
e.c
om/i
nd
ex
.ph
p/IJ
ECE
Hybrid
m
ethod for a
chievin
g Pa
ret
o front on
econ
omic
emission
dispatch
K. Rajesh
, N.
Visa
li
Depa
rtment
o
f
E
le
c
tri
c
al a
nd
Ele
ct
roni
cs
Engi
n
eering,
JN
TUA
Co
ll
eg
e
of
Engi
n
eering,
Ind
ia
Art
ic
le
In
f
o
ABSTR
A
CT
Art
ic
le
history:
Re
cei
ved
Ma
r 18,
2019
Re
vis
ed Feb
4,
2020
Accepte
d
Fe
b 12,
2020
In
thi
s
pape
r
h
y
brid
m
et
hod
,
Modifie
d
Nondom
ina
te
d
Sort
ed
Gene
tic
Algorit
hm
(MN
SG
A
-
II)
and
Modifie
d
Populat
ion
Var
ia
n
t
Diffe
ren
t
ia
l
Evol
uti
on(MP
VD
E)
have
b
ee
n
pla
c
ed
in
eff
e
ct
in
ac
h
ie
ving
the
best
opti
m
a
l
soluti
on
of
Mul
ti
objecti
v
e
e
con
om
ic
emiss
ion
loa
d
dispatch
o
pti
m
iz
ation
proble
m
.
In
thi
s
te
chn
ique
latte
r
,
one
is
used
to
enf
orc
e
the
assig
ned
per
ce
n
t
of
the
popul
at
i
on
and
the
r
emaini
ng
with
the
form
er
one.
T
o
over
com
e
the
pre
m
at
ur
e
c
onver
gence
in
a
n
opti
m
iz
a
ti
on
p
r
oble
m
dive
rsit
y
p
rese
rving
oper
at
or
is
emplo
y
ed,
from
the
tra
deof
f
cur
v
e
the
best
opti
m
al
soluti
on
is
pre
dicted
using f
uzzy
set
th
eor
y
.
Thi
s m
et
hodolog
y
va
li
da
te
d
on
I
EE
E
30
bus
te
st
s
y
stem
wit
h
six
gene
rat
ors
,
IEE
E
118
bus
te
st
s
y
stem
with
fourte
en
gene
ra
tors
and
with
a
fort
y
gene
ra
tors
te
st
s
y
stem.
The
so
lut
ions
are
dissim
il
it
ude
wi
th
the
exi
sting
m
et
ahe
urist
ic
m
et
hods
li
ke
Stre
ngth
Pare
to
Evol
uti
on
ar
y
Algorit
hm
-
II,
Multi
obj
ec
t
ive
diff
ere
nt
ia
l
evol
ut
i
on,
Multi
-
obje
c
ti
ve
Partic
le
Sw
arm
optim
iz
at
ion
,
Fuz
z
y
cl
ust
eri
ng
p
article
sw
arm
opti
m
iz
ation,
No
ndom
ina
te
d
sort
i
ng
genetic algori
thm
-
II.
Ke
yw
or
d
s
:
Cost f
unct
ion
Eco
no
m
ic
load dis
patch
Mult
iobject
ive
functi
on
Trad
e
off
c
urve
Copyright
©
202
0
Instit
ut
e
o
f Ad
vanc
ed
Engi
n
ee
r
ing
and
S
cienc
e
.
Al
l
rights re
ser
v
ed
.
Corres
pond
in
g
Aut
h
or
:
K.
Ra
j
es
h,
EEE
Dep
a
rtm
e
nt,
JNT
UA Coll
eg
e of E
ng
i
neer
i
ng,
An
a
nth
a
puram
u,
A
ndhra P
radesh,
Ind
ia
.
Em
a
il
:
20
16ra
je
sh75@gm
ai
l.
com
1.
INTROD
U
CTION
In
a
c
omm
on
par
la
nce
of
power
syst
em
s
m
ul
ti
ob
j
ect
ive
econom
ic
e
m
issi
on
loa
d
dis
patch
i
s
a
n
op
ti
m
iz
ation
pro
blem
wh
ic
h
con
cer
ns
wit
h
incom
patible
obj
ect
ives
li
ke
cost
an
d
em
issi
on
.
Be
st
powe
r
gen
e
rati
on
wit
h
m
ini
m
u
m
e
m
issi
on
is
the
desire
d
ta
sk
w
hich
le
ad
s
to
t
he
opti
m
iz
at
ion
pro
blem
wit
h
m
any
const
raints.
P
ower
ge
ner
at
i
on
with
opti
m
u
m
incre
m
ental
f
uel
co
st
can
be
com
patible
with
the
al
loca
ti
on
of
gen
e
rato
rs
i
n
t
he
the
rm
al
plant.
T
he
vast
use
of
c
om
bu
sti
bl
e
fo
s
sil
f
uels
i
n
the
rm
al
plan
ts
pro
du
ce
s
ha
rm
fu
l
gases
li
ke
CO
,NO
2
,S
O
2
le
ad
ing
t
o
c
on
ta
m
inati
on
of
at
m
os
phe
re
wh
ic
h
is
a
m
ajo
r
is
su
e
from
a
glo
ba
l
war
m
ing
per
s
pe
ct
ive.
W
it
h
t
he
awa
ren
e
ss
of
gase
ous
poll
utant,
t
he
Un
it
ed
Stat
es
has
pas
sed
a
Cl
ean
Ai
r
Act
Am
end
m
ents
of
1990
[1
]
ha
s
m
ade
enf
orc
e
m
ent
to
al
l
th
e
powe
r
plants
to
f
ollo
w
the
strat
egies
to
m
ini
m
iz
e
the
e
m
issi
on
s
fr
om
the
the
rm
al
po
we
r
pl
ant.
The
inco
rpor
at
io
n
of
em
issi
on
obj
ect
ive
with
Eco
nom
ic
Disp
at
c
h(
E
D)
pro
blem
has
be
com
e
a
m
ul
tio
bject
ive
with
it
s
co
m
plexity
.
Ma
ny
te
chni
qu
es
wer
e
de
velo
ped
for
the
s
olu
ti
on
of
m
ulti
ob
j
e
ct
ive
nonlinea
r
pro
blem
s
on
e
of
t
he
possi
ble
thin
gs
to
c
onve
rt
m
ult
iob
je
ct
ive
to
m
on
o
obj
ect
i
ve
.
Linea
r
Pro
gra
m
m
ing
te
ch
ni
qu
e [
2]
consi
de
red
as
a
sin
gle
obj
ect
ive w
it
h
a
lot of
ass
um
pt
ion
s
and
does
n’
t
pr
ov
i
de
the
easy
so
luti
on.
A
no
t
her
a
ppr
oach
[
3]
co
nv
e
rsion
of
m
ulti
ob
j
ect
i
ve
to
m
on
o
ob
j
ect
ive
is
i
m
ple
m
ented
by
us
in
g
su
it
able
weig
hts
and
dr
a
wb
ac
ks
associat
ed
wit
h
this
on
e
it
req
ui
res
m
ult
iple
ru
ns
to
achieve
the
tra
de
-
off
cu
rv
e
,
to
com
pen
sat
e
this
ε
con
strai
nt
m
et
ho
d
was
dev
el
ope
d
[
4]
con
si
der
in
g
a
no
t
he
r
obj
ect
ive
as
a
const
raint
with
it
s
l
i
m
i
ta
ti
on
of
ε
le
vels,
bu
t
t
his
is
a
tim
e
consum
ing
with
the
proc
ess
of
wea
k
non
dom
inated
so
luti
on
s
.
Lat
er
a
novel
fu
zzy
op
ti
m
iz
ation
te
ch
nique
[5
]
was
ap
plie
d
but
du
e
to
l
ack
of
com
po
sit
ion
i
n
di
recti
ng
t
he
searc
h
to
wa
rd
s
t
he
tra
de
-
off
cu
r
ve
w
hi
ch
is
not
pre
f
err
e
d
one
.
A
fu
zzy
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 meth
od
for ac
h
ie
vi
ng
Pareto
fro
nt
on ec
onomic
em
issi
on
dis
pa
tc
h
(
K. Ra
je
s
h
)
3359
sat
i
sfacti
on
m
axim
iz
ing
dec
isi
on
m
aking
[6
]
ap
proac
h
i
m
ple
m
ented
sat
isfact
or
il
y
bu
t
the
exte
nsi
on
of
obj
ect
ives
has
becam
e
a
te
dio
us
proces
s.
A
no
vel
ap
pro
ach
intr
oduc
ed
in
[
7]
w
hich
is
a
form
idable
a
nd
exh
i
biti
ng
unti
m
el
y con
ve
rg
e
nce c
har
act
e
risti
c.
