TELKOM
NIKA
, Vol. 11, No. 2, Februa
ry 2013, pp. 653~658
ISSN: 2302-4
046
653
Re
cei
v
ed Au
gust 27, 20
12
; Revi
sed
De
cem
ber 2
3
, 2012; Accepte
d
Jan
uary 11,
2013
Agent-Based Automatic Shore Operating Scheduling
for a Container Terminal
Nann
an Yan
*
, Yuan Zhou
Shan
gh
ai Marit
i
me Univ
ersit
y
266
0 Pu Do
ng
Da Da
o, 86-02
1-58
711
77
0
*Corres
p
o
ndi
n
g
author, e-ma
i
l
: nn
yan
@
shmt
u.edu.cn
A
b
st
r
a
ct
T
h
is p
ape
r establis
he
d
berth al
loc
a
ti
on a
nd q
uay
crane sch
ed
ulin
g
mo
dels
of shor
e
oper
ating for a contain
e
r ter
m
i
nal. F
u
rther
mor
e
, the st
ructure of the bert
h
- quay cran
e
sched
uli
ng ag
ent
is also giv
e
n
.
F
i
nally, an exa
m
pl
e usin
g gen
etic al
go
rithm to reso
l
v
e the berth
alloc
a
tio
n
mod
e
l
and a
g
e
n
t techno
logy is sh
ow
n. The experi
m
ent res
u
lt show
s that the use of
intellige
n
t theory an
d
techno
lo
gy can provid
e an
effective w
a
y for
contain
e
r te
rmi
nal o
per
atin
g sched
uli
ng.
Ke
y
w
ords
: ber
th alloc
a
tio
n
, quay cran
e sche
duli
ng, a
gent
Copy
right
©
2013 Un
ive
r
sita
s Ah
mad
Dah
l
an
. All rig
h
t
s r
ese
rved
.
1. Introduc
tion
The
contain
e
r
termi
nal
operating
sched
uling
inclu
d
e
s
be
rth allo
cation
a
nd qu
ay
cra
ne
sched
u
ling. Berth all
o
catio
n
mea
n
s
that b
e
rt
h
s
are
assign
ed
for a
ship
be
fore the
arrival
of the ship at
terminal
s. Q
uay
crane
scheduli
ng is d
e
fined a
s
the
allocatio
n
of a reaso
n
a
b
le
numbe
r of q
uay crane
s for ea
ch b
e
rt
h
and the
sui
T
able lo
ading
and unl
oadin
g
ord
e
r for
qu
ay
cra
n
e
s
. Due
to the rand
o
m
ness,
real
-time an
d
com
p
lexity of co
n
t
ainer l
oadi
ng
and
unlo
adi
ng
operation
s
, the buildi
ng o
f
a complete
analytical m
odel for
co
ntainer te
rm
inal ope
rati
ng
sched
uling i
s
very difficult
. With the
creation
a
nd
d
e
velopme
n
t
of many relat
ed theo
rie
s
a
n
d
optimizatio
n tech
niqu
es in
the field of
artifi
cial intel
ligen
ce, man
y
new metho
d
s in o
peratio
n
sched
uling
field a
r
e
presented. A
ne
w intelli
gent
sched
uling
method
ba
se
d on
multi-a
gent
system (MAS) provide
s
a new method
for the resolving of sch
edulin
g pro
b
l
e
ms.
In
recent
years,
some
schola
r
s
hav
e studie
d
the
appli
c
ation of
theorie
s an
d
technol
ogie
s
based o
n
ag
ent and MAS
in the field of port.
Reb
o
llo e
s
ta
blish
ed a
containe
r term
inal
prod
uctio
n
manag
eme
n
t system b
a
s
ed
on MA
S.In this system, variou
s resou
r
ces,
su
ch
as
q
uay
cra
nes, ship
pla
n
s,
are
ma
p
ped
to corresp
ondi
ng
a
gents.
