TELKOM
NIKA Indonesia
n
Journal of
Electrical En
gineering
Vol.12, No.6, Jun
e
201
4, pp. 4609 ~ 4
6
1
6
DOI: 10.115
9
1
/telkomni
ka.
v
12i6.543
7
4609
Re
cei
v
ed
De
cem
ber 2
8
, 2013; Re
vi
sed
F
ebruary 28,
2014; Accept
ed March 1
4
, 2014
Emergency Resource Scheduling Problem Based on
Improved Particle Swarm Optimation
Wu Kaijun
1
*, Shan Yaz
h
ou
1
, Lu Huai
w
e
i
2
1
School of Elec
tronic an
d Infor
m
ation En
gi
ne
erin
g, LanZ
h
o
u
Jiao T
ong Uni
v
ersit
y
,
Lan Z
h
o
u
73
00
70, Chi
n
a
2
School of Ph
ysics and Soft
w
a
re Eng
i
n
eeri
n
g, LanZ
ho
u Jia
o
T
ong Unversit
y,
Lanz
ho
u 730
0
70, Chi
n
a
*Corres
p
o
ndi
n
g
autnor, e-ma
i
l
:
w
k
j
@
mai
l
.lzjt
u
.cn
A
b
st
r
a
ct
Emer
ge
ncy res
ource sch
ed
uli
ng is a kin
d
of NP
combi
natio
n prob
le
m w
h
ic
h possess
es i
m
p
o
rtant
practica
l valu
e. In order to overco
me the pr
o
b
le
ms suc
h
as lon
g
co
mputi
n
g time a
nd e
a
s
y
to fall into loc
a
l
best for tra
d
iti
ona
l h
euristic
opti
m
i
z
at
ion
al
gorith
m
,
an I
m
prove
d
Partic
l
e
Sw
arm Opti
mi
z
a
t
i
o
n
a
l
gor
i
t
hm
(IPSO) is prop
osed
,
the
al
go
rithm uses
the
rand
o
m
icity
a
nd st
a
b
le
tend
entio
usn
e
ss c
haracter
i
stics
of
clou
d mo
del, a
dopts different inerti
a
w
e
ig
ht
g
ener
ating
meth
ods i
n
d
i
fferent
grou
ps, the
se
archi
ng
abi
lity
of
the a
l
gor
ith
m
i
n
loc
a
l
an
d g
l
oba
l situ
atio
n i
s
bal
anc
ed
effectively. In t
h
e
pa
per,
th
e a
l
go
ri
thm
i
s
u
s
ed to
solve th
e e
m
er
gency res
ourc
e
sche
dul
ing
p
r
obl
em, th
e mathe
m
atic
mod
e
is estab
lish
e
d
an
d the so
lut
i
o
n
alg
o
rith
m is d
e
v
elo
ped. T
he s
i
mu
lati
on res
u
lt
s of exa
m
pl
e s
how
that the al
gorith
m
h
a
s fa
ster search s
p
eed
and stron
ger o
p
timi
z
a
ti
on ab
il
ity than GA and PSO algorith
m
.
Ke
y
w
ords
: e
m
erge
ncy resour
ce sched
uli
ng,
par
ticl
e sw
arm
opti
m
i
z
at
ion, cl
oud
mo
del
Copy
right
©
2014 In
stitu
t
e o
f
Ad
van
ced
En
g
i
n
eerin
g and
Scien
ce. All
rig
h
t
s reser
ve
d
.
1. Introduc
tion
With the
ra
pid d
e
velop
m
ent of
so
cial e
c
on
omy and
in
cre
a
s
ing
level
s
of urb
an
mode
rni
z
atio
n, all
kind
s o
f
emergen
cie
s
h
a
ve
eno
rmous influ
e
n
c
e
s
o
n
urb
a
n
functio
n
s.
Some
unexpe
cted
events will seriou
sly
affe
ct
the
life
saf
e
ty of the p
eople,
su
ch
as
earth
qua
kes,
tsunami
s
, g
a
s
le
ak.
Lo
sse
s
stemme
d from ineffe
ctive eme
r
g
e
n
c
y logi
stics a
ccount for 15%
to
20% of the e
n
tire casualty and fina
nci
a
l
losse
s
cau
s
e
d
by su
dde
n
outbursts of n
a
tural
cala
mities
and hum
an
-contrived di
sa
sters.