Re
cent
d
e
velo
pm
ent
te
nd
s
towa
r
ds
E
volut
ion
a
ry
Al
gorithm
s
(EA
)
f
or
so
l
ving
EE
L
D
pro
blem
s
com
pen
sat
ing
the
dow
ns
ide
of
the
cl
assic
a
l
m
et
ho
ds
to
ob
ta
in
Pa
reto
set
.
These
al
gorithm
s
are
based
on
the
popula
ti
on
wh
ic
h
ge
ner
at
e
s
opti
m
a
l
so
lut
ion
on
a
sin
gle
r
un
ha
ving
i
ts
own
m
erit
s.
N
ondom
inate
d
so
rtin
g
Gen
et
ic
Algori
thm
[8
]
is
app
l
ie
d
w
hich
ex
hi
bited
the
s
uboptim
al
so
luti
on
of
res
ult
an
d
c
on
s
um
ing
m
or
e
tim
e
for
the
e
valuat
ion
.
N
SGA
-
I
I
[9
]
e
xh
i
bits
el
it
is
m
,
rankin
g
bu
t
fail
s
in
it
s
un
i
form
i
ty
of
t
he
tra
de
-
off
,
a
nd
this
can
be
c
om
pen
sat
ed
us
i
ng
dy
nam
ic
crowd
i
ng
distanc
e
an
d
m
or
e
MOE
As
a
re
Stre
ngth
Pa
reto
Ev
olut
ion
ar
y
Algorithm
[10
]
,
Niche
d
par
e
to
ge
netic
al
gorithm
[11],
Mult
iobject
ive
Partic
le
swa
r
m
op
tim
iz
at
io
n
[
12]
,
Mult
iobject
ive
θ
-
Partic
le
swa
rm
op
tim
iz
a
ti
on
[
13
]
,
M
ulti
ob
j
ect
iv
e
Dif
fere
ntial
Evo
l
utio
n
[
14
]
,
sem
idefinite
Pr
og
ram
m
ing
[15],
sel
f
ada
ptive
le
arn
i
ng
bat
al
go
rithm
[16],
Mult
iobject
ive
Gl
ob
al
Be
st
Ar
ti
fici
al
Be
e
Colo
ny
[
17]
,
Gr
ey
Wo
l
f
O
pt
i
m
iz
at
ion
[18
]
.
In
a
n
M
OE
A
a
set
of
so
l
utions
a
re
ob
t
ai
ned
,
wh
ic
h
util
iz
es
the
fu
zzy
set
theor
y
to
ac
hieve
the
best
com
pr
om
ise
so
luti
on
[
19]
.
In
this
pa
pe
r
,
two
obj
ect
iv
es
are
consi
der
e
d
si
m
ul
ta
neo
usl
y
without
co
nve
rting
t
o
a
sin
gle
obj
ect
i
ve
and
la
yo
uted
with
m
od
ific
at
ion
of
NSGA
-
I
I
a
nd
PVDE.
Com
bin
ing
tw
o
or
m
or
e
al
gorithm
s
are
re
ferred
t
o
as
hy
br
idiza
ti
on
w
hich
is
a
su
cc
essfu
l
one
i
n
so
lvi
ng
en
vir
onm
ental
econo
m
ic
load
disp
a
tc
h
pr
ob
le
m
.
The
m
ai
n
ta
sk
of
the
hy
br
i
d
m
et
hod
is
t
o
pro
vid
e
div
e
rsified
Par
et
o
f
ront.
In
t
his
pa
pe
r
a
hy
br
i
d
m
et
ho
d
w
her
e
us
e
d
with
the
c
om
bin
at
i
on
of
MP
VDE
an
d
MNSGA
-
I
I
is
e
m
plo
ye
d
w
hich
is
at
the
init
ia
l
sta
ge
half
of
the
popula
ti
on
is
ap
plied
wi
th
MN
SGA
-
II
wh
e
r
e
non
-
dom
inate
d
sorti
ng
is
im
plem
ented
an
d
the
rem
ai
nin
g
half
of
t
he
popula
ti
on
is
c
arr
ie
d
with
refreshe
d
popula
ti
on
c
oncept
of
sem
i
-
i
nter
qu
a
rtil
e
ran
ge
a
nd
m
idq
uar
ti
le
range
t
o
el
i
m
inate
i
mm
ov
able
local
op
ti
m
a
us
in
g
MPV
DE
and
finall
y,
Eucli
dean
dista
nce
is
evaluate
d
f
or
the
ne
xt
gen
e
rati
on
of
the
po
pu
la
ti
on.
This
hybri
d
m
et
ho
d
is
te
ste
d
on
I
EEE
30
bus
syst
e
m
with
six
gen
e
rato
rs
a
nd
IEEE
118
bus
syst
e
m
with
fo
urt
ee
n
gen
e
rato
rs
a
nd
40
gen
e
rato
rs
syst
e
m
.
The
rem
ai
nd
er
of
t
he
pap
e
r
is
organ
iz
e
d
as
f
ollow
s:
in
sect
ion
2,
the
m
at
he
m
a
tical
fo
rm
ulati
on
of
EEL
D,
in
sect
io
n
3
on
m
ulti
ob
j
ect
iv
e
o
ptim
iz
ation
and
i
n
sect
io
n
4
on
the
br
ie
f
desc
ription
on
MNS
GA
-
I
I
an
d
MP
VD
E
a
nd
hybr
id
m
et
ho
d
is
presente
d
in
sec
ti
on
5
res
ults
of
three
te
st case
s sho
w
ing
bette
r pe
rfor
m
ance in
c
om
par
ison
with
existi
ng m
et
ho
ds
is
presente
d.
2.
PROBLE
M
F
ORMUL
ATI
ON
The
c
om
b
ined
eco
no
m
ic
e
m
i
ssion
dis
patch
is
inco
rporat
ed
with
t
wo
co
nt
rad
ic
to
ry
ob
j
ec
ti
ves,
c
os
t
and
em
issi
on
with
co
ns
trai
nts.
The
quad
rati
c
equ
at
io
n
relat
ed
to
the
c
os
t
functi
on
of
ev
ery
ind
i
vidual
pow
e
r
gen
e
rati
on
un
it
is g
i
ven as
=
∑
(
)
=
1
(1)
(
)
=
(
+
+
2
)
(2)
wh
e
re
F
k
(P
Gk
)
is
k
th
gen
e
rato
r
cost
functi
on.
The
a
k
,
b
k
,
a
nd
c
k
are
re
ferr
ed
to
k
th
un
it
f
uel
cost
coe
ff
i
ci
ents.
n
is
the
no.
of
gen
e
rato
rs.
is
the
total
fu
el
cost
in
doll
a
r
s
per
hour.
Th
e
belo
w
(
3)
is
a
com
bin
at
ion
of
qu
a
drat
ic
and e
xponentia
l f
un
ct
ion
s
represe
nt
s the em
issi
on
levels
of
w
hich
al
l ha
rm
fu
l t
ox
ic
gases
.
(
)
=
∑
(
(
+
+
2
)
+
exp
(
)
=
1
)
(3)
wh
e
re α
k
, β
k
, γ
k
,
,
are the
em
issi
on co
e
ff
ic
ie
nt
o
f
k
th
unit
. T
he
constraints
ar
e ex
plaine
d
in
the
fo
ll
owin
g.
2.1.
P
ower
b
al
an
ce eq
uatio
ns
The Po
wer bal
ance e
qu
at
i
on is re
pr
ese
nted
in (4
)
∑
=
+
=
1
(4)
her
e
,
P
Gk
,
P
D
,
and
P
LOSS
are
r
epr
ese
nted
t
he
real
power
outpu
t
at
k
th
ge
ne
rator,
total
la
od
dem
and
an
d
powe
r
losses i
n
tra
nsm
issi
on
li
ne re
sp
ect
ively
. T
he
pow
e
r
l
os
s is
appr
ox
im
at
ely cal
culat
ed usin
g (5).
=
∑
∑
+
∑
+
=
1
=
1
=
1
(5)
wh
e
re,
B
lk
, B
k
o
, B
00
are
powe
r
loss c
oeffici
ent
s or B c
oeffici
ents.
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
.
4
,
A
ugus
t
2020
:
3358
-
3366
3360
2.2.
Power
c
apacit
y
li
mi
ts
c
on
s
tr
aint
s
≤
≤
(6)
wh
e
re
P
Gkm
ax
a
nd P
Gkm
in
are th
e
m
axi
m
u
m
an
d
m
ini
m
u
m
v
al
ues of o
utput
powe
r
at
t
he k
th
gen
e
rato
r.
2.
3.