T
he
purp
o
se of this
system is to obtain the shorte
st ship
stayin
g
time in po
rt [1]. Thurst
on pro
p
o
s
ed
a
distrib
u
ted ag
ent techn
o
log
y
to study the
coo
r
din
a
tion
of
contai
ner terminal han
dlin
g operation
s
,
whi
c
h mainl
y
resolves
the proble
m
of
t
he dispatchi
ng of quay cran
es and tru
c
ks.
In this stu
d
y, a sim
u
latio
n
syste
m
ba
sed
on
Java
wa
s devel
o
ped [2]. Co
n
s
ide
r
ing th
e
yard
plan
s and the tran
spo
r
tation plans,
Satoshi
Ho
shin
o esta
blish
ed an
automated
guided
vehicle
(AGV
) tra
n
spo
r
tation
system
based
on
a
gent to th
e
goal
of efficiency [3]. Pa
per
[4] propo
se
d
an optimiza
t
ion model b
a
se
d on (MAS) to re
solve the p
r
oble
m
of the
automatic
co
ntainer te
rmin
al prod
uctio
n
dispat
chin
g.
2. A D
y
namic Berth Ass
i
gnment Mo
del
Some
referred
parameters
in
the
dynamic
b
e
rt
h a
ssi
gnm
ent
mod
e
l
are:
the
arriving
time
for ship
s, th
e sta
r
t job
time of
ship
s at
be
rths,
the contain
e
r
loa
d
ing/u
n
l
oadin
g
time, the
depa
rture time of ship
s.Th
e followin
g
s
are a
s
sume
d: (1) the
di
fferent arriva
l times of shi
p
s
are
con
s
id
ered a
s
und
ete
r
mine
d varia
b
les, (2) to d
e
termin
e the
maximum a
c
cepT
able
wai
t
ing
time for
s
h
ips acc
o
rding to different ships
,
(3)
the
b
e
rths mu
st
meet the
phy
sical
con
d
itio
ns
of a ship (water de
pth a
nd length
)
, (4
) ea
ch
ship
has o
ne a
nd
only one b
e
rt
hing op
po
rtun
ity
and (5
) ea
ch
berth is o
b
tai
nable o
n
ly for one shi
p
.The
objective fun
c
tion is a
s
the
following:
Evaluation Warning : The document was created with Spire.PDF for Python.
TEL
K
654
ident
assi
g
whe
n
i. X
ijk
berth
sho
u
l
each
to ti
less
t
mea
n
L
j
m
e
assi
g
solv
e
3. Q
u
the
q
Rule
oper
a
K
OM
NIKA
V
The de
s
c
ifies the
nu
m
g
n
ed to
be
r
n
s
h
i
p
j
a
equals to 1
i.
The co
n
s
The ab
o
l
d be eq
ual
t
ship shoul
d
j
A
-
b
0
Expressi
o
me req
u
ire
m
t
han the
acc
e
D
W
j
i
(
Expressi
o
n
s the dept
h
I
P
x
j
i
)
(
Expressi
o
e
ans
the le
n
In this
m
g
n
ed fo
r
eac
h
e
the pro
b
le
m
u
a
y
Cran
e
Quay cr
a
QCS
=
(T
a
Task me
a
q
uay crane
m
ean
s
the
a
tions
.
V
ol. 11, No.
2
c
ripti
ons of
m
be
r set o
f
s
h
r
th i.b
j
mea
n
rriv
e
s. C
ij
if ship j is
a
s
t
r
ain
t
s are:
o
ve expres
s
t
o the total
a
d
be
served
j
wt
j
,
j
A
o
n
(4)
en
s
u
m
ents a
nd
e
pT
able ma
x
i
xijk
,
0
)
o
n (5) en
su
r
h
of berth i
i
x
ij
k
,
0
o
n
(6)
en
su
n
gth of shi
p
m
od
el
, the sh
h
be
r
t
h a
r
e
u
m
in this
pap
e
Schedulin
g
a
ne
sched
ul
a
sk
, Qc
, Rul
e
a
ns the
set
o
set
whi
c
h
r
u
l
e
s
e
t
w
2
, Februa
ry
vari
ables in
h
ips at po
rt.
n
s the time
means t
h
a
ssi
gne
d to
b
s
io
n (2) e
n
a
mount of al
at least on
e
V
u
re
s t
hat e
a
the impo
rt
a
x
imum wai
t
i
n
k
V
j
B
,
,
e
s
the dept
h
, D
j
means
t
k
V
j
B
,
,
re
s t
hat th
e
p
j
.
