Study Report

rumblecleverΤεχνίτη Νοημοσύνη και Ρομποτική

1 Δεκ 2013 (πριν από 3 χρόνια και 11 μήνες)

323 εμφανίσεις

2010
-
8
-
27





Study Report

Summary of my lab life in POSTECH


Submitted by:

Hong Li

Ho
me University:

Logistics

Engineering,
School

of

Mechanical

Engineering
,

University of Science

& Technology Beijing, China

Exchange Lab:
Logistics lab, Industrial & Management Engineering,

POSTECH, Korea

Exchange Date:
07/16/2010 to 09/01/2010

Advisor:
Prof. Byung
-
In Kim, Industrial & Management Engineering,

POSTECH, Korea

2010

Hong Li

USTB

2010
-
8
-
30

Study
Report

2010


-

1
-



C
ONTENTS
:


1.
Traffic Simulation, TRANSYT
·············
·
·············
··········
··

2

2.
Data C
ollection of Marine Trade between Korea and China

···
·········
·

4

2.
1

Korean Ports

····
··································
··········

13

2.2 Chinese Ports

················································

14

2.3 Distance
····················································

17

2.4
Variety and quality indicators of fuel

····························

17

2.5
Major Vessel
s

in
China
·································
·
······

18

2.6
Achieved Port
Volume

·········································

19

3.
Balanced Clustering
: Affinity Propagation & K
-
means

··
·········
··
···

20

4.
Additional
Gains

······
···················
·
·········
·········
····

22

References

·············································
·
·
···
·····

2
3
















Study
Report

2010


-

2
-



LIST OF FIGURES


Figure 1.1
Original

Pattern
··········································

4

Figure 1.2
TRANSYT Simulation

··
··································

5

Figure 2.1 Marine trade routes of the world

······
·····················

10

Figure 2.2 The first route from Weihai to Inchon

·······················

10

Figure 2.3 Location of Yan Tai and Inchon

·
···························

11

Figure 2.4 The routes of the fo
ur tunnels

·························
····

12

Figure 2.5 Ports in Korea

···················
·······················

13

Figure 2.6
Ports in China

········
··································

14

F
igure 3.1

A&R Relationship Between Data Points
··
···················

20


















Study
Report

2010


-

3
-



LIST OF TABLES


Table

2
.1
Korea's Foreign Trade

Statistics

······························

7

Table
2
.2
Korea
n

e
xports to major trading

partner
s

············
·
·········

8

Table
2
.3
Korea
n

im
ports to major trading
partners

····················
·

9

Table
2
.4
List of basic data for Low
-
Carbon

·····
······················

12

Table 2.
5

Ports in Korea

············································

13

Table 2.
6

Ports in China

································
·
···········

15

Table 2.
7

Distance between k
orean

ports and chinese ports

····
·
··········

16

Table 2.
8

Fuel oil quality indicators

of China Fuel Company

·············

1
7

Table 2.9

Ships in use

·····························
·
················

1
8

Table 2.10

Volume of Shanghai Port

(
Korea & China
)
···············
·
···

19

















Study
Report

2010


-

4
-


P
ART 1.

Traffic Simulation
, TRANSYT


My lab life
starts

with

a

traffic simulation by us
ing

an off
-
line computer program
called TRANSYT.

TRANSYT
is
software

us
ed

to simulate the traffic conditions. It
could be used to
determin
e

and
study optim
al

fixed time, co
ordinated
, traffic signal timings in any
network of roads for which the average traffic flows are known. A traffic model of the
network calculates a
Performance Index (PI) in monetary terms, which is a weighted
sum of all vehicle delay
s

and stops. An
optimizing

routine systematically alters signal
offsets and/or allocation of green times to search for the timings which reduce the P.I.
to a minimum valu
e.

After reading the user guide of TRANSYT 13

[1]

and
doing
some test
s
,

I

d
learn
t that
TRANSYT is based on GUI so we could input and simulate our pattern easily and
vividly.