Va
l
ve
p
oin
t
ef
fect o
f
c
os
t
f
u
nctio
n
W
it
h
th
e
inc
orporati
on
of
the
se
ef
fects
in
th
e
li
near
pro
ble
m
of
optim
iz
ation
is
tra
ns
fi
gured
to
a
no
n
conve
x
pro
blem
with
cost
f
unct
ion
incl
ud
i
ng
t
he
ri
pp
le
cu
rv
e
.
T
he
m
od
i
fied
quad
rati
c
equ
at
io
n
with
these
eff
ect
s is
re
pr
e
sented
in (7
)
.
(
)
=
+
+
2
+
|
(
sin
(
(
−
)
)
)
|
(7)
3.
EVOLUTI
ONAR
Y
AL
GO
R
ITH
MS
Ev
olu
ti
onary
a
lgorit
hm
s
ov
erco
m
e
the
dr
aw
backs
associat
ed
with
cl
assic
al
m
e
tho
ds.
In
this
pape
r
,
a h
yb
rid
m
et
ho
d
im
ple
m
ented
to ac
hieve
ef
fici
ent, go
od d
i
ve
rsity
an
d faste
r
c
onverge
nce
rate.
3.1.
Modifie
d
NSGA
-
II
In
M
NSGA
-
I
I
the
Eucli
dea
n
distance
is
im
plem
ented
in
t
he
sel
ect
ion
process
of
eac
h
fron
t
in
2N
Popu
la
ti
on
of
the
com
bin
ed
par
e
nt
an
d
off
sp
ri
ng
popula
ti
on
.
Eucli
dea
n
distance
im
pl
e
m
ents
m
idp
oi
nts
to
enh
a
nce t
he diversity
in
eac
h fron
t.
P
opulati
on m
e
m
ber
s which a
re
ou
t
of
the r
a
nge of E
uc
li
dean
distanc
e w
il
l
be
discar
ded
a
nd
rest
incl
ude
in
P
opulati
on
N,
duri
ng
t
he
sel
ect
ion
proc
ess,
sel
ect
ed
m
e
m
ber
s
Ni
e
xceeds
the N t
he
n best fitness m
e
m
ber
s ar
e
selec
te
d, an
d
if t
he
sel
e
ct
ed
po
pu
la
ti
on is less t
han N, a
new po
pu
la
t
ion
i
s
gen
e
rated
whic
h
is
within
Eu
c
li
dean
dis
ta
nce
.
Eucli
dea
n dist
ance
,
=
(
√
∑
(
−
)
2
=
1
)
/
(
−
1
)
;
i=
1,
2,
3......
(8)
If
Ni >
N
the
n be
st fit
ness p
opul
at
ion
m
e
m
ber
s selec
te
d.
If
Ni <
N
the
n ne
w
r
an
dom
p
opulati
on
is
g
e
ne
rated
w
hich
s
houl
d
li
e in t
he r
ang
e
of E
uclid
ean
distance.
The
ran
ge
of
E
uclidean
dista
nc
e
is
de
velo
pe
d
as (
(
,
+
,
,
−
)
,
w
her
e
is
sel
ect
ed
s
uc
h
that t
he
m
axi
m
um
p
op
ulati
on m
e
m
ber
s sh
oul
d
exist wit
hin
t
he
ra
ng
e
of
E
uc
li
dean
d
ist
anc
e, the
app
li
cat
ion
of
Eucli
dea
n dist
ance is t
o
e
nh
a
nc
e the
div
e
rsity
am
on
g
the
po
pu
la
ti
on m
e
m
ber
s.
3.2.
Modifie
d
pop
ulat
i
on
vari
ant dif
feren
tial e
vo
lu
tio
na
r
y al
go
ri
th
m
In
P
VD
E
[
20]
the
i
niti
al
popula
ti
on
is
re
fr
e
sh
e
d
us
in
g
t
he
inter
qu
a
rtil
e
r
ang
e
co
nce
pt
a
t
the
init
ia
l
sta
ge
to
m
ake
an
ef
fici
ent
al
gorithm
.
Scal
in
g
fact
or
an
d
c
r
os
s
ov
e
r
pro
ba
bili
ty
var
ia
ti
on,
place
a
vital
ro
le
in
the pr
oductio
n of o
ff
s
pr
i
ng. In
MPV
DE
t
he
c
on
ce
pt c
oeffici
ent of
Q
uar
ti
le
dev
ia
ti
on is
use
d.
S
tep
1:
T
he
i
niti
al
h
al
f
ra
ndom
p
op
ulati
on i
s g
e
ner
at
e
d.
St
ep
2:
A
tw
o
r
ow
vect
or
is
ini
ti
al
iz
ed
for
scal
ing
facto
r
and
cr
os
s
over
pro
bab
il
it
y
para
m
et
ers
in
the
range
[0,1
]
a
nd
it
m
i
nim
u
m
siz
e
is
[1x
3]
vecto
rs.
Evaluati
on
of
m
axi
m
u
m
,
m
i
nim
u
m
and
m
edian
value
s
for
eac
h param
et
er is d
et
e
rm
i
ned.
St
ep
3:
In
it
ia
ll
y vector
d
if
fer
e
nce
Var
m
i
n
is evalu
at
e
d wit
h ge
ner
at
io
n
li
m
it
s it
s d
iff
eren
ce
is m
ultip
li
ed wit
h 0.01.
Var
is
cal
c
ulate
d
us
i
ng
t
he
con
ce
pt
of
c
oe
ff
ic
ie
nt
of
Q
uar
ti
le
de
viati
on
with
eac
h
colum
n
of
pa
r
ent
popula
ti
on
.
Var
=
1
−
2
1
+
2
;
(9)
wh
e
re
Q
1
is t
he
sem
i i
nterq
uart
il
e ran
ge
and
Q
2
is t
he
m
id quar
ti
le
r
a
nge
Q
1
=
(H
i
gh qua
rtil
e
–
L
ow qua
rtil
e)/2;
(10)
Q
2
= (
High
qua
rtil
e + Low
quarti
le
)/2;
(
11)
W
it
h new l
im
its
ℎ
ne
w re
fr
es
hm
ent p
op
ulati
on is e
valuate
d wit
h
ste
p
1.
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 meth
od
for ac
h
ie
vi
ng
Pareto
fro
nt
on ec
onomic
em
issi
on
dis
pa
tc
h
(
K. Ra
je
s
h
)
3361
St
ep
4:
Evaluate
t
he
f
it
ness
of
t
he
r
efr
es
h
par
e
nt
popula
ti
on
an
d
fin
d
t
he
best
on
e
w
hich
give
s
the
m
ini
m
u
m
fitness
value
a
nd up g
rad
at
io
n of scal
ing fac
tor (sf) a
nd cr
osso
ver
pro
ba
bili
ty
(
cp)
is
do
ne
.
Find the m
utati
on and c
r
os
s
over,
t
he
sel
ect
io
n
is m
ade fro
m
the m
erg
e
popula
ti
on
of
offs
pr
i
ng and
par
e
nt.
Upgradati
on
of the
par
am
et
er is don
e
.
3.3.
Hy
brid
met
hod
In
t
his
pa
pe
r
th
e
com
bin
at
ion
of
M
NSGA
-
I
I
and
MPV
DE
i
s
us
e
d
as
a
hy
bri
d
m
et
ho
d
f
or
the
s
ol
vi
ng
env
i
ronm
ental
econom
ic
load
disp
at
c
h
with
two
obj
ect
ives
cost
an
d
em
is
sion.
I
n
P
VDE
[20]
two
obje
ct
ives
are
c
onve
rted
into
m
ono
obje
ct
ive
us
i
ng
t
he
weig
hted
s
um
m
et
ho
d
but
in
this
pa
pe
r
c
os
t
f
unct
ion
an
d
e
m
issi
on
funct
ion
is
co
ns
ide
red
as
tw
o
ob
j
ect
ives.
I
n
thi
s
pap
e
r
the
en
ti
re
popu
la
ti
on
is
sp
li
te
d
int
o
two
half’s,
t
he
fi
rs
t
half
of
the
popula
ti
on
is
app
li
ed
wit
h
MNSGA
-
I
I
a
nd
sec
ond
half
of
the
po
pu
l
at
ion
is
encou
ntere
d
by
MPVDE
to
achieve
t
he
be
st
op
ti
m
al
so
luti
on
the
ste
ps
in
the
ap
plica
ti
on
of
M
NSGA
-
I
I
an
d
MPVDE f
or s
ol
vin
g en
vir
onm
ental
eco
nom
ic
load
d
is
patch
is as foll
ows
In
it
ia
li
ze
the
num
ber
of
obj
e
ct
ives,
decisi
on
var
ia
bles
,
a
nd
loa
d
dem
and,
m
axi
m
u
m
nu
m
ber
of
it
erati
on
s
and
al
so
c
os
t
and
em
issi
on
coeffic
ie
nts.