P
i
mea
n
s
ip'
s
a
r
riving
t
u
n
determin
e
d
e
r.
g
Model
ing
model
c
e
)
o
f all loadin
g
ca
n be s
c
hich sh
oul
d
2013 : 653
–
exp
r
ession
U stands fo
r
whe
n
ship
h
e pe
riod
o
b
erth i while
n
su
re
s t
he
l arriving
sh
e
time.
a
c
h
s
h
i
p
m
a
nce of
the
s
n
g time wt
j
.
U
k
h
of be
rth is
t
he shi
p
'
s
dr
a
U
e
len
g
t
h
of
s
the length
t
ime to
port,
d
dyn
a
mic f
a
c
a
n
be exp
r
g
and
unloa
d
c
he
dule
d
by
followed b
–
658
(1)
:
B me
a
r
the job nu
j
ac
cept
s
o
f loading/u
n
X
ijk
equals t
total
a
m
o
u
ips. The ab
u
s
t
be ser
v
s
hip. The
wa
i
not less th
a
a
ft depth.
ship
j
i
s
l
e
s
of berth
i
.
the be
rt
hin
g
a
ct
or
s.
S
o
w
r
esse
d as
a
d
ing t
a
s
ks o
all loadin
g
y quay cra
a
n
s
the ber
t
mber set
o
serv
i
c
e.
A
j
n
l
oadin
g
tim
e
o
0 if ship j
u
n
t
o
f
s
h
i
p
s
o
v
e expres
v
ed after
a
i
ting time of
e
a
n the ship'
s
s
s th
an t
he
g
location
an
w
e u
s
e th
e g
e
a
triple:
f be
rthin
g
s
g
and unlo
a
n
e
sched
uli
n
ISSN: 230
2
t
h nu
mbe
r
s
o
f ship j wh
means th
e
e
for ship j a
t
is n
o
t
assi
g
n
s
se
rved in
b
si
o
n
(3) e
n
rriving
acc
o
e
ac
h
s
h
ip
m
u
s
draft de
p
length
of
b
d the
quay
c
e
netic algo
ri
t
s
hip
s
. Qc
m
a
di
ng oper
a
n
g and
qu
ay
(
2
-4
046
(1)
s
et. V
ich is
e
time
t
be
rth
n
ed
to
(2)
(3)
b
erth
s
n
su
r
e
s
(4)
o
rding
u
st
be
(5)
p
th. W
i
(6)
b
erth
i
.
c
ra
ne
s
t
hm to
(7)
m
eans
a
tions
.
cr
a
n
e
(8)
)
Evaluation Warning : The document was created with Spire.PDF for Python.
TEL
K
A
unlo
a
of ta
s
bloc
k
indic
a
the p
quay
waiti
n
quay
need
(1
≤
q
≤
t
a
sk
loadi
n
((p
ri
i
=
the d
then
cabi
n
cra
n
e
4. T
h
st
ru
c
t
to t
h
agen
exch
a
berth
quay
sho
u
l
desi
g
of t
h
com
m
res
p
o
inter
p
K
OM
NIKA
A
gent-Ba
s
e
d
Task
i
=(
ti
d
Tid
i
is th
e
a
ding
ta
sk.
d
s
k
i
. bl
ock
i
is
t
k
of ta
sk
i
.
c
a
tes
the b
e
rio
r
ity
of
tas
k
qc
j
=(q
c
id
j
qcid
j
i
n
d
cran
e j. pos
j
n
g for q
uay
c
Suppo
se
c
r
an
es
.