We choose a
series

of
intersections of Pohang
downtown

as our optimization

s

original pattern.
Figure 1
.1

shows the detailed traffic conditions of our original
pattern.

Figure 1
.1

Original

Pattern

[2]

Study
Report

2010


-

5
-


We

want to use TRANSYT to make a simulation with 5
nodes
,

one bottleneck

5x

and
type is give way, d
rive on
the

right.

Figure 2 is
the simulated traffic

conditions

of our original
pattern.

Figure
1.2

TRANSYT Simulation













TRANSYT could do something to optimize:

a.

Simulate the traffic conditions then integrate the whole graph

s
parameters

and
find if there

re

some collision and some man
-
made mistakes;

b.

If there

re

some overloaded jams, it will show you the up/down road flow gain in
percentage in that we could find the data flooding link and correct it;

c.

It will sum the flow apparent loss in percentage for analys
is;

d.

S
how the traffic condition of each stage clearly, also the stage for every node;

e.

I
t will provide you a generate
d

report including a summary of the simulation;

Additionally, I also
read another user guide is about The TRANSYT
-
VISSIM
L
INK

[
3
]

an optional
component

of TRANSYT 13.

VISSIM is

a micro
-
simulation
. If

TRANSYT passed the signal settings to VISSIM,
then VISSIM organize
s

a model for it so that could
optimize

the signals. When using
Study
Report

2010


-

6
-


TRANSYT
-
VISSIM
L
INK screens in TRANSYT 13 could
normally be used to view
and links edit, but flows and connections are disabled.

When the TRANSYT
-
VISSIM LINK works, it

ll
calculate

a performance index (PI)
which could be the average net delay or a particular journey time and so on. The
TRANSYT
receive
s

all of
the candidate times, and analyze
s the differences between

them, find the
effect

to PI then back to VISSIM
. S
o
VISSIM runs again to make
improvement
. Until the signal timings
are

optimized
. Then export

it

for further study
and evaluation of the net.

The
correspond
ences of VISSIM to TRANSYT:

VISSIM signal controller TRANSYT node.

VISSIM signal group TRANSYT link.

L
ink timings, stage definitions and timings are automatically
calculate
d

from the
VISSIM green times.

S
ome other data for TRANSYT should be set by ourselves in order to
optimize
.
Because

they

do not exist

in VISSIM and the default parameters are always
unrealistic
.

There

re still a variety of data in VISSIM and we c
an see the
m but can

t
manipulated.




PART 2.

Data collection of Marine Trade between Korea and China


My advisor Prof. Kim

is the
K
orean

research leader

of a cooperative
research

project.
So I learn to do some
research

about this project a
ssessment of
inter
-
regional division
and cooperation

among China, Korea and Japan in formulating a low
-
carbon marine
trade and logistics system.

Now, more than 2 / 3 of the total transport volume of international trade achieved
through the ocean, while
C
hinese total import and export about 90% of all use the sea
transport.
Since 2004 China has overtaken the United States and Japan to become
Study
Report

2010


-

7
-


South Korea

s largest trading partner.

This year
at the end of A
ugust, the total amount
is 98,071
,400,000 U.S. dolla
rs
the
trade
of
China to Korea

[4]
.

Tables
2
.1, 2.2
and
2
.
3

show
s

detailed information about the trade of Korea and Korea to China:


Table
2
.1

Korea's Foreign Trade

Statistics

[
5
]