Le
t
k=
[k
1
,
k
2
,
k
3
,
k
4
,
k
5
...]
w
he
re
k
is
the
decisi
on
vect
or
a
nd
each elem
ent co
r
res
ponds to
the
ou
t
pu
t
pow
er
of
eac
h ge
ne
rator.
Gen
e
rate t
he
r
a
ndom
p
opulati
on such
that it
has
t
o
sat
isfy e
qu
al
it
y co
ns
tra
int i
n
(
4
)
.
MNSGA
-
I
I
is
ap
plied
f
or
the
fi
rst
par
t
of
th
e
init
ia
l
popula
ti
on
and
us
in
g
(
2
)
an
d
(
3
)
eval
uate
the ob
j
ect
ive
f
un
ct
io
n o
f bo
t
h cost a
nd em
iss
ion
.
Fr
om
the
existi
ng
po
pu
la
ti
on
init
ia
l
ly
identify
the
nondomi
nated
ind
i
viduals.
Pr
e
dict
the
nond
om
inated
so
rtin
g suc
h
t
ha
t t
he
ra
nkin
g
s
el
ect
ion
m
et
ho
d
is i
m
ple
m
ent
ed
to
h
i
gh
li
gh
t
the r
a
nkin
g
f
r
onts.
Fo
r
t
he
evalu
at
ion
of
cr
owding
distance
,
a
so
rted
popula
ti
on
of
eac
h
obj
ect
i
ve
functi
on
value
i
n
the inc
reasin
g order
is re
quire
d.
The
ge
ne
rati
on
of
half
of
the
init
ia
l
po
pula
ti
on
of
t
he
offs
pri
ng
is
evaluat
ed
by
pe
rfo
rm
i
ng
t
he
m
utati
o
n
and cr
os
s
over.
Evaluate
t
he
obj
ect
iv
e
f
un
ct
i
on
s
from
(
2
)
a
nd
(
3
)
f
or
the
nex
t
half
of
th
e
popula
ti
on
usi
ng
MP
V
DE,
for
this p
opulati
on
r
ef
res
hm
ent is don
e
b
y
us
i
ng
the conce
pt
of
sem
i i
nterquar
t
il
e and
m
id qu
arti
le
.
Pr
e
dicti
ng
the
best
popula
ti
on
w
hich
gi
ves
m
ini
m
u
m
cost
value
an
d
perf
or
m
s
the
cro
ss
ov
e
r
an
d
m
utatio
n
for
the
g
e
ne
rati
on of
offs
pr
i
ng.
Me
rg
e
both
th
e
offs
pr
i
ngs
th
at
will
eq
ualise
to
t
he
popula
ti
on
in
num
ber
.
Com
bin
e
pa
re
nt
popula
ti
on
a
nd
offsprin
g
t
hat lea
ds
t
o
twic
e t
he pop
ulati
on.
Sele
ct
the
best
ind
ivi
du
al
s
of
popula
ti
on
ba
s
ed
on
eucli
dea
n
distance
s
uc
h
that
the
best
non
-
dom
inant
on
e
placed
i
n
the re
posit
or
y. In
t
he
nex
t
it
erati
on
it
is
ta
ken
as
a
new
p
a
ren
t popu
la
ti
on
a
nd
t
he
process r
e
pea
ts
un
ti
l
it
reac
hes
m
axi
m
u
m
gen
erati
on.
Gen
e
r
al
ly
in
the
m
ul
ti
ob
j
ect
i
ve
c
om
bin
ed
en
vir
onm
ental
econ
om
ic
load
disp
at
c
h
so
luti
ons
le
a
ds
to
m
ulti
ple
op
tim
al
po
ints
in
w
hich
t
he
best
opti
m
a
l
i
s
to
be
obta
in
ed,
for
this
purp
ose
a fuzzy
set t
he
or
y
[21] is i
m
plem
ented.
3.4.
Be
st
op
tim
al
value
Fu
zzy
set
theo
r
y
i
s
helpf
ul
in
achievin
g
the
be
st
so
luti
on
fro
m
the
par
et
o
optim
al
set
fo
r
t
he
decisi
on
m
aker
[
22
]
, d
e
ci
sion
m
aker
de
ci
sion
is
inacc
ur
at
e
in n
at
ure.
k
th
obj
ect
ive fun
ct
io
n
of
t
he
so
luti
on
in p
ar
et
o
set
F
k
is i
ndic
at
ed by a m
e
m
ber
shi
p
f
unct
ion µ
k
is de
fine
d
a
s
=
{
1
,
≤
−
−
,
<
<
0
,
≥
(12)
w
he
re
F
k
m
in
a
nd
F
k
m
ax
are
t
he
m
ini
m
u
m
and
m
axi
m
um
values
of
k
th
obj
ect
ive
func
ti
on
,
the
nor
m
al
iz
ed
m
e
m
ber
sh
ip
f
unct
ion µ
i
=
∑
=
1
∑
∑
=
1
=
1
(13)
Non
do
m
inate
d
so
l
ution
s
in
nu
m
ber
is represe
nt
ed
a
s M, t
he best s
olu
ti
on
ha
ving m
axi
m
um
v
al
ue
of µ
i
.
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
.
4
,
A
ugus
t
2020
:
3358
-
3366
3362
4.
RESU
LT
S
A
ND D
I
SCUS
S
ION
This
hy
br
id
m
et
hod
is
te
ste
d
with
thre
e
te
st
cases.
Ca
se
1
i
t
relat
ed
to
IE
EE
30
bus
te
st
syst
e
m
with
six
ge
ner
at
or
s
with
an
d
wit
hout
los
ses
an
d
Ca
se
2
ap
plied
to
IEEE
118
bu
s
te
st
syst
em
with
14
generator
s
with
an
d
with
ou
t
valve
poin
t
loading
ef
fec
t
and
Ca
se
3
i
s
i
m
ple
m
ented
on
a
40ge
ne
r
at
or
s
syst
em
without
losses.
I
n
al
l
th
e
cases
cost
an
d
em
issi
on
obje
ct
ive
functi
on
are
not
conve
r
te
d
into
a
sin
gle
obj
ect
ive
a
nd
not
consi
der
e
d
a
s
a
co
ns
trai
nt
in
the
opti
m
izati
on
pro
blem
.
MATL
AB
program
was
de
velo
ped
on
Int
el
i3
process
or w
it
h
2G
B R
AM, a
n op
e
rati
ng syst
e
m
is
W
I
NDO
Ws 1
0.
4.1.
C
as
e
1
IEEE
30
bu
s
te
st
syst
e
m
havi
ng
a
l
oa
d
de
m
and
of
2.8
34pu
with
a
ba
se
of
10
0MV
A
and
ha
ving
Popu
l
at
io
n
nu
m
ber
40
an
d
t
he
m
axi
m
u
m
no
of
it
erati
on
s
is
100.
T
he
t
rad
e
off
c
urve
f
or
th
e
EE
L
D
without
losses
is
sho
w
n
in
F
i
gure
1
a
nd
with
los
ses
is
sh
ow
n
in
F
i
gure
2.
T
he
m
i
nim
u
m
fu
el
cost
of
607.1
265
($
/
hr)
is
achieve
d
f
or
t
he
em
issio
n
of
0.203
1
(to
n/hr)
an
d
the
m
axi
m
um
fu
el
cost
of
637.8
721(
$/
hr
)
with
the
m
ini
m
u
m
e
m
issi
on
of
0.
1942(t
on
/
hr)
as
sh
ow
n
in
T
able
1
the
exec
ution
ti
m
e
ta
ken
with
ou
t
los
ses
is
9.202
0
sec
an
d
with
t
he
los
ses
the
l
owest
cost
value
is
614.2
687
(
$/h
r
)
f
or
the
em
is
sion
val
ue
of
0.2
00
9
(to
n/h
r
)
is
s
hown
i
n
T
a
ble
2
and
t
he
ti
m
e
ta
ken
for
the
com
pu
ta
ti
on
is
9.840sec
.
Th
e
cost
an
d
em
issi
on
coeffic
ie
nts as
well
as m
axi
m
um
an
d
m
ini
m
um
v
al
ues of
powe
r ge
ner
at
io
n
is t
a
ken as in
[
23
]
.
Figure
1. IEE
E
30 bus t
est
sys
tem
w
it
ho
ut
lo
sses
Figure
2. IEE
E
30 bus t
est
sys
tem
w
it
h
losse
s
Table
1.