Emerge
ncy reso
urce
sch
e
duling i
s
a speci
a
l
goo
ds
transpo
rting
activity which
aims at
providin
g e
m
erge
ncy
re
so
urce fo
r e
m
ergen
cie
s
, ma
x
i
mizing
efficie
n
cy a
nd
mini
mizing
da
ma
ges
cau
s
e
d
by
di
sa
st
er
s.
E
m
e
r
gen
cy
re
sou
r
ce s
c
he
du
ling
is
an im
po
rta
n
t co
mpon
ent
in d
ealing
wit
h
accide
nts,
wh
ich i
n
time
weigh
s
g
r
eat v
a
lue i
n
redu
ci
ng lo
sse
s
, p
r
eventing
se
conda
ry di
sa
sters
and maintai
n
i
ng so
cial
stab
ility.
When
a natu
r
al di
sa
ster-stricken
spot is
in nee
d of
countermea
s
u
r
es from
eme
r
gen
cy
resou
r
ce tra
n
sp
orting
sy
stem, eme
r
g
ency vehi
cle
s
should
del
iver eme
r
ge
ncy re
so
urce
to
deman
d sp
ot within the least of time. Emergen
cy
resou
r
ce sch
edulin
g is a typical ca
se
of
optimizin
g co
mbination, a tough is
su
e concerni
ng NP
. It contains g
r
eat co
mplexi
ties. Questio
n
s
of
the
simil
a
r kind are hot studied
by diff
erent b
r
a
n
ch
es of
sci
en
ce,
su
ch a
s
o
p
e
r
ations
re
sea
r
ch
,
applie
d math
ematics, com
puter
appli
c
at
ion, graph th
eory a
nd
co
mmuni
cation
& tran
spo
r
tation,
etc. The time it takes to be
solved is g
r
o
w
i
ng exp
one
ntially as its d
i
mensi
on is e
x
pandin
g
, whi
c
h
is ha
rd
to
be ha
ndle
d
with traditi
onal m
a
the
m
at
ic m
e
tho
d
s. In
re
ce
nt years, h
euri
s
tic
optimizatio
n algorith
m
s
su
ch a
s
Gen
e
tic Algorit
hm,
Ant Colony Algorithm a
nd
Particle Swarm
Optimizatio
n
have been
wi
dely applied t
o
solving such kind of pro
b
lems [1
-3]; however, the
s
e
algorith
m
s all
have
drawba
cks li
ke
lon
g
comp
uti
ng ti
me, ea
sy to f
a
ll into
stag
n
a
tion a
nd
un
able
to do fu
rthe
r co
mputing.
Therefore,
h
o
w to
con
s
truct o
p
timize
d
algo
rithm
si
mple in
form
ation
and high in
optimizin
g preci
s
ion enj
oys gre
a
t si
gni
fican
c
e in solving emerg
ency re
so
urce
sched
uling p
r
oblem.
Particle
swa
r
m o
p
timizati
on (PSO
) i
s
a set of ne
w intelligent
o
p
timization
al
gorithm
s
whi
c
h i
s
simp
le calculation,
fast
conve
r
g
ence a
nd
ea
sy to implem
e
n
t, etc [4]. Th
e alg
o
rithm
h
a
s
been a
pplied
to VRP [5-6] and a
c
hieve
d
very good re
su
lts, but it exist such pro
b
lems a
s
bei
ng
Evaluation Warning : The document was created with Spire.PDF for Python.
ISSN: 23
02-4
046
TELKOM
NI
KA
Vol. 12, No. 6, June 20
14: 4609 – 4
616
4610
easy to stag
nate and fall
into local opti
m
um. Theref
ore, in this p
aper, IPSO is put forward
to
solve the em
erge
ncy reso
urce sche
duli
ng pro
b
lem.
The re
sult of experim
ents i
ndicates that
the
algorith
m
ca
n
efficiently sol
v
e the proble
m
.
2. The Math
ematic Mod
e
of Emergen
c
y
Resourc
e
Scheduling Problem
2.1. Problem Descrip
tion
There are ple
n
ty of emerg
ency vehi
cle
s
in
the eme
r
g
ency rescu
e
system. Th
e vehicle
s
will
set off from the
rescue
center
and deliver
go
ods to several
cri
s
i
s
-attack
ed sites (quantity=R).