Korea's Foreign Trade Statistics

(2001
-
2010.6)

million U.S. dollars

Date

Total

Increased
%

Export

Increased
%

Import

Increased
%

2001

291,537

-
12.4

150,439

-
12.7

141,098

-
12.1

2002

314,597

7.9

162,471

8

152,126

7.8

2003

372,644

18.5

193,817

19.3

178,827

17.6

2004

478,307

28.4

253,845

31

224,463

25.5

2005

545,657

14.1

284,419

12

261,238

16.4

200
6

634,847

16.3

325,465

14.4

309,383

18.4

2007

727,780

14.6

371,327

14.1

356,453

15.2

2008

857,282

17.7

422,007

13.6

435,275

22

2009

686,618

-
19.9

363,534

-
13.9

323,085

-
25.8

2010
,
1
-
6

425,387

37.1

221,514

34.4

203,873

40.2

In


J a n

6 2,2 1 5

3 5.2

3 0,7 4 1

4 5.5

3 1,4 7 4

2 6.4


F e b

6 4,0 5 5

3 3.5

3 3,0 4 0

3 0.1

3 1,0 1 5

3 7.2


M a r

7 2,9 4 8

4 0.8

3 7,4 2 7

3 4.2

3 5,5 2 0

4 8.5



A p r

7 4,8 2 1

3 5.5

3 9,3 5 9

2 9.8

3 5,4 6 2

4 2.6


M a y

7 3,9 3 5

4 4.3

3 9,0 2 6

4 0.3

3 4,9 0 9

4 9.1


J u n

7 7,4 1 3

3 3.7

4 1,9 2 1

3 0.1

3 5,4 9 3

3 8.2








Study
Report

2010


-

8
-



Table
2
.
2

Korea
n

exports to major trading partners

[
6
]

Export of Korea

(Jan
-
Jun,

2010)


Unit: Million dollars

Nation(Region)

Amount

Increase
d
%

Percentage
%

Total

221,514

34.4

100.0

China

55,648

48.7

25.1

USA

23,512

32.1

10.6

Japan

12,834

31.6

5.8

Hong Kong

12,369

44.0

5.6

Singapore

7,559

24.5

3.4

Chinese Tai Wan

6,651

73.4

3.0

Germany

5,646

48.6

2.6

India

5,181

42.1

2.3

Mexico

4,339

33.1

2.0

Indonesia

4,287

70.4

1.9

Vietnam

4,266

37.2

1.9

Brazil

3,685

108.0

1.7

Australia

3,141

41.9

1.4

Thailand

3,102

57.0

1.4

Malaysian

3,062

69.2

1.4

(Increased means
compared with

last year.)







Study
Report

2010


-

9
-



Table
2
.
3

Korea
n

im
ports to major trading partners

[
7
]

Import of
Korea

(Jan
-
Jun
,

2010)

U n i t: Mi l l i o n d o l l a r s

N a t i o n ( R e g i o n )

A m o u n t

C o m p a r e d %

P e r c e n t a g e %

T o t a l

2 0 3,8 7 3

4 0.2

1 0 0.0

C h i n a

3 3,6 8 9

3 4.7

1 6.5

J a p a n

3 0,9 0 6

3 9.3

1 5.2

U S A

2 0,1 9 7

5 8.3

9.9

S a u d i A r a b i a

1 2,5 3 7

5 2.2

6.2

A u s t r a l i a

8,7 6 7

2 5.2

4.3

I n d o n e s i a

7,0 0 3

7 9.4

3.4

G e r m a n y

6,6 4 0

2 3.4

3.3

C h i n e s e T a i Wa n

6,3 4 2

4 9.4

3.1

U n i t e d A r a b E m i r a t e s

6,2 3 6

6 6.6

3.1

Q a t a r

5,7 3 8

3 1.5

2.8

K u w a i t

5,3 6 8

6 0.2

2.6

R u s s i a

4,9 5 6

1 1 8.4

2.4

Ma l a y s i a n

4,9 0 9

2 7.0

2.4

I r a n

4,0 2 5

6 4.4

2.0

S i n g a p o r e

3,7 3 3

-
3.2

1.8



C h i n a
,
Ko
rea
,

Japan
,
Russia

and
Australia

always share

one
marine trade shipping
route, especially among China, Korea and Japan.