C
om
par
iso
n
of
cost a
nd em
issi
on
r
e
su
lt
s of case
1
without l
osse
s
Me
th
o
d
MN
MN
S
GA
-
II
+
MP
VD
E
S
P
E
A2
[
2
4
]
MODE
MO
P
S
O
F
C
P
S
O
[
2
4
]
C
o
st
E
m
is
si
o
n
C
o
st
E
m
is
si
o
n
C
o
st
E
m
is
si
o
n
C
o
st
E
m
is
si
o
n
C
o
st
E
m
is
si
o
n
P
G1
0
.2227
0
.4076
0
.1097
0
.4060
0
.1162
0
.4151
0
.1194
0
.3979
0
.1070
0
.4097
P
G2
0
.3719
0
.4609
0
.2993
0
.4589
0
.2865
0
.4604
0
.3072
0
.4258
0
.2897
0
.4550
P
G3
0
.5589
0
.5380
0
.5243
0
.5378
0
.5605
0
.5409
0
.4907
0
.5268
0
.5250
0
.5363
P
G4
0
.7362
0
.3892
1
.0162
0
.3834
1
.0098
0
.3
8
0
8
1
.0041
0
.3984
1
.0150
0
.3842
P
G5
0
.5247
0
.5307
0
.5245
0
.5378
0
.5060
0
.5298
0
.5212
0
.5336
0
.5300
0
.5348
P
G6
0
.4196
0
.5076
0
.3598
0
.5101
0
.3550
0
.5070
0
.3914
0
.5515
0
.3673
0
.5140
C
o
st
6
0
7
.1265
6
3
7
.8721
6
0
0
.109
6
3
8
.264
6
0
0
.2071
6
3
8
.9388
6
0
0
.2823
6
3
6
.904
5
6
0
0
.132
6
3
8
.358
E
m
is
si
o
n
0
.2031
0
.1942
0
.2221
0
.1942
0
.2219
0
.1942
0
.2205
0
.1944
0
.2222
0
.1942
Table
2.
C
om
par
iso
n of
cost a
nd em
issi
on
r
e
su
lt
s of
c
ase
1
with lo
sses
Me
th
o
d
MN
S
GA
II
+
MP
VD
E
S
P
E
A2
[2
4
]
MODE
MO
P
S
O
F
C
P
S
O
[
2
4
]
C
o
st
E
m
is
si
o
n
C
o
st
E
m
is
si
o
n
Co
st
E
m
is
si
o
n
C
o
st
E
m
is
si
o
n
C
o
st
E
m
is
s
ion
P
G1
0
.2606
0
.4048
0
.1189
0
.4107
0
.1318
0
.4169
0
.1145
0
.3872
0
.1130
0
.4063
P
G2
0
.4012
0
.4625
0
.3085
0
.4635
0
.2925
0
.4478
0
.3240
0
.4209
0
.3145
0
.4586
P
G3
0
.5391
0
.5436
0
.5200
0
.5447
0
.5287
0
.5488
0
.5086
0
.5271
0
.582
6
0
.5510
P
G4
0
.7042
0
.4110
1
.0081
0
.3903
0
.9934
0
.3847
1
.0525
0
.4545
0
.9860
0
.4084
P
G5
0
.5310
0
.5380
0
.5286
0
.5444
0
.5362
0
.5635
0
.5488
0
.5950
0
.5264
0
.5432
P
G6
0
.4180
0
.4942
0
.3742
0
.5155
0
.3786
0
.4995
0
.3191
0
.4828
0
.350
0
.4974
C
o
st
6
1
4
.2687
6
4
0
.2599
6
0
5
.548
6
4
6
.190
6
0
6
.276
6
4
4
.1750
6
0
7
.8919
6
3
7
.6409
6
0
7
.786
6
4
2
.896
E
m
is
si
o
n
0
.2009
0
.1942
0
.2208
0
.1942
0
.2197
0
.1942
0
.2247
0
.1947
0
.2201
0
.1942
L
o
ss
e
s
0
.0201
0
.0351
0
.0272
0
.0335
0
.0309
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 meth
od
for ac
h
ie
vi
ng
Pareto
fro
nt
on ec
onomic
em
issi
on
dis
pa
tc
h
(
K. Ra
je
s
h
)
3363
4.2.
Ca
se
2
It
is
i
m
ple
m
ented
on
IEEE
11
8
bus
syst
em
with
14
ge
ner
a
tors
with
a
nd
without
Valve
po
i
nt
loa
ding
eff
ect
is
exh
ibi
te
d.
Total
De
m
and
was
20
00M
W
[
25
]
with
a
popu
la
ti
on
of
40
an
d
crosso
ver
pr
ob
a
bi
li
ty
and
m
uta
ti
on
val
ue
s
are
0.8
5
a
nd
0.166.
C
om
par
ison
is
s
how
n
i
n
T
a
ble
3
with
ou
t
valve
point
loading
e
ff
ect
and
in
T
able
4
wit
h
valve
po
i
nt
loadi
ng
the
c
orrespo
nd
i
ng
Pa
r
et
o
op
ti
m
al
so
luti
on
s
are
s
hown
in
F
i
gure
3
an
d
F
igure
4.
F
or
a
m
ini
m
u
m
e
m
i
ssion
of
27
76.
1633
(t
on
/
hr)
th
e
cost
val
ue
is
8968.
1794(
$/hr)
is
ob
ta
ine
d
ha
ving
the
exec
utio
n
tim
e
of
9.906
sec
in
case
of
without
val
ve
po
i
nt
an
d
wit
h
val
ve
poi
nt
m
ini
m
u
m
e
m
i
ssion
2754.
2895(t
on
/
hr)
a
nd r
el
at
ed
cost
value
is
9004.
3651
($
/
hr)
w
it
h
e
xec
utio
n
ti
m
e o
f
10.
57
2s
ec.
Figure
3. IEE
E
118 bus test
s
yst
e
m
w
it
ho
ut
valve p
oin
t
Figure
4. IEE
E
118 bus test
s
yst
e
m
w
i
th v
al
ve po
i
nt
Table
3.
C
om
par
iso
n of
cost a
nd em
issi
on
r
e
su
lt
s of c
ase
2
without
valve
po
i
nt
Table
4.
C
om
par
iso
n of Cost
and
em
issi
on
re
su
lt
s of c
ase
2 wit
h valve
poi
nt
Me
th
o
d
s
Uni
t
MN
S
GA
-
II
+
MP
VD
E
S
P
E
A2
NSGA
-
2
MO
P
S
O
C
o
st
E
m
is
si
o
n
C
o
st
E
m
is
si
o
n
Co
st
E
m
is
si
o
n
C
o
st
E
m
is
si
o
n
P
G1
2
8
4
.052875019457
1
9
7
.626370894082
3
2
9
.7295
2
3
6
.8771
2
4
0
.0744
2
1
9
.0736
3
2
5
.8810125
2
4
1
.12950
P
G2
1
9
2
.18732935442
1
5
1
.576676960222
1
5
0
.0293
1
5
0
.0267
3
0
2
.1989
1
5
0
.6923
1
5
0
1
5
0
P
G3
1
3
0
1
3
0
9
5
.3
3
3
7
1
2
9
.1020
9
5
.4
5
5
8
1
2
4
.9328
1
3
0
1
3
0
P
G4
1
2
8
.3185708334
1
3
0
1
2
1
.6660
1
2
7
.7472
6
9
.4
0
9
3
1
2
8
.1623
1
3
0
1
3
0
P
G5
1
8
5
.626065115719
1
6
3
.104211030723
2
5
0
.0010
1
5
0
.6224
2
4
7
.4409
1
5
3
.9634
2
0
2
.8506499
1
7
8
.9299
P
G6
1
7
9
.495514517557
2
5
1
.690474265875
1
8
2
.3107
2
8
3
.9287
2
3
6
.9518
2
8
4
.5399
1
9
4
.7000195
22
2
.3214
P
G7
1
3
8
.68198357291
1
8
0
.541835667428
1
8
4
.7832
1
8
4
.8378
1
8
5
.3728
1
8
7
.2257
1
3
5
1
7
9
.9619
P
G8
1
3
9
.637661586537
1
7
5
.029843754326
1
5
8
.0345
1
5
3
.9764
1
6
0
.0652
1
6
0
.9197
1
0
9
.6862982
1
4
5
.6571
P
G9
1
6
2
1
6
0
.430587427345
1
2
3
.2748
1
2
6
.1220
7
6
.8
0
9
0
1
5
6
.0239
1
6
2
1
62
P
G10
1
6
0
1
6
0
1
5
5
.5005
1
5
9
.6728
1
3
7
.0618
1
5
5
.8425
1
6
0
1
6
0
P
G11
80
80
6
0
.4
5
2
0
7
9
.7
9
0
2
6
0
.5
7
4
6
6
8
.8
1
1
9
7
9
.9
9
9
8
2
5
5
80
P
G12
80
80
7
1
.0
1
8
5
7
9
.6
4
0
0
6
3
.6
7
6
3
7
7
.8
4
6
6
80
80
P
G13
85
85
6
2
.9
1
3
9
8
2
.6
5
7
2
7
1
.5
1
9
0
7
9
.1
7
9
6
8
4
.8
8
2
1
9
4
4
85
P
G14
55
55
5
4
.9
5
2
4
5
4
.9
9
9
5
5
3
.3
9
0
2
5
2
.7
8
5
8
55
55
C
o
st
8
7
8
6
.2
1
9
8
8
1
4
9
3
1
8
9
0
0
4
.3
6
5
1
2
5
1
1
8
7
7
9
3
4
2
.9
3
8
4
9
8
3
9
.5
1
3
4
9
4
1
5
.7
1
4
4
1
0
1
3
3
.174
9
7
0
1
.4
9
2
3
3
5
8
9
0
8
.9
3
7
2
E
m
is
si
o
n
3
2
1
2
.8
0
7
1
4
4
2
7
5
7
2
7
5
4
.2
8
9
5
1
6
6
9
3
8
5
3
6
0
5
.6
4
5
1
2
8
5
0
.8
9
6
6
4
7
9
9
.8
3
2
6
2
8
5
0
.4
5
7
8
3
2
5
5
.1
6
6
7
7
2
2
8
2
7
.1
3
2
0
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
.