After delivery
,
the vehicles will
return to the center. The maximu
m capacity for each vehi
cl
e is
k
P
(
K
k
,...,
2
,
1
). Find
drivi
ng route
s
fro
m
the
cent
er
to ea
ch
de
mandin
g
site
meeting
th
e
following requirem
ents
[7]:
1) Overl
oad i
s
forbid
den.
2) Delive
r
y must be finishe
d
as soon a
s
possibl
e.
3) Suppli
e
s required by ea
ch dem
andi
n
g
site mu
st be delivere
d
.
4) the cost of emergen
cy c
enter mu
st be
at minimum.
Supplem
enta
r
y conditio
n
s:
1) Adeq
uate
vehicle
s
are available in e
m
er
g
e
n
c
y ce
nter, and the
cap
a
city of each h
a
s
been give
n in
the discussio
n
;
2) T
here i
s
only o
ne e
m
erg
e
n
c
y ce
nter a
nd it
s location
ha
s b
een
give
n in th
e
discu
ssi
on;
3) Th
ere i
s
o
n
ly one type
vehicle in th
e
cent
e
r
. Each cri
s
i
s
-atta
cked
site is
se
rved by
only one vehi
cle. Every site must
be
co
vered by the driving route;
4) the
g
r
o
s
s dema
nd
of
each d
e
man
d
ing
site
alo
ng the
ro
ute
ca
nnot
su
rp
ass the
cap
a
city of the vehicle;
5) In order t
o
simplify the case, only the
time consumed on road will be
considered,
irres
p
ec
tive of the s
e
rvice time.
2.2. Mathem
atic Mode
Here is the m
a
thematic m
o
de built ac
co
rding to the de
scription a
b
o
v
e:
(1)---obj
ectiv
e
function (2
)-(11
)
---con
st
raint co
nditio
n
s
ijk
ij
M
ks
is
j
t
ijk
f
k
M
k
k
y
d
c
c
x
z
0
min
(1)
N
j
y
S
iM
k
ijk
,
1
(2)
S
h
M
k
y
y
s
i
hik
s
i
ihk
,
,
0
(3)
0
0
N
jM
k
jk
y
(4)
M
k
Q
q
y
k
j
s
iN
j
ijk
,
(5)
0
Q
q
y
j
M
kS
iN
j
ijk
(6)
jik
t
=
ijk
ij
v
d
(7)
M
k
S
j
i
y
t
t
t
jik
jik
jk
ik
,
,
,
(8)
Evaluation Warning : The document was created with Spire.PDF for Python.
TELKOM
NIKA
ISSN:
2302-4
046
Em
ergency Reso
urce Sche
duling Proble
m
Based
on Im
prove
d
Part
icle Swa
r
m
…
(Wu Kaiju
n)
4611
M
k
N
i
l
t
i
ik
,
,
(9)
M
k
S
j
i
y
ijk
,
,
,
1
,
0
(10)
M
k
x
k
,
1
,
0
0
(11)
In the mode:
a)
ij
d
refers to the dista
n
ce b
e
twee
n nod
e
i
and
no
de
j
(
,
2
,
1
,
0
,
S
j
i
). Whe
n
0
,
j
i
,
ij
d
refers to the emergen
cy cente
r
;
b)
ijk
v
refers to the averag
e sp
eed of vehicl
e
k
from nod
e
i
to node
j
;
c)
ik
t
refers
to the time vehic
l
e
k
need
s to go to the deman
d
i
ng site
i
;
d)
ijk
t
refers
to the time vehic
l
e
k
need
s to go from no
de
i
to
node
j
;
e)
i
l
refers
to the lates
t
time of arrival of vehic
l
e
k
;
f)
k
n
refers to the total numbe
r of deman
din
g
sites that vehicl
e
k
needs
to serve. Wh
en
0
k
n
, it means tha
t
the vehicle doe
s
not parti
cipate in the
missi
on.
g)
k
R
refers to the set of dema
nding
sites
se
rved by the vehicl
e;
h)
f
k
c
refers
to the fixed c
o
s
t
of vehic
l
e
k
whe
n
driving;
i)
t
ijk
c
refers
to the unit c
o
s
t
of vehic
l
e
k
from node
i
to node
j
;
j)
k
Q
refers to th
e loadin
g
ca
p
a
city of vehicl
e
k
;
k)
i
q
refers to the deman
d of node
i
;
l)
N
refers
to the s
e
t of c
r
is
is-attack
ed
s
i
tes
.