F
o
s to be sche
Suppo
se
≤
n) is waitin
g
{qc
j
=(
qci
d
{qc
p
=(
qci
d
The ba
si
c
if ((bay
i
=
=
an
d t
he
u
n
g task
.*/
if ((bay
i
=
=
=
4) an
d
(p
eck sho
u
ld
b
if ((b
ay
i
=
((
pr
i
i
=1
) and
n
sh
ould
be f
if ((bay
i
<
b
e
s ca
n't
wor
k
h
e Structu
r
A
fter fini
s
t
ure of the B
Whe
n
a
h
e task
ag
e
t.
Therefo
r
a
nge with
th
-
q
uay cr
an
e
crane
ass
i
l
d
be
ab
le
g
ne
d the str
u
h
e BQSA a
m
uni
cati
on
i
o
ns
ib
le
fo
r
c
p
reter i
s
r
e
d
Automatic
S
d
i
, io
i
,deck
i
,
s
e
se
rial num
b
d
ec
k
i
mean
s
t
he block nu
c
n
i
indi
cat
e
s
e
ginnin
g
tim
k
i
.
, s
t
atus
j
, po
s
d
i
c
at
es the s
j
in
dic
a
t
e
s t
h
c
ra
ne j. s
j
i
n
the
r
e
are
r ta
s
k
i
=
(
t
i
d
dule
d
.
task
i
(1
≤
i
≤
n
)
g
for the s
e
r
d
j
, status
j
, p
o
d
p
, status
p
,
p
c
scheduli
ng
=
bay
q
) and (
i
u
nl
oadi
ng
ta
=
bay
q
) and
(
ri
q
=
3
)
)
/
*
i
f
b
e finishe
d
b
=
=ba
y
q
)
a
n
(pri
q
=2
)) /*
i
inished befo
b
ay
q
) and
(|
b
k
a
c
ross
eac
h
r
e of the
B
s
hin
g
buildi
n
QSA s
hould
ship arr
i
ve
e
nt by th
e
a
r
e,
the B
Q
e outsi
de
w
o
e
a
s
s
i
gnme
n
i
gnm
ent
an
d
to ma
na
ge
u
cture of th
re s
h
o
w
n
a
i
nte
r
face
a
n
c
ommunicati
n
e
sp
on
sible
f
I
S
S
hore Op
era
t
s
bay
i
, block
i
,
b
e
r
o
f
t
a
s
k
i
.
s
that task
i
i
s
mber in the
th
e
num
b
e
e of taski.
s
j
, n
j
, s
j
)
erial numbe
r
h
e current
b
n
dicate
s the
e
n tasks
w
a
it
i
d
i
, io
i
, dec
k
i
,
)
i
s
waiting f
o
r
vi
ce of qu
a
y
o
s
j
, t
j
) tas
k
i
p
os
p
, t
p
) tas
k
rul
e
s ca
n b
e
i
o
i
==1
)
and
(
sk are at t
h
e
(
(i
o
i
==0)
and
f
unloading
efore the
un
l
d
((io
i
==
1
)
f loading t
a
s
re the loa
d
i
n
b
ay
i
-bay
q
|>
=
h
anoth
e
r.
*/
B
erth
-Qu
a
y
n
g th
e bert
h
be de
sign
e
d
s
,
the task
a
ge
nt platf
o
r
Q
SA shoul
d
o
rl
d.In addit
i
o
n
t and
sch
e
d
sched
uli
n
g
the rule
s
o
e BQSA as
a
s the fol
l
o
w
n
d th
e
me
s
n
g with the
f
or the ex
p
S
SN: 2302-4
0
t
ing
Schedul
i
ybay
i
, c
n
i
, s
i
io
i
means
w
s
on
the
de
c
k
yar
d
of tas
k
e
r
of
co
nt
a
t
i
indicat
e
s
r
of qu
ay cr
a
b
ay of quay
e
arliest idle
t
i
ng
for
loa
d
,
bay
i
, c
n
i
, s
i
,
o
r the s
e
rv
i
c
（
y
cran
e p
1
≤
=(t
i
d
i
, io
i
, d
e
k
q
=(
tid
q
, io
q
,
d
e
expressed
(
io
q
==
0
)) the
e
same
bay
,
(i
o
q
==0)
)
a
n
t
a
sk
s ar
e
a
l
oa
ding ta
sk
)
a
n
d
(
i
o
s
ks
are at
n
g task on th
e
8)
a
nd (
p
os
j
Crane Sch
e
h
-quay
cran
e
d
.