4
,
A
ugus
t
2020
:
3358
-
3366
3364
4.3.
Ca
se
3
The
pro
po
se
d
m
et
ho
d
is
a
ppli
ed
to
the
40
ge
ner
at
or
s
syst
em
an
d
the
res
ul
ts
are
show
n
i
n
the
T
able
5
and
t
he
co
rr
es
pondin
g
Paret
o
grap
h
is
sho
wn
in
F
ig
ur
e
5
sho
wing
the
best
com
pr
om
ise
so
luti
on
.
The
total
dem
and
was
1050
0M
W
[
26
]
with
the
po
pu
l
at
ion
of
40
ex
hi
bited
with
10
0
it
erati
on
s
an
d
the
trade
off
c
urve
is
ob
ta
ine
d
f
or
th
e
trai
ls
of
30
tim
es
and
for
the
m
ini
m
u
m
e
m
issi
on
value
112444.
1310
(
ton
/h
r)
t
he
ach
ie
ve
d
cost
val
ue
12
4483.
5238
(
$/hr
)
an
d
the
ti
m
e
ta
ken
f
or
the
execu
ti
on
of
t
he
Ma
tl
ab
pro
gr
am
is
17
.
233
sec
.
Figur
e
6
sho
w
s
the
grap
h
between
best
val
ue
s
from
ref
res
hed
popula
ti
on
ver
ses
num
ber
of
it
erati
on
s
.
Fr
om
the
grap
h,
it
is
pr
e
di
ct
ed
t
hat
the
best
value
of
the
cost
f
unct
ion
value
over
it
e
rati
on
100
re
aches
the
m
ini
m
u
m
cost
val
ue
wh
i
ch
is
us
ed
f
or
m
uta
ti
on
an
d
cro
ss
over
eval
uation
in
MP
VD
E
.
T
he
flo
w
c
har
t
sh
ow
n
in
F
ig
ure7
giv
e
s
the
m
ajo
r
ste
ps
i
n
the
im
ple
m
entat
ion
of
a
hybr
i
d
m
et
ho
d
f
or
s
olv
i
ng
eco
no
m
ic
e
m
issi
on
disp
a
tc
h
prob
le
m
.
In
it
ia
ll
y
m
od
ifie
d
non
-
dom
inant
so
rtin
g
ge
ne
ti
c
al
go
rithm
–
II
is
app
li
ed
fol
lowe
d
by
m
od
ifie
d
P
opulati
on
Var
i
ant
Diff
e
re
ntial
Evo
luti
on
an
d
in
both
the
m
et
ho
ds
non
-
do
m
inant
so
rti
ng
is
app
li
ed
f
or
t
he
gen
e
rati
on
of
offsprin
g’s
an
d
in
this
stream
el
it
is
m
con
cept
is
at
tribu
te
d
t
o
m
ake
the
bes
t
on
e
in
the
reposit
ory
w
hich
e
nha
nces
the
co
nv
e
rg
e
nce
rate.
The
sel
ect
io
n
process
is
ord
ered
by
the
c
r
owding
distance
ope
ra
tor
f
or
im
pr
ov
ing
the
un
i
for
m
sp
read
ov
e
r
trade
off
c
urve.
I
n
MN
SGA
-
II
sim
ulate
d
bin
ary
cro
ss
over
a
nd
the
poly
no
m
ia
l
m
utati
on
is
carried
over
f
or
the
ge
ner
at
i
on
of
c
orres
pondin
g
of
f
sprin
g’s
an
d
the
la
tt
er
m
eth
od
MPV
DE
best
value
w
hi
ch
produce
m
ini
m
u
m
fitness
is
pr
edict
ed
a
m
on
g
the
re
f
resh
e
d
popula
ti
on
w
hich
is
us
e
d
in
th
e cal
culat
ion
s
of m
utati
on
and c
ro
ss
over
and
gen
e
rati
ng a
trust
worthy
re
su
lt
s.
Figure
5. 40
ge
ner
at
or
s
s
yst
em
w
it
h
no
l
os
s
es
Figure
6.
Be
st
value ve
rses n
o of i
te
rati
on
s
Table
5.
C
os
t a
nd
em
issi
on
re
su
lt
s of
case
3
Metho
d
Un
it
MNSG
A
-
I
I
+
M
P
VDE
Co
st
E
m
iss
io
n
Co
st
E
m
iss
io
n
PG1
1
1
2
.5912
1
7
7
5
8
1
5
2
1
0
3
.1491
1
3
4
4
7
7
3
3
PG2
1
550
4
6
5
.7093
5
3
3
9
6
3
1
9
PG2
113
1
0
6
.916
4
7
9
6
6
3
2
7
4
PG2
2
550
4
5
0
.8983
5
2
7
2
9
2
7
PG3
119
1
1
5
.2054
9
4
2
2
2
9
6
7
PG2
3
5
2
5
.9715
6
5
6
7
6
9
7
1
4
7
5
.8423
0
9
3
0
2
8
5
8
PG4
1
8
5
.9339
9
2
5
4
2
5
1
1
1
6
3
.6363
9
4
6
2
9
7
8
2
PG2
4
550
4
8
7
.3049
8
7
5
1
2
0
4
6
PG5
97
96
PG2
5
5
3
3
.5252
3
5
2
3
0
1
0
5
4
5
2
.2363
2
4
8
4
0
2
8
2
PG6
1
3
2
.7229
0
9
1
3
0
9
7
3
1
3
3
.9377
4
3
7
7
5
7
2
6
PG
26
550
4
4
1
.1147
8
1
1
6
2
7
4
7
PG7
300
2
8
9
.4637
5
6
0
1
8
9
7
5
PG2
7
10
3
9
.65
5
4
6
4
9
6
3
5
5
6
3
PG8
300
2
9
7
.2746
1
2
6
1
9
7
8
9
PG2
8
1
4
.14
1
2
4
9
3
1
0
6
5
8
8
1
6
.70
2
2
1
4
9
9
6
7
3
5
9
PG9
300
2
6
7
.8701
5
7
8
1
9
9
6
3
PG2
9
1
0
.43
3
4
6
3
2
7
9
2
1
3
4
2
6
.45
5
7
0
9
6
7
5
3
8
5
5
PG1
0
1
8
5
.6707
1
0
2
3
6
0
1
7
2
7
2
.3587
9
2
2
1
6
0
0
4
PG3
0
97
92.