}
,...,
2
,
1
|
{
R
r
r
N
;
m)
M
refe
rs to the set of eme
r
gen
cy vehicl
es.
}
,...,
2
,
1
|
{
K
k
k
M
;
n)
S
refers to the sum of nod
es.
O
N
S
.
0
k
x
and
ijk
y
are defi
ned a
s
follows:
0
1
e
m
e
r
g
en
c
y
veh
i
cl
e k
i
s
i
n
u
s
e
0o
t
h
e
r
w
i
s
e
k
x
;
1
v
e
h
i
c
l
e
k
dr
i
v
e
f
r
o
m
n
ode
i
t
o
n
ode
j
0o
t
h
e
r
w
i
s
e
ij
k
ij
y
,
and
。
3. Impro
v
ed
Particle S
w
a
r
m Optimiz
a
tion
3.1. Particle S
w
arm Opti
miz
a
tion
Inspired by
bird
s sea
r
chi
ng food i
n
real
world, P
S
O is p
r
op
o
s
ed
by Kenn
edy and
Eberh
a
rt i
n
1
995, a
s
a
ne
w-b
o
rn al
gorit
hm b
a
sed
on
group
theo
ry
, the b
a
si
c
pri
n
cipl
e of
PSO is
to fix individ
ual actio
n
throug
h sh
ari
n
g informatio
n
and individu
al’s o
w
n exp
e
rien
ce a
m
o
ng
grou
ps to get
the optimal solution of the probl
em.
The mathem
aticl descripiti
on of PSO is: group co
nst
i
tuted by
n
part
i
cle
s
sea
r
ch in
D
dimen
s
ion
sp
ace.
The
lo
ca
tion of the
i
pa
rticle
is expre
s
sed
as:
)
,...,
,
(
2
1
iD
i
i
i
x
x
x
X
, an
d the
corre
s
p
ondin
g
spee
d i
s
e
x
presse
d a
s
:
)
,...,
,
(
2
1
iD
i
i
i
v
v
v
V
, and
i
P
=
)
,...,
,
(
2
1
iD
i
i
p
p
p
is
the
optimal
lo
cati
on sea
r
ched
by
the
i
parti
cl
e so fa
r, an
d
g
P
=
)
,...,
,
(
2
1
gD
g
g
p
p
p
is the
opti
m
a
l
locatio
n
sea
r
che
d
by the whol
e parti
cl
es rig
h
t now. The upd
ate equation
s
of the
d
d
i
me
ns
io
nal
spe
ed
1
k
id
v
and lo
cation
1
k
id
x
of particle
i
at the
k
+1
lteration a
r
e e
x
presse
d as f
o
llows:
Evaluation Warning : The document was created with Spire.PDF for Python.
ISSN: 23
02-4
046
TELKOM
NI
KA
Vol. 12, No. 6, June 20
14: 4609 – 4
616
4612
)
(
)
(
2
2
1
1
1
k
id
k
gd
k
id
k
id
k
id
k
id
x
p
r
c
x
p
r
c
wv
v
(12)
1
1
k
id
k
id
k
id
v
x
x
(13)
Whe
r
e
w
is the inertia
weight;
1
c
and
2
c
are
the a
c
cele
ration
con
s
ta
nt;
1
r
and
2
r
a
r
e
t
he
rand
om ind
e
p
ende
nt numb
e
rs b
e
twe
en
0 and 1.
3.2. Cloud M
odel
Proposed by
academi
c
ian
Li Deyi [8], the cl
oud model combines th
e view of probability
theory and fu
zzy set theory, descri
b
e
s
the unce
r
ta
in
ty of the con
c
ept of the natural lan
gua
ge
and provide
s
an effective mean
s for the informat
ion
processin
g
b
y
adopting bo
th the qualitative
and qu
antitative methodolo
g
y.
Normal clo
u
d
model is the rand
om set with
stable ten
den
cy followi
ng norm
a
l distribution
rule. The
r
e
are such three normal di
stributio
n rul
e
s hid
den in
the normal
clou
d model
as
)
,
,
(
2
2
3
Hn
En
Ex
N
, three num
e
r
ical fe
ature
s
of cloud a
r
e
expre
s
sed b
y
expected v
a
lue
Ex
,
entropy
En
and ul
tra-e
n
tropy
Hn
res
p
ec
tively.