r
e
que
s
t
o
f
r
m. The BQ
S
d
h
a
v
e
c
o
o
n, the BQS
A
e
duling. Al
s
g
sho
u
ld
b
e
o
f the loadi
n
sho
w
n in
F
w
in
g: (1)
c
o
s
sag
e
int
e
r
p
outsi
de wo
r
p
lanatio
n o
f
0
46
i
ng for a Co
n
, t
i
, pri
i
)
w
hether tas
k
k
o
r
i
n
t
h
e
c
a
k
i
. Ybay
i
indi
c
a
iners
waiti
the time n
a
ne
j. s
t
a
t
u
s
j
cra
ne j.
n
j
t
ime for qua
y
d
in
g/unlo
a
di
n
,
t
i
, pri
i
) (
1
≤
i
≤
c
e of qu
a
y
c
r
）
≤
p
≤
m.
e
ck
i
, bay
i
,
c
d
eck
q
, bay
q
,
as
:
n ((p
r
i
i
=1)
a
unloadi
ng
t
n
d ((de
ck
i
==
a
t the s
a
me
in the ca
bin
.
o
q
=
=
1
)
)
a
n
the same
b
e
deck.*/
j
<pos
p
) t
hen
e
du
ling Ag
e
e
assig
n
me
n
f
berthi
n
g
a
S
A receives
o
mmuni
cati
o
A
shoul
d
be
s
o, the
solu
t
e
st
ore
d
.
A
t
n
g/unlo
adin
g
F
igure 1. Th
e
o
mm
uni
c
ati
o
p
reter. Th
e
r
ld an
d the
f
the rec
e
i
v
n
tainer Term
i
k
i
i
s
t
h
e
l
o
a
a
bin. sbay
i
c
ates th
e b
a
ng fo
r l
o
a
eeded by fi
n
indi
cat
e
s t
h
e
i
s
the qu
a
y
cran
e j.
n
g
and
th
er
e
≤
n), pri
i
=
0,
w
r
ane j
（
1
≤
j
≤
m
c
n
i
, s
i
,
t
i
,
pri
i
cn
q
, s
q
, t
q
, p
r
a
nd (pri
q
=
2
t
ask shoul
d
b
1
)
a
n
d
(
d
bay, the
u
.
*/
d ((de
ck
i
==
1
b
ay, the lo
a
((
pr
i
i
=1)
an
d
e
nt (BQSA)
n
t and
sche
d
a
nd lo
adi
ng/
u
t
he task
req
o
n capabiliti
e
able to sto
r
e
t
i
o
n algo
rith
t
the
sa
me
g
ope
ration.
e
func
tions
o
f
o
n co
ntrol
u
communic
a
i
n
tegral pa
r
v
ed messa
g
ina
l
(Nanna
n
din
g
t
a
sk
o
i
s
the bay
n
u
a
y
numb
e
r
d
i
ng/unlo
adi
n
ishing task
i
.
e
cu
rr
ent
st
a
ntity of the
e
ar
e
m av
a
w
hi
ch m
ean
s
m
. Suppose
i
)}
(
r
i
q
}
(
2
)
)
/
*
i
f
l
o
b
e finish
ed
b
d
eck
q
==
0)
))
u
nloadin
g
ta
)a
nd (de
c
k
q
=
a
di
ng task
i
d
(pr
i
q
=1
))/
*
d
uling m
o
d
e
u
nlo
a
ding
i
s
ues
t
from th
e
e
s for infor
m
e
the
model
m for the
t
i
me, the
B
I
n
summa
r
f
the c
o
mp
o
u
nit includi
n
a
tion interf
a
r
t of the me
g
e
.