8
3
6
4
6
4
9
7
3
3
5
8
PG1
1
1
0
1
.3998
5
6
4
6
7
7
1
4
2
4
3
.9106
1
9
0
1
8
2
1
3
PG3
1
190
1
6
1
.5469
4
2
8
3
4
6
2
4
PG1
2
1
2
8
.9996
8
4
9
8
8
9
8
8
2
8
3
.5050
2
9
0
8
3
9
3
8
PG3
2
1
5
7
.7449
1
1
6
4
8
8
7
4
1
6
9
.8345
7
4
7
4
0
4
9
8
PG1
3
125
3
7
0
.6473
2
1
7
4
8
8
4
1
PG3
3
190
1
7
8
.6597
9
4
8
2
4
3
0
7
PG1
4
2
4
4
.1594
1
7
4
9
6
8
6
9
4
2
7
.2089
4
4
1
1
5
7
3
7
PG3
4
200
1
9
4
.9725
2
6
0
5
3
0
0
6
PG1
5
2
7
6
.5526
3
7
1
8
3
8
3
3
7
9
.7764
9
6
0
7
8
5
7
5
PG3
5
200
1
9
3
.4869
5
5
1
8
2
6
4
9
PG1
6
4
2
6
.3493
5
9
6
1
3
7
5
6
3
9
4
.2369
7
3
7
3
9
7
1
8
PG3
6
200
1
8
8
.9952
2
2
7
8
0
5
5
4
PG1
7
4
8
0
.6584
3
6
5
2
6
6
7
8
4
0
9
.9054
8
3
6
2
5
3
4
PG3
7
110
9
9
.72
7
2
3
7
0
5
2
5
7
7
3
PG1
8
500
4
2
1
.9695
6
0
0
3
2
6
2
2
PG3
8
110
100
.3944
9
3
1
4
1
1
2
4
PG1
9
4
7
4
.5123
2
0
2
8
6
5
0
5
4
7
2
.5865
1
4
6
5
8
0
6
7
PG3
9
110
1
0
2
.8197
5
0
0
3
1
8
6
3
PG2
0
550
4
7
3
.6613
2
7
0
6
1
8
0
7
PG4
0
4
8
7
.6330
3
2
6
2
2
1
8
7
4
3
7
.5857
2
6
2
2
9
1
6
6
Co
st
E
m
iss
io
n
1
1
9
6
3
3
.3135
7
9
5
9
6
3
8
0
1
8
9
.3591
5
6
8
2
7
1
2
4
4
8
3
.5238
5
5
6
8
6
1
1
2
4
4
4
.1310
7
1
8
0
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 meth
od
for ac
h
ie
vi
ng
Pareto
fro
nt
on ec
onomic
em
issi
on
dis
pa
tc
h
(
K. Ra
je
s
h
)
3365
Figure
7
.
Flo
w
ch
a
rt of
hybri
d
m
et
ho
d
5.
CONCL
US
I
O
N
The
m
ulti
ob
j
e
ct
ive
econom
i
c
e
m
issi
on
load
di
sp
at
c
h
optim
iz
ation
pro
bl
e
m
has
been
so
lve
d
with
m
any
heurist
ic
an
d
novel
m
et
h
ods.
I
n
this
pap
e
r,
tw
o
c
onflic
ti
ng
ob
j
ec
ti
ves
are
co
ns
i
der
e
d
sim
ultaneousl
y
and
t
he
com
bin
at
ion
of
M
N
SGA
-
II
a
nd
M
PVDE
is
hi
gh
l
igh
te
d
a
s
one
of
th
e
hybri
d
m
et
ho
d
s
i
n
ac
hieving
the
be
st
opti
m
al
so
luti
on.
It
i
s
te
ste
d
on
IE
EE
30
bus
te
st
syst
e
m
with
6
ge
ner
at
or
s
an
d
IEEE
118
bus
te
s
t
syst
e
m
with
14
ge
nerat
ors
and
with
sta
nda
rd
40
ge
ner
at
ors
syst
em
.
The
res
ults
obta
in
ed
are
c
om
par
ed
wit
h
the
e
xisti
ng
m
et
hods
.
I
n
al
l
t
he
t
hr
ee
te
st
ca
ses
po
pu
la
ti
on
par
am
et
er
an
d
the
m
axi
m
u
m
nu
m
ber
of
it
er
at
ion
s
100
rem
ai
n
sam
e
in
achievin
g
the
Paret
o
optim
al
set
.
Man
y
nu
m
ber
s
of
trai
ls
of
about
30
we
re
im
ple
m
ented
for
al
l
m
e
t
ho
ds
li
ke
SPE
A
-
II,
MOPS
O,
an
d
MODE
,
f
or
a
chievin
g
t
he
optim
al
so
luti
on
a
m
on
g
this
H
ybr
i
d
m
et
ho
d
a
c
hie
ve
d
the
b
e
st o
ptim
al
so
luti
on
w
it
h
good c
onve
rg
e
n
ce.
REFERE
NCE
S
[1]
A.
El
-
Keib
,
H.
Ma,
J.
Hart
.
,
“
E
conomic
dispatc
h
in
vie
w
of
th
e
cl
e
an
ai
r
act
of
1990,
”
IE
EE
Tr
ans.
Powe
r
Syst
.
,
vol.
9
,
no
.
2
,
pp
.
972
–
978,
1994
.
[2]
A.
Fara
g,
S.
Al
-
Baiy
at,
T
.
Chen
g.
,
“
Ec
onom
ic
l
oad
dispatch
m
ult
iobjective
op
tim
iz
at
ion
pro
ce
d
ure
s
using
li
near
progra
m
m
ing
technique
s,
”
I
EE
E
Tr
ans.
Powe
r S
yst
.,
vol
.
10
,
no
.
2,
pp
.
731
–
738
,
1995.
[3]
J.S
Dhill
on,
S.C
Patt
i,
D.P
K
otha
ri
.
,
“
Stochas
ti
c
ec
onom
ic
e
m
ission
loa
d
di
spatc
h
,
”
El
e
ct
ri
cal
Powe
r
System
s
Re
search
,
vol
.
2
6,
pp
.
179
-
186
,
1993.
[4]
Y.T
.
Hs
ia
o
,
H.
D.
Chia
nge
,
L
iu
,
Y.L.
Chen
.
,
“
A
computer
package
for
opt
ima
l
m
ult
i
-
obj
ec
t
ive
VA
R
pla
nning
in
la
rge
sca
le
pow
e
r
s
y
st
ems
,
”
IE
E
E
Tr
ansacti
ons
on
Powe
r S
ystem
s
,
vol. 9, no. 2,
pp.
668
-
676,
19
94.
[5]
D.
Sriniva
san,
C.
S.
Chang,
A.C.
Liew.,
“
Multi
obj
ec
t
ive
Ge
ner
ation
sche
du
le
using
fuz
z
y
opti
m
al
sea
rc
h
te
chn
ique
,
”
I
EE
Proce
ed
ings
-
Gene
ration, Tr
ansm
ission and
Dist
ributi
on
,
vol. 14
1,
no
.
3
,
pp
.
233
-
242,
1994
.
[6]
C.
M.
Huang,
H.
T.
Yang,
C
.
L
.
Huang.
,
“
Bi
-
obj
ective
power
dispa
tc
h
using
fuz
z
y
sati
sfac
ti
on
-
m
ax
imizi
ng
decisio
n
appr
oac
h
,
”
IE
EE Tr
ansacti
ons on
Powe
r S
yste
ms
,
vol.
12
,
no
.
4
,
pp
.
1715
-
1721
,
19
97.
[7]
D.
B.
Das,
C.
Pat
var
dhan
,
“
N
ew
m
ult
i
-
obje
c
t
ive
stocha
sti
c
sea
rch
t
ec
hniqu
e
for
ec
onom
ic
loa
d
dispat
ch
,
”
IEE
Proceedi
ngs
-
Gene
ration, Tr
ansm
ission and
Distributi
on
,
vol
.
145
,
no
.
6
,
pp
.
747
-
752,
1998
.
[8]
M.
A.
Abido,
“
A
novel
m
ult
io
bje
c
ti
ve
evol
u
tionar
y
a
lgori
thm
for
env
ironmen
ta
l
ec
onom
ic
power
dispatch,
”
El
e
ct
ric
Pow
er
S
yste
ms
Re
sear
ch
,
vol
.
65
,
no
.
1
,
p
p.
71
-
81
,
2003
.
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
.
4
,
A
ugus
t
2020
:
3358
-
3366
3366
[9]
D.
Kal
y
anmo
y
,
A.
Prata
p
,
S.
Agarwal
,
and
T
.
Me
y
a
riva
n
.
,
“
A
fast
and
eliti
st
m
ult
i
-
obj
ec
t
ive
genetic
algorith
m
:
NSGA
II,
”
IEEE
Tr
ansacti
on
on
Ev
olutionar
y
Co
mpu
tat
ions
,
vo
l.
6,
no
.
2
,
pp
.
182
-
197,
2002
.
[10]
M.
A.
Abido.
,
“
Envi
ronm
ent
a
l/Economic
Pow
e
r
Dispatc
h
Us
ing
Multi
objecti
v
e
Evol
uti
on
ar
y
Algorit
hm
s,”
IE
E
E
Tr
ansacti
ons on Power
Syst
ems
,
vol.
18
,
no
.
4
,
pp
.
1529
-
1537
,
20
03
.
[11]
M.
A.
Abido
.
,
“
A
nic
hed
Pare
to
genetic
a
lgori
th
m
for
envi
ronm
ent
a
l/
e
conomic
power
dispatch,
”
E
le
c
tr.
Powe
r
Syst
.
Re
s.