3.3. Cloud Adaptiv
e
Adjustmen
t
Stra
teg
y
The si
ze of p
a
rticle
swarm
is set as
N
, the fitness value of particl
e
i
X
is
i
f
after
iterating
k
times; the avera
ge fitness value is
N
i
i
avg
f
N
f
1
1
; when the avera
ge v
a
lue of the
fitness value
better than
avg
f
is
'
avg
f
and
infe
rior to
avg
f
is
'
'
avg
f
, the fitness val
ue of t
h
e
optimal particle is
min
f
; let
'
avg
f
Ex
,
1
min
'
)
(
c
f
f
En
avg
,
2
c
En
He
,
'
En
=no
r
mrnd(
En
,
He
),
w
is cal
c
ulate
d
as form
ula (14),
Wherein,
as a co
ntrol
para
m
eter.
"
"
'
2
'
2
'
8
.
0
)
)
(
2
)
(
exp(
*
5
.
0
8
.
0
2
.
0
avg
i
avg
i
avg
i
avg
i
f
f
f
f
f
En
Ex
f
f
f
w
(14)
4. Solution to Emergenc
y
Resource Scheduling
Problem Bas
e
d on IPSO
4.1. Coding
Strateg
y
of
Particles
The
p
o
sition
of
parti
cle correspon
ding with
t
he an
swe
r
to the q
u
e
s
tion is th
e
key
idea of
PSO. In
the paper
, a ne
w particl
e enco
d
ing of 3-dime
nsi
onal
vectors X is con
s
tructe
d to
emergen
cy re
sou
r
ce sche
d
u
ling pro
b
lem
.
In
the vector
X
, the first dimensi
on
r
X
of particles is
rescu
e
ce
nter informati
on; the se
con
d
dimensi
on
s
X
of particle
s
is veh
i
cle inform
ation; the third
dimen
s
ion
t
X
of particle
s
is traveling di
stan
ce of the vehi
cle.
4.2. Decodin
g
Stra
teg
y
of
Particles
To get to the travel path of
the vehicle o
r
de
r,
t
X
must be
adjuste
d. Adjustment fun
c
tion
can
be
got
a
c
cordi
ng to
t
he
size
seq
u
ence of ve
ct
or
t
X
, that is to
say, findin
g
out
t
X
o
f
t
he
vehicle fo
r
cu
stome
r
i
first, and the
n
sort
ed from
sm
all
to larg
e in
accordan
ce
wit
h
t
X
, thus
the
driving path o
r
de
r of vehicl
e
i
is determi
ned. For exa
m
ple, if
there
are 10
cu
sto
m
ers, 2 re
scue
cente
r
s, one of
re
scue cen
t
ers ha
s
2
ve
hicle
s
,
t
he
other
ha
s 3 ve
h
i
cle
s
. If the p
o
sition ve
cto
r
X
of a parti
cle i
s
shown a
s
Table 1, the
n
vector
X
of the positio
n after adj
ustme
n
t
is sho
w
n
a
s
Table 2. So the co
rrespon
ding v
ehi
cle routing
s
are a
s
follows:
Evaluation Warning : The document was created with Spire.PDF for Python.
TELKOM
NIKA
ISSN:
2302-4
046
Em
ergency Reso
urce Sche
duling Proble
m
Based
on Im
prove
d
Part
icle Swa
r
m
…
(Wu Kaiju
n)
4613
rescu
e
ce
nter 1: vehicle 1: 1
→
2
v
ehic
l
e 2: 4
→
8
rescu
e
ce
nter 2: vehicle 3: 3
v
ehic
l
e 4: 6
→
7
→
5
v
ehic
l
e 5: 10
→
9
T
able 1. Vector
X
before Adj
u
sti
n
g
customer
1 2 3 4
5 6 7
8 9 10
r
X
1 1 2 1
2 2 2
1 2 2
s
X
1 1 3 2
4 4 4
2 5 5
t
X
0.3 1.2 0.6 2.
3
4.8 1.1 2.9
3.9 2.8 1.7
T
able 2. Vector
X
after Adjusting
customer
1 2 3 4
5 6 7
8 9 10
r
X
1 1 2 1
2 2 2
1 2 2
s
X
1 1 3 2
4 4 4
2 5 5
t
X
1 2 1 1
3 1 2
2 2 1
4.3.