Figure
2.1

show
s

marine trade
routes of the world:



Study
Report

2010


-

10
-



Figure
2
.1

Marine trade routes of the world

[4]


China
and

Korea launched their first s
hipping line officially
in

1990
.

I
t’s from Weihai
I

to Inchon

[4]
.

Figure
2
.
2

The first route from Weihai to Inchon


Study
Report

2010


-

11
-



In

2006, China and Korea signed an agreement to combine the land and sea transport,
the first truck got on the ship at Inchon port.
It means the seamless logistics between
Korea and China started

[4]
.

Between China and Ko
rea the route from Yantai I

to Inchon
(K) is
called the star route
for mar
ine trade of the two countries. I
t

s very important
. (
Blue one is Yan Tai)

Figure
2
.
3

Location of Yan Tai and Inchon



According to the news

[8]
,
China and
Korea are preparing to build a
c
ross
-
h
arbor
t
unnel
.

I
t means
that
the logistics between the two countries will be more
efficien
t
.
Here are the three plans of the tunnels from Korea to China:


1

Inchon to Weihai

341miles




2

Hwaseong to Weihai

373miles




3

Pyeongtaek

to Weihai

386miles




[4:
Ongjin

(North Korea) to Weihai (221miles)
]


The figure 2
.4

shows the routes of these plans.



Study
Report

2010


-

12
-


Figure
2
.
4

The routes of the four tunnels


Korea Internati
onal Trade Association estimate
d

at 2009 that South Korea and China

s
annual trade volume

in 2013 may exceed 200 billion U.S. dollars which is 3 years
later than the two nations planned

[9]
.


In my collection, I just focused on the emission of carbon
-
dioxide.
If we know the
distance, the ship
characters
, the fuel and the amount of trade, then we could
calculator

the carbon emissions.
So

it needs some basic data:

Table
2
.
4

List of basic data for Low
-
Carbon

Data

Status

Korean

Ports



Chinese

Ports




Distance



Fuel



Ship



Port
Volumes

Some of



Study
Report

2010


-

13
-


2.
1

Korean Ports

Figure 2.
5

shows the marine ports in Korea.

Figure 2.
5

Ports in Korea



Here are 16 major ports of Korea list in table 2.
5
.

Table 2.
5

Ports in Korea

Korea
n

Ports

1

Seoul

9

Masan

2

P
usan

10

Mokpo

3

Changhang

11

Mukho

4

Cheju

12

Pohang

5

Chinha
e

13

Gwangyang

6

Daepori

14

Samch
e
ok

7

Inchon

15

Ulsan

8

Kunsan

16

Yosu


Study
Report

2010


-

14
-


2.2 Chinese Ports

Figure 2.1 shows the marine ports in Korea.

Figure 2.
6

Ports in China















Study
Report

2010


-

15
-



Following table shows

ports in
China
, there are

57
in
list.