(2) re
s
n
Yan)
655
(9)
o
r the
u
mbe
r
in
th
e
ng
. S
i
pri
i
is
(10
)
(11)
a
tus
of
t
a
sks
a
ilable
s
task
i
tas
k
q
(
12)
13)
o
ading
b
efore
then
sk
on
=
=0
))
)
i
n the
Quay
e
l, the
s
sent
e
ta
s
k
m
ation
of the
be
rth
-
B
QSA
r
y, w
e
o
ne
nts
g th
e
a
ce i
s
s
s
a
ge
s
ou
rce
Evaluation Warning : The document was created with Spire.PDF for Python.
ISSN: 23
02-4
046
TELKOM
NIKA
Vol. 11, No. 2, Februa
ry 2013 : 653 – 658
656
manag
eme
n
t cente
r
. It is respon
sibl
e for acce
pting
the regi
strat
i
on of age
nts and
re
sou
r
ce
informatio
n query. (3) ta
sk buffer pool. Whe
n
seve
ra
l tasks arrive,
they can be put into the task
buffer pool waiting for bein
g
assi
gne
d. (4) job co
ntrol
module. It is arrang
ed to resolve the be
rth
allocation a
n
d
quay
crane
sched
uling
problem a
c
co
rd
ing to
the
sol
u
tions
sto
r
ed
in the m
o
d
e
l
libra
ry. (5) le
arnin
g
mod
u
l
e
. BQSA record
s and
su
mmari
ze
s th
e allocation
and sched
uling
results, and add
s new knowl
edge to
kno
w
le
dge d
a
taba
se. (6
) berth / quay crane d
a
taba
se. It
is a
r
range
d t
o
sto
r
e th
e inf
o
rmatio
n of b
e
rths an
d q
u
a
y crane
s
wh
ich
are m
a
n
aged
by B
Q
SA.
(7) kno
w
ledg
e/quay
cran
e sch
edulin
g
rule
d
a
taba
se. It i
s
re
spo
n
si
ble fo
r
stori
ng
ea
ch
allocation
an
d
sched
uling
result.
(8) mo
del
d
a
taba
se.
Different
solutio
n
s fo
r
berth
and
qu
ay-
cra
ne allo
cati
on and
sched
uling problem
s are
stored i
n
model data
base.
Figure 1. The Structu
r
e of
the
Berth - Q
uay Cra
ne Scheduli
ng Age
n
t
5. Solution Exa
m
ples
We
ta
ke
a
contai
ner
te
rminal
which
ha
s
3
b
e
rt
hs
a
nd
is
p
r
ovided with 1
2
qu
ay
cra
n
e
s
for
an
example.
The lo
adin
g
/u
nloadi
ng
sp
e
ed
of
the
q
uay cra
ne i
s
2
7
co
ntai
ners
per h
our. Th
e wo
rk d
a
ta of 12 ship
s i
s
sho
w
n a
s
T
able 1. The
maximum accepT
able
wa
iting
time for ships is from 3-10 h
o
u
r
s . The loadi
ng/
unloa
ding ta
sks of ea
ch
ship is sho
w
n
as
Table 2.