Process Des
c
ription of
Algorithm
Step 1
:,
initialize species
set the si
ze
n
of each
particl
e swarm an
d alg
o
rithm
para
m
eter (a
ccelerating
fa
ctor
1
c
2
c
、
, maximum numbe
r of iteration
s
max
N
an
d dimen
s
ion
of
particl
es
D
, individual extrem
um
pbest
and glob
al extremum
gbest
).
Step 2
:
E
a
ch
part
i
cle
i
X
in the particl
e swa
r
m is exe
c
ute
d
as follo
ws:
a) Update th
e sp
eed a
nd
positio
n
i
X
with the formula
(9) an
d form
ul
a (10
)
, amo
n
g
of
them,
w
is cal
c
ulated with th
e formula (11
)
;
b) Cal
c
ul
ating
the fitness value
i
f
of
i
X
;
c
)
If
i
X
is better than the fitness value of
pbest
, then
pbest
sho
u
ld
be upd
ated to the
curre
n
t positi
on of
i
X
;
d) If
i
X
is
better than th
e fitn
ess valu
e of
gbest
, then
gbest
sh
ould
be u
pdate
d
t
o
the
curre
n
t positi
on of
i
X
;
Step 3
:
Ju
dg
e wh
ether th
e cu
rrent iteration num
be
r
G
is eq
ual to
or g
r
eate
r
th
an the
iteration n
u
m
ber
max
N
, if not sat
i
sfied, return
Step2, other
wise, the current optimal
solution
s is
output.
5. Analy
s
is o
f
the Simulation Results of Example
5.1. Example 1
A cri
s
is-attacked a
r
e
a
ha
s 8 deman
d si
tes nee
ding
reso
urce d
e
livered f
r
om em
erge
ncy
resou
r
ce cent
er. There are
5 vehicles. T
he loadin
g
ca
pacity is 8 ton for each vehicle. The fixed
co
st of
each
is
80
yuan.
Th
e
averag
e spee
d i
s
20km/h
o
u
r
.
The avera
g
e
driving
co
st
is
10yuan/
km. T
he di
stan
ce b
e
twee
n site
s,
the dem
and
i
q
(unit: ton)
an
d the late
st a
rrival time
i
l
Evaluation Warning : The document was created with Spire.PDF for Python.
ISSN: 23
02-4
046
TELKOM
NI
KA
Vol. 12, No. 6, June 20
14: 4609 – 4
616
4614
(unit: minute
)
)
8
,...,
2
,
1
(
i
are di
spl
a
ye
d in Tabl
e 4 (0 rep
r
e
s
ent
s
for the cente
r
, 1-8 represe
n
ts
for the s
i
tes
)
.
Compute
the
exampl
e
with
GA,
PSO
and IPSO resp
ectively on the sam
e
comp
uter. Th
e
optimal sche
duling pla
n
is
that 3 cars pa
rticipate
in th
e task. Th
e driving route, di
stan
ce an
d cost
are
sh
own in
Table
4. O
b
jective fun
c
ti
on
chan
ge
s
according
to
iteration
and
its tend
en
cy is
recorded in Fi
gure 1.
Table 3. The
Data of Di
stri
but
ion Center and Di
sa
ster Locatio
n
Cr
isis-
a
ttacked
sites
Cr
isis-
a
ttacked si
tes
0
1
2
3
4 5 6 7 8
0
0
4
6
7.5
9
20 10 16
8
1 4
0
6.5
4
10
5
7.5
11
10
2
6
6.5 0 7.5
10
10
7.5
7.5
7.5
3
7.5
4
7.5
0
10
5 9 9
15
4
9
10 10 10
0
10
7.5
7.5
10
5
20
5 10 5
10
0
7
9
7.5
6
10
7.5
7.5 9
7.5 7
0
7
10
7
15
11
7.5 9
7.5 9
7
0
10
8
8
10
7.5 15
10
7.5 10
10
0
i
q
3
3
1
3 2 4 1 4
i
l
40
60
30
80 80 50 90 30
Table 2. The
Optimal Sch
e
duling Schem
e of Example 1
vehicles driving
routes
driving
time
driving time
total cost
vehicle1
0—1—3—
2—0
64.5
21.5
295
vehicle2
0—6—4—
0
79.5
26.5
345
vehicle3
0—8—5—
7—0
118.5
39.5
475
total
262.5
87.5
1115
Figure 1. Performa
nce Co
mpari
s
o
n
s of
GA, PSO and IPSO
5.2. Example 2
A cri
s
is-atta
cked a
r
ea
has
20 de
mand
si
tes
needi
ng resource d
e
live
r
ed from
emergen
cy reso
urce
ce
nter. Th
er
e
are 4 vehi
cle
s
.