Table 2.
6

Ports in China

Chinese Ports

Number

Chinese

English

Number

Chinese

English

1

八所港

BASUO PORT

30

泉州港

QUANZHOU PORT

2

北海港

BEIHAI PORT

31

日照港

RIZHAO PORT

3

重庆港

CHONGQING PORT

32

三亚港

SANYA PORT

4

大连港

DALIAN PORT

33

上海港

SHANGHAI PORT

5

丹东港

DANDONG PORT

34

汕头港

SHANTOU PORT

6

防城港

FANGCHENG PORT

35

深圳港

SHENZHEN PORT

7

福州港

FUZHOU PORT

36

石岛港

SHIDAO PORT

8

广州港

GUANGZHOU
PORT

37

台北

TAIBEI

9

海口港

HAIKOU PORT

38

太仓港

TAICANG PORT

10

香港

HONGKONG

39

台中

TAICHUNG

11

花莲

HUA LIEN

40

台州港

TAIZHOU PORT

12

江阴港

JIANGYIN PORT

41

天津港

TIANJIN PORT

13

京唐港

JINGTANG PORT

42

威海港

WEIHAI PORT

14

锦州港

JINZHOU PORT

43

温州港

WENZHOU PORT

15

九江港

JIUJIANG PORT

44

武汉港

WUHAN PORT

16

高雄

KAOHSIUNG

45

芜湖港

WUHU PORT

17

基隆

KEELUNG

46

厦门港

XIAMEN PORT

18

莱州港

LAIZHOU PORT

47

洋浦港

YANG PU PORT

19

连云港港

LIANYUNGANG
PORT

48

阳江港

YANGJIANG PORT

20

龙口港

LONGKOU PORT

49

烟台港

YANTAI PORT

21

龙眼港

LONGYAN PORT

50

营口港

YINGKOU PORT

22

澳门

MACAO

51

漳州港

ZANGZHOU PORT

23

马尾港

MAWEI PORT

52

张家港港

ZHANGJIAGANG
PORT

24

南京港

NANJING PORT

53

湛江港

ZHANJIANG PORT

25

南通港

NANTONG PORT

54

镇江港

ZHENJIANG PORT

26

宁波港

NINGBO PORT

55

中山港

ZHONGSHAN
PORT

27

青岛港

QINGDAO PORT

56

舟山港

ZHOUSHAN PORT

28

秦皇岛港

QINHUANGDAO
PORT

57

珠海港

ZHUHAI PORT

29

钦州港

QINZHOU PORT


Study
Report

2010


-

16
-



2.3 Distance

Distance

of Korea Ports and Chinese Ports

Table 2.
7

Distance between
K
orean

ports and
C
hinese ports

[10]

DISTANCE

PORT A

PORT B

Nautical miles

大连港
DALIAN

釜山
PUSAN

546

大连港
DALIAN

仁川
INCHON

285

大连港
DALIAN

平泽
Pyeongtaek

280

青岛港
QINGDAO

釜山
PUSAN

502

青岛港
QINGDAO

仁川
INCHON

328

青岛港
QINGDAO

平泽
Pyeongtaek

329

上海港
SHANGHAI

釜山
PUSAN

330

上海港
SHANGHAI

仁川
INCHON

331

上海港
SHANGHAI

平泽
Pyeongtaek

332

连云港
LIANYUNGANG

釜山
PUSAN

333

连云港
LIANYUNGANG

仁川
INCHON

334

连云港
LIANYUNGANG

木浦
Mokpo

335

石岛港
SHIDAO

釜山
PUSAN

336

石岛港
SHIDAO

仁川
INCHON

337

石岛港
SHIDAO

平泽
Pyeongtaek

338

天津港
TIANJIN

釜山
PUSAN

772

烟台港
YANTAI

仁川
INCHON

271


2.4

Variety and quality indicators of fuel

Marine fuel oil is made from the heavy oil and light distillated oil. It has a large
amount of heat, good combustion performance, storage
stability

with
small corrosion
s
,
so it
has an excellent and wide range of usage. It is
great
horsepower

with
medium
and low speed marine diesel engine
s
most economic and ideal fuel, but also can be
used as boiler fuel
for
small nozzle.

The main specificati
ons are:

RMD15
-
-

120

Marine fuel
; quite
Ray'
s
1st viscosity (100 ° F) 1000 seconds, for
high horsepower, the low
-
speed marine diesel engine
s
.

R
ME2
5
-
-

1
8
0

Marine fuel
; quite
Ray'
s
1st viscosity

(100 ° F) 1
5
00 seconds, for
Study
Report

2010


-

17
-


high horsepower marine diesel
engine
s
.

RM
G35
-
-

380

Marine fuel
; quite
Ray'
s
1st viscosity (100 ° F)
3
000 seconds, for
high horsepower marine diesel engine
s
.