Table 1. The
Basic
Wo
rk
Data of Arriving Ships
Arr
i
ving
Shi
p
s
/
Vesse
ls
Arr
i
vin
g
Time
Th
e
Numbe
r
of
Conta
i
ne
rs
Da
y
Momen
t
Sh
ip
1
Th
e
1st
D
a
y
01:
4
0
96
0
Sh
ip
2
Th
e
1st
D
a
y
05:
2
0
17
90
Sh
ip
3
Th
e
1st
D
a
y
08:
3
0
98
0
Sh
ip
4
Th
e
1st
D
a
y
13:
0
0
18
25
Sh
ip
5
Th
e
1st
D
a
y
22:
0
0
10
60
Sh
ip
6
Th
e
1st
D
a
y
23:
0
0
98
0
Sh
ip
7
Th
e
2n
d
D
a
y
05:
0
0
15
25
Sh
ip
8
Th
e
2n
d
D
a
y
06:
0
0
10
35
Sh
ip
9
Th
e
2n
d
D
a
y
20:
0
0
10
80
Sh
ip
10
Th
e
2n
d
D
a
y
21:
0
0
89
0
Sh
ip
11
Th
e
3r
d
D
a
y
18:
0
0
16
70
Sh
ip
12
Th
e
3r
d
D
a
y
20:
0
0
15
80
The tested b
a
si
c paramet
ers of
gen
etic algorithm are
that: (1
) pop
ulation si
ze is 100. (2)
the times of
here
d
ity iterat
ion is 1
000.
(3) ge
netic
crossove
r p
r
obability is
0.9. (4)
gen
e
t
i
c
variation probability is 0.05.
Evaluation Warning : The document was created with Spire.PDF for Python.
TELKOM
NIKA
ISSN:
2302-4
046
Agent-Ba
s
ed
Autom
a
ticShore Op
erating
Scheduli
ng for a Co
ntaine
r Term
inal (Nanna
n Yan)
657
Table 2. The
Examples of
Loadi
ng/Unl
o
ading Ta
sks
Task
No.
Task
Bay
No.
Deck/Cab
i
n
Load
ing
/
Unload
ing
Th
e SequeceNumbe
r
in
One Ba
y
Th
e N
u
mbe
r
of
Conta
i
ne
rs
I
01D
01
D
I
1
1
I
01H
01
H
I
2
1
I
02H
02
H
I
2
31
E02H
02
H
E
4
12
I
03D
03
D
I
1
2
I
05D
05
D
I
1
1
...
...
...
...
...
...
I
30D
30
D
I
1
7
I
30H
30
H
I
2
1
E30H
30
H
E
4
15
E30D
30
D
E
4
17
I
34D
34
D
I
1
14
I
34H
34
H
I
2
12
Suppo
se all
berth
s satisfy the arriving
ship
s.
Acco
rd
ing to the a
s
sumed
exp
e
rime
nt
para
m
eters,
the be
st val
ue i
s
39.2
2
after
ru
nnin
g
the
gen
etic a
l
gorithm prog
ram.
Th
e
be
st
chromo
som
e
is 1
4 8
12
0
2 6
9 1
3
0
3
5 7
10. So
ship 1,
ship
4,
shi
p
8 a
n
d
ship
12
are
assign
ed for berth No. 1 whi
c
h a
c
qui
re
s 4 quay cran
es. Ship 2, sh
ip 6, ship 9 a
nd shi
p
13 are
assign
ed for
berth
No. 2
whi
c
h a
c
q
u
ires 4
quay
cra
nes. Ship
3,
Ship 5, ship
7 and
shi
p
10
are a
s
sign
e
d
for be
rth
No. 3
whi
c
h al
so
acq
u
ire
s
4 q
u
a
y
cran
es. T
he di
spat
chi
n
g
arrang
ement
is sh
own as
Table 3.
Table 3.Th
e Example of the Best Beth Allocatio
n
Usin
g Geneti
c
Algoritm
Arr
i
ving
Ships
/
Vessels
Th
e
Ber
t
hi
ng
Berth
Arr
i
ving
Tim
e
Tim
e
(
hours)
Depa
rtu
r
e
Tim
e
Load
ing
/
Unloading
Speed
Volum
e
s/ Hour
)
Day
Momen
t
Day
Momen
t
Sh
ip
1
1
Th
e 1
st
Day
01:
4
0
9. 33
Th
e 1
st
Day
10:
5
0
10
5
Sh
ip
2
2
Th
e 1
st
Day
05:
2
0
13.
6
6
Th
e 1
st
Day
19:
0
0
13
1
Sh
ip
3
3
Th
e 1
st
Day
08:
3
0
9. 25
Th
e 1
st
Day
17:
4
5
10
6
Sh
ip
4
1
Th
e 1
st
Day
13:
0
0
13.