,
vo
l. 2
5,
no
.
2
,
pp
.
97
–
105,
2003
.
[12]
B.
Zha
o
and,
and
Y.J
.
Cao,
“
Multi
ple
obj
ective
p
ar
ti
cle
sw
arm
opti
m
iz
at
ion
t
ec
hn
iq
ue
for
ec
onom
ic
loa
d
di
spat
ch,”
J.
Zhe
ji
ang
Univ
ersity
S
ci
en
ce
,
v
ol.
6A
,
no
.
5
,
pp
.
420
–
427,
2005.
[13]
T.
Nikam,
H.
D
oagou
-
Mojarra
d
,
“
Multi
obje
ctive
ec
onom
ic
/
emissi
on
dispat
ch
b
y
m
ult
iobj
ectiv
e
θ
-
par
ticle
sw
arm
opti
m
iz
ation,
”
I
ET
Gene
ration
,
Tr
ansm
i
ss
ion
&
Distributi
on
,
vol
.
6
,
no
.
5
,
pp
.
36
3
–
377,
2012
.
[14]
B.
V.
Babu,
and
B.
Anbara
su,
“
Multi
-
object
ive
diffe
ren
ti
a
l
evolution
(MO
DE):
an
evol
ut
iona
r
y
al
gorit
hm
for
m
ult
i
-
obje
c
ti
ve
opti
m
i
za
t
ion
probl
ems
,
”
3rd Int
ernati
o
nal
Conf
ere
nce
on
(CIRAS
-
2005),
Singapor
e, p
p.
13
–
16
,
2005
.
[15]
A.
M.
Jubril
,
O.
A.
Kom
ola
fe
a
nd
K.
O.
Al
awode
.
,
"S
olvi
ng
M
ult
i
-
Obje
ctive
E
conomic
Dispat
ch
Problem
Via
Sem
ide
fini
t
e
Pro
gra
m
m
ing,
" i
n
I
EE
E
Tr
ansacti
o
ns on
Powe
r
Sys
te
ms
,
vol
.
28
,
no
.
3
,
pp
.
2056
-
20
64,
Aug.
2013.
[16]
T.
Niknam.
R
Aziz
pana
h
.
,
Mohs
en
Za
re,
Bahman
Bahmani
-
Firouzi
.
,
“
Reser
ve
Constrai
n
ed
D
y
nam
ic
Envi
ronm
ent
a
l/Economic
Dispa
tc
h:
A
New
Multi
objecti
v
e
Self
Adapti
ve
L
ea
rn
i
ng
Bat
Algorit
h
m
,
”
IEE
E
S
ystem
s
Journal
,
vo
l
.
7,
no.
4
,
pp
.
763
-
7
76,
2013
.
[17]
S.
Ma,
Y.
W
an
g
and
Y.
Lv.
,
"M
ult
iobj
e
ct
iv
e
Envi
ronm
ent
/
Econom
ic
Pow
er
Dispatc
h
Us
ing
Evol
uti
on
ar
y
M
ulti
obje
c
ti
ve
Optimi
za
t
ion,
"
in
I
EE
E
Ac
c
ess
,
vol
.
6
,
p
p.
13066
-
13074
,
2018.
[18]
Niti
sh
chopr
a,
Gourva
kum
ar,
shivani
m
eha
ta.,
“
Mu
lti
-
obje
ctiv
e
Ec
onom
ic
Emiss
ion
Loa
d
Dispatc
h
using
Gr
e
y
W
olf
Optimiza
t
i
on,
” in
I
JA
E
RS
,
vol.
3
,
no
.
11
,
pp
.
26
-
33
,
2016
.
[19]
M.
A.
Abido.
,
“
Multi
objecti
v
e
E
volut
iona
r
y
Alg
orit
hm
s
for
El
ectric
Pow
er
Dispatc
h
Problem,
”
I
EE
E
Tr
ansacti
o
n
on
Ev
o
lut
ionary
Computati
o
ns
,
v
ol.
10
,
no
.
3
,
Jun
.
2006
.
[20]
R.
Sw
ai
n,
P.Sarka
r,
KC.Mehe
r
,
CK.
Chanda
.
,
“
Populat
ion
var
ia
nt
diff
ere
nt
ia
l
evol
uti
on
base
d
m
ult
iobj
ec
t
iv
e
ec
onom
ic
emiss
ion
loa
d
dispatch,
”
Int
Tr
ans E
lect
r E
nerg
Syst
.
,
vo
l.
27
,
no
.
10
,
pp
.
1
-
25,
Apr.
2017.
[21]
M.
Sakawa
,
H.
Ya
no,
T
.
Yum
in
e.
,
“
An
int
er
acti
ve
fuz
z
y
sa
ti
sfi
cing
m
et
hod
for
m
ult
iobj
ectiv
e
l
i
nea
r
progr
amm
ing
proble
m
s a
nd
it
s a
ppl
ic
a
ti
on
,
”
IE
EE
Tr
ans.
S
yst.
,
Man,
Cyb
ern
.
,
v
ol.
SM
C
-
17,
no.
4
,
pp
.
654
–
661
,
1987.
[22]
R.
Fuller
and
C.
Car
lsson,
“
Fuzz
y
m
ultiple
c
rit
er
ia
d
ec
ision
m
aki
ng:
Re
ce
nt
developm
ent
s,”
Fuzzy
Se
t
Syst
.
,
vol.
78
,
pp
.
139
–
153,
1996
.
[23]
Dun
-
wei
Gong,
Yong
Zha
ng
,
Cheng
-
li
ang
Qi.
,
“
Envi
ronm
ent
a
l/
e
conomic
power
disp
at
ch
using
a
h
y
br
i
d
m
ult
i
-
objecti
v
e opti
m
iz
ation
al
g
orit
hm
,
”
Elec
tri
c
al
Pow
er
and
En
ergy
Syst
ems
,
vo
l.
32
,
no
.
6
,
pp
.
6
07
–
614,
2010
.
[24]
H.
Abdulla
h
k
ubba,
and
S.
Sabah
sam
i,
“
A
Modifie
d
Strengt
h
Pare
to
Evol
uti
on
ar
y
A
lgori
thm
2
base
d
Envi
ronm
ent
a
l
/
Ec
onom
ic
Pow
e
r
Dispatc
h
,
”
Jou
rnal
of Engi
ne
ering,
vo
l.
20,
no.
3,
pp
.
105
-
123
,
2014.
[25]
Kam
al
Manda
l,
and
N.
Chakrabor
t
y
,
“
Eff
e
ct
o
f
Control
Para
m
et
ers
on
Diffe
r
e
nti
al
Evol
ut
ion
base
d
Com
bine
d
Ec
onom
ic
Emis
sion
Dispatc
h
with
Valve
-
Poi
nt
Loa
ding
and
Tra
nsm
ission
Loss,”
Inte
rnat
i
onal
Journal
o
f
Eme
rging Ele
ct
ric Powe
r S
yste
m
s
,
vol. 9, no. 4, p
p.
1
-
18
,
2008
.
[26]
M.
Basu,
“
E
con
om
ic
envi
ronm
e
nta
l
d
ispatch
usi
ng
m
ult
i
-
object
i
ve
diff
ere
nt
ia
l
e
volut
ion,”
Appl
i
ed
Soft
Computi
ng
,
pp.
2845
-
2853
,
Dec
.
2010.
BIOGR
AP
H
I
ES
OF
A
UTH
ORS
K.
Rajesh
cur
ren
tly
pursuing
Ph.D
degr
ee
at
JN
TUA,
Ananth
apur
amu,
Andhra
Prade
sh,
India.
His
are
a
of
int
ere
st
inc
lud
es
opti
m
iz
ation
tech
nique
s
on
Ec
onom
ic
loa
d
dispat
ch
,
Rene
wabl
e
ene
rg
y
sour
ce
s
a
nd
Dem
and
side
m
ana
gement.
Dr.
N.
Visali
is
rec
e
ive
d
Ph
.
D fr
om
JN
TUA,
An
at
hapur
amu.
She
is pr
ese
n
tly
wor
king
as
Profess
o
r
and
PR
INCIP
A
L,
JN
TUACE,
Kali
kiri,
Ch
it
to
or
district,
And
hra
Prade
sh
,
In
dia
.
Her
area
o
f
int
er
est
in
cl
udes
Ec
onom
ic
loa
d
sche
duli
ng
using
opti
m
iz
a
ti
on
tec
hnique
s,
Con
trollers
for
FA
CTS
Devic
es,
opt
imiz
at
ion
techniqu
es
in
pl
ac
emen
t
of
ca
pa
ci
tor
,
D
emand
side
m
an
agem
ent
,
DG
.
Evaluation Warning : The document was created with Spire.PDF for Python.