(Seconds means the time needed flow out 50ml from the viscosity
measuring tools

Table 2.
8

Fuel oil quality indicators

of China Fuel Company

[11]






Name


Heavy diesel

Light diesel

Test Methods


Standard

RMD



120

RME



180

RMG



380

20


0














Number

15

25

35

(≈DMC,









DMO)










Kinematic
viscosity

100

,

mm2/s≤

15



25



35



20.5



(50

)

3.0
-
8.0



(20

)

GB/T

265

GB/T

11137

ISO

3104

Kinematic
viscosity

50

,

mm2/s≤



120



180



380





Ray's
viscosity

(37.8

)≤

800

950

2000

160
0

3800

360
0

135

30
-
40



C
arbon
residues

wt%,≤

14

15

18

0.5

10%

0.4

GB/T

268

ISO

6615


W
ater

wt%,

0.8

1

1

1

痕迹

GB/T 260

ISO 3733

D
ensity


kg/m3,≤

985

991

991

实测

实测

GB/T 1884

ISO 3765

ISO 821(6.2)


Sulfur

wt%,≤





3.5

3.5

3.5

0.5

(GB/T

387)



1

(GB/T

380)

GB/T

11140

ISO

8754









Study
Report

2010


-

18
-


2.5
Major Vessel
s

in
China


㈵⁤楦ieren琠獨tps⁡ e⁩ c汵ded



Table 2.
9

Ships in use

[11]

Type Name

Number of Containers

TEU


Main engine fuel
consumption(
ton
/d
ay
)

102 TEU normal
container ship

102

2.9

118 TEU normal container ship

118

-

120TEU offshore multi
-
purpose
vessel

120(66/54)

3.76

128 TEU normal container ship

128(80/48)

5.1

152 TEU normal container ship

152(84/68)

-

171TEU refrigerated container ship

171

5.8

200 TEU

normal container ship

210

9.4

230TEU full container ship

230

7.85

319 TEU normal container ship

319

16.4

345 TEU normal container ship

345

11.1

358TEU multi
-
purpose container
vessel

358

16.72

405 TEU normal container ship

405

15.3

412 TEU normal
container ship

412

16.4

500 TEU normal container ship

504

-

500t refrigerated transport vessel

500

-

502 TEU container / multipurpose
ship

505

15.7

504 TEU normal container ship

-

-

610 TEU normal container ship

614

19.17

650 TEU normal container
ship

779

23.81

818 TEU normal container ship

865

22.85

900 TEU normal container ship

-

-

1200 TEU normal container ship

1270

52.8

1236 TEU no hatch container ship

1236

94

1700 TEU normal container ship

1714

62.5

15500t container ship

1436

-


Study
Report

2010


-

19
-


2.6
Achieved Port
Volume


Table 2.
10

Volume of Shanghai Port

Korea & China

[1
2
]

Num
ber

Route

Route
Code

Ship
Capacity
(TEU)

Frequency

Arrive

Depart

Main Stops

1

Pusan

SKZ3

350

Per week

Fri.

Fri.

Shanghai

Pyeongtaek

Inchon

2

Pusan

SK02

650

Per week

Thur.

Fri.

Shanghai/Pusan/Ulsan/G
wangyang/Ulsan/Pusan/S
hang

3

Pusan

NK1

480

Per week

Wed.

Thur.

Shanghai/Pusan

4

Pusan

CJS

500

Per week

Wed.

Thur.

Ningbo/Shanghai/Pusan/
Komaki

/
Muroran /
Sakata

/Pusan/Ulsan/Ningbo

5

Inchon

ISS

1000

Per week

Sat.

Sun.

Shanghai/
Inchon
/Shanghai

6

Pusan

SK01

700

Per week

Tue.

Wed.

Shanghai/ Pusan /Komaki
Sendai /onahama /Naka

Pusan / Ulsan
/Gwangyang / Shanghai

7

Inchon

SKZC

253

Per week

Tue.

Wed.