3
3
Th
e 2
nd
Day
2:1
0
13
0
Sh
ip
5
3
Th
e 1
st
Day
22:
0
0
9. 83
Th
e 2
nd
Day
7:5
0
10
8
Sh
ip
6
2
Th
e 1
st
Day
23:
0
0
9. 5
Th
e 2
nd
Day
8:3
0
10
3
Sh
ip
7
3
Th
e 2
nd
Day
05:
0
0
11.
3
3
Th
e 2
nd
Day
19:
1
0
13
5
Sh
ip
8
1
Th
e 2
nd
Day
06:
0
0
9. 66
Th
e 2
nd
Day
15:
4
0
10
7
Sh
ip
9
2
Th
e 2
nd
Day
20:
0
0
8. 5
Th
e 3
rd
Da
y
4:3
0
12
7
Sh
ip
10
3
Th
e 2
nd
Day
21:
0
0
1. 25
Th
e 3
rd
Da
y
7:1
5
87
Sh
ip
11
1
Th
e 3
rd
Da
y
18:
0
0
12.
5
Th
e 4
th
Day
6:3
0
13
4
Sh
ip
12
2
Th
e 3
rd
Da
y
20:
0
0
13.
8
3
Th
e 4
th
Day
9:5
0
11
4
5. Conclusio
n
This pa
per u
s
e
s
a
gent
te
chn
o
logy to
build
th
e
structure
of the
Be
rth-Q
uay
cra
ne
Sched
uling
Agent(BQSA). The berth
alloca
tion mo
del and th
e q
uay crane
scheduli
ng rule
s of
the BQSA are also sh
o
w
n in this p
aper. Exper
i
m
ent re
sult sho
w
s t
hat the use of agent
techn
o
logy a
nd othe
r rel
a
ted intelligent
theory ca
n
provide a
n
effective sup
port for sh
ore
operating sch
edulin
g for a contai
ner te
rminal.
Referen
ces
[1] Rebo
llo, J
u
lia
n,
Carrascosa, Botti.
A Multi-Ag
ent Sys
t
em for th
e
Auto
matio
n
of
a Port C
onta
i
ne
r
T
e
rmi
nal.
Auto
nomo
u
s Age
n
ts 2000
w
o
rksh
op on Ag
ents i
n
Industr
y
.
Bar
c
elo
na, Spa
i
n. 200
0.
Evaluation Warning : The document was created with Spire.PDF for Python.
ISSN: 23
02-4
046
TELKOM
NIKA
Vol. 11, No. 2, Februa
ry 2013 : 653 – 658
658
[2] T
hurston HU.
Distrib
uted
Agent Arc
h
itect
u
re for P
o
rt Au
tomati
on
. 26th Internatio
na
l
C
o
mputer
S
o
ft
w
a
r
e
and Ap
plic
atio
ns Confer
enc
e. Oxford. UK. 2002; 81-
90.
[3] Satoshi H
o
shin
o, Jun Ota, Akiko Sh
in
ozaki, Hi
dek
i Hashim
oto.
D
e
sig
n
of an A
G
V T
r
ansporta
tio
n
System
by C
o
nsid
erin
g Ma
n
age
ment Mo
de
l
in
an
ACT
Intelli
ge
nt Auton
o
m
o
u
s Syste
m
s
. Book Ed
itors
IOS Press. 2006.
[4] Miguel
Re
boll
o
, Vice
nte
Julia
n,
Carl
o
s
Carrascos
a
, Vicente Botti.
A Multi-Agent System
for the
Auto
matio
n
of a P
o
rt Co
nta
i
ner T
e
r
m
in
al.
26th
Intern
ati
ona
l C
o
mput
e
r
Soft
w
a
re
an
d Ap
plic
atio
ns
Confer
ence. Oxford. UK. 20
0
2
; 81-90.
Evaluation Warning : The document was created with Spire.PDF for Python.