The l
oadi
ng
ca
pa
city is
20 ton f
o
r
e
a
ch
vehicle. The
fixed cost of
eac
h is 10
0 yuan. The av
erag
e sp
eed
is 25km/hou
r. The avera
ge
driving
co
st i
s
1
2
yua
n
/km
.
The
co
ordi
n
a
te of
e
m
erg
ency
re
scue
cente
r
i
s
0 p
o
int (3km,
4km),
positio
n co
ordinate
s
of de
mand
sites
(unit: km), the
deman
d
i
q
(unit: ton)and th
e
latest arrival
0
500
1
000
1500
1100
1150
1200
1250
1300
1350
1400
1450
i
t
er
at
i
o
n
to
ta
l
c
o
s
t
:
y
u
a
n
GA
PS
O
IP
S
O
Evaluation Warning : The document was created with Spire.PDF for Python.
TELKOM
NIKA
ISSN:
2302-4
046
Em
ergency Reso
urce Sche
duling Proble
m
Based
on Im
prove
d
Part
icle Swa
r
m
…
(Wu Kaiju
n)
4615
time
i
l
(unit: minute)
(1
,
2
,
.
.
.
,
2
0
)
i
are di
spl
a
yed in T
abl
e 5 (0
represents fo
r the
center, 1
-
20
rep
r
e
s
ent
s for the site
s).
Table 5. The
Data of Emergen
cy Cente
r
and Emerg
e
n
cy Lo
cation
order
number
x
coordinates
y
coordinates
demand
latest
arri
v
a
l
time
order
number
x
coordinates
y
coordinates
demand
latest
arri
v
a
l
time
0 3
4
1 3.1
7.6
4
60
11
8.4
8
2
30
2 1
8.6
2
15
12
4.7
9.3
1
30
3 6.3
5.7
1
60
13
1.4
6.8
2
15
4 2
3.3
1
45
14
0.3
6.3
2
30
5 5.3
6.6
5
10
15
3.5
0.5
3
60
6 6
9.8
4
60
16
2.4
1.6
4
50
7 0.6
9.4
2
30
17
8.2
3.7
4
45
8 3
4.5
2
10
18
7.3
8.4
3
60
9 1.9
9.8
3
20
19
8.5
5.2
1
50
10 3.7
7.1
2
45
20
7.8
8.8
3
20
Table 6. The
Optimal Sch
e
duling Schem
e of Example 2
vehicles driving
routes
driving
time
driving time
total cost
vehicle1
0—5—20
—11—
19—17—1
8
—3
—0
56.42
23.51
382.12
vehicle2
0—8—13
—14—
4—16—15
—0
47.55
14.8
277.55
vehicle3
0—2—7—
9—12
—6—10—
1—0
46.65
19.44
333.24
total
150.62
57.74
992.91
Comp
ute th
e
example
with
GA, PSO a
n
d
IPSO re
sp
ectively on the
same
com
pute
r
. The
optimal sche
duling pla
n
is
that 3 cars pa
rticipate
in th
e task. Th
e driving route, di
stan
ce an
d cost
are
sh
own in
Table
6. O
b
jective fun
c
ti
on
chan
ge
s
according
to
iteration
and
its tend
en
cy is
recorded in Fi
gure 2.
Figure 2. Performa
nce Co
mpari
s
o
n
of GA, PSO and IPSO
It is o
b
viou
s f
r
om th
e
re
sul
t
of the t
w
o
e
x
perime
n
ts th
at the
algo
rithm p
r
op
osed
in th
e
thesi
s
ca
n d
e
tect the opti
m
al solutio
n
to em
erg
e
n
cy reso
urce sche
duling
problem swiftly and
acc
u
rately. I
t
s
effic
i
enc
y
is
better than that
of GA a
nd PSO. It is a new meth
od to solve the
emergen
cy re
sou
r
ce sche
d
u
ling problem
.
0
50
0
1
000
15
00
80
0
10
00
12
00
14
00
16
00
18
00
20
00
22
00
it
e
r
a
t
i
o
n
to
t
a
l c
o
st:
y
u
a
n
GA
PS
O
IP
S
O
Evaluation Warning : The document was created with Spire.PDF for Python.
ISSN: 23
02-4
046
TELKOM
NI
KA
Vol. 12, No. 6, June 20
14: 4609 – 4
616
4616
6. Conclusio
n
To solve the
proble
m
of NP in the emerg
e
n
c
y re
sou
r
ce sche
duling, an i
m
prove
d
particl
e
swarm optimatio
n
algorith
m
h
a
s bee
n a
pplie
d
to the
eme
r
g
ency
re
sou
r
ce sch
eduli
ng
is
prop
osed in
the thesi
s
. The simul
a
tio
n
results
of example indi
cate that the
algorithm
can
efficiently sol
v
e eme
r
ge
ncy re
sou
r
ce
scheduli
ng p
r
o
b
l
em throug
h t
he
comp
ari
s
o
n
of GA
an
d t
h
e
PSO algorith
m
.
Ackn
o
w
l
e
dg
ements
This
wo
rk
was fina
nci
a
lly sup
porte
d b
y
t
he Nation
al Social S
c
i
ence foun
dat
ion (No.
12CGL0
04),
sci
en
ce
a
nd techn
o
logy suppo
rt
proje
c
t
of Gan
s
u
p
r
ovince (No.
1304FK
C
A09
7
),
sci
entific
re
se
arch p
r
oje
c
t
of Gan
s
u
pro
v
ince
edu
cati
on d
epa
rtme
nt (No. 20
13
A-052
) a
nd y
outh
sci
en
ce fund
proje
c
t of Lan
zho
u
Jia
o
ton
g
University (No. 201
100
5).
Referen
ces
[1]
Ren J,
Xi
HW
, Shi
XF
. Geneti
c
Algor
ithm for
V
ehic
l
e Sc
hed
ulin
g Pro
b
l
e
m
of Cit
y Emerg
e
n
c
y
Lo
gistic
s
Distributi
on.
Jo
urna
l of Millitar
y Transportatio
n
Univ
ersity.
2011; 13(
9): 70-
73.
[2]
T
ang LS, Chen
g W
M
, Liang J, et al. Researc
h
on Ant Col
o
n
y
Cl
usterin
g
Al
gorithm for the
Problem of
Emerge
nc
y
Lo
gistics
Distrib
ut
ion
. Ra
ilw
ay T
r
ansp
o
rt and Ec
ono
my.
20
08;
30(9): 66-
69
.
[3]
W
u
KJ, W
ang T
J
.
T
he model
and
optimiz
ati
on a
l
gor
it
hm of
multi-rescu
e c
enter em
erge
n
c
y
mater
i
al
s
disp
atchin
g
w
i
t
h
time limits.
Computer En
gi
n
eeri
ng an
d App
licatio
ns.
20
12;
48(30): 19-
23.
[4]
Kenn
ed
y J
Eb
erhart R.
P
a
rticle Swar
m
Opt
i
mi
z
a
tion.
Proc
eed
ings
of IEE
E
Intern
ation
a
l
Co
nfere
n
ce
on Ne
ural N
e
tw
o
r
ks. 199
5; 1942-
194
8.
[5]
Liu Z
X
. Ve
hicl
e sched
uli
ng o
p
timizati
on in l
ogistics d
i
strib
u
tion b
a
se
d on
particle s
w
ar
m optimizati
o
n
algorithm.
Jour
nal of W
uha
n
Univers
i
ty of Scienc
e an
d T
e
chno
logy.
2
009;
32(6): 615-
61
8.
[6]
ZHANG YB V
GQ. S
t
udy
of Phy
s
ical Distribution R
outing
Optimization Problem Based
on Hy
brid PSO
Algorit
hm.
Packing e
ngi
ne
erin
g.
2007; 2
8
(5): 10-1
2
.
[7]
Guo HL. Stu
d
y
on Mu
lti-Obje
ctive Veh
i
cle
R
outin
g Pro
b
le
m in ur
ban Em
erge
nc
y Lo
gist
ics, Baod
ing
,
Heb
e
i Un
iversit
y
. 20
10.
[8]
Li D Y, M
e
n
g
HJ, Shi
XM. E
m
bershi
p
cl
ou
ds an
d mem
b
ershi
p
clo
ud
g
ener
atorsr.
Co
mp
uter R&
D
.
199
5; 32(6): 15
-20.
Evaluation Warning : The document was created with Spire.PDF for Python.