Shanghai / Inchon

8

Pusan

STC

1200

Per week

Fri.

Sat.

Shanghai /Kunsan/ Pusan
/
Ulsan /Gwangyang

9

Pusan



933

Per week

Mon.

Tue.

Shanghai / Pusan /
太仓
/
Shanghai

10

Pusan

PHS

500

Per week

Tue.

Wed.

Shanghai / Inchon /
Pyeongtaek /Kunsan/
Shanghai

11

Inchon



184

Per week

Fri.

Fri.

Shanghai / Inchon

12

Pusan



480

Per week

Sat.

Sun.

Shanghai / Pusan / Ulsan
/Gwangyang

13

Pusan





Per week

Mon.

Tue.

Shanghai / Pusan

14

Pusan



431

Per week

Tue.

Wed.

Shanghai /Pusan/ Ulsan
/Gwangyang

15

Inchon



210

Per week

Mon.

Tue.

Shanghai / Inchon


Study
Report

2010


-

20
-


PART 3.

Balanced Clustering
: Affinity
Propagation & K
-
means


I think the purpose of cluster
ing is to analyze

the data by finding their features and
create

an accurate
definition

or pattern for th
e clusters. It m
ake
s

some improvement
s

at the same time.
Clustering data based on
the

measure
ment

of

similarity is a critical
step in scientific

data analysis and in engineering systems
. I

d learnt two different
algorithms

the other

is AP

and another

is K
-
means

[1
3
]

[14]
.

There

are

a lot of m
ethods for algorithm to analyze

clusters

such as
p
artitioning

m
ethods,
h
ierarchical
m
ethods,
d
ensity
-
b
ased
m
ethods
, grid
-
b
ased
m
ethods

and
model
-
b
ased
m
ethods.

AP is a new clustering algorith
m. Here are the key points of this algorithm:



Exemplar
,

the center of cluster
.



Similarity,
similarity

of two

data points is
s(
i,
k
)
, indicates
how well the data
point

with index k is suited

to be the exemplar for data point i.

For points
X
i
and
X
k
,

s(i,

k) =

||
X
i



X
k
||
2
.



Preference
,

indicates the preference of i as a cluster center

P (i) or S (i, i)
.



Responsibility, R (i, k)

indicates how well data point k be the cluster center of
data point i.



Availability, A (i, k) indicates hoe well data point
i

choose data point k as it

s
cluster center.

Responsibility and Availability are sent between data points
as their message,
figure 3
.1

shows their relationship between data points.


Figure 3
.1
A&R Relationship
between

Data Points




Study
Report

2010


-

21
-


Functions to calculate R
and

A

R(
i
, k)=
S(i, k)

-

max{A(i, j)+S(i, j)}

(j

{1,2,

,N,

j≠k})

A(i,

k)=min{0,R(k,k)+

j
k
j
R
))}}
,
(
,
0
{max(

(j

{1,2,…,

N,

j≠i,

j≠k})

R(k,k)=P(k)
-
max{A(k,j)+S(k,j)} (j

{1,2,…
,N,

j≠
k})



Damping factor, it

s function is

convergence. For every iteration A and R should
be update
d

form the last A and R.


i
R


(1
-
lam) *
i
R
+lam *
1

i
R

lam
)
1
,
5
.
0
[



i
A

=

(1
-
lam) *
i
A
+lam *
1

i
A


lam
)
1
,
5
.
0
[



K
-
means

should input the first series cluster centers as start points. Then it will make
some improvement by iteration. The clusters will be ma
de by the rules: data
points
have

high similarity in same cluster and low similarity in different cluster. Every
cluster has their own centriod shows the cluster

s center of gravity.

K
-
means has a high dependence

on the fi
rst series cluster centers which were given by
ourselves

at the beginning. So the
consequences

of this algorithm

is very sensitive.
However, compared with the AP algorithm K
-
means could get a output quickly, so in
some situation K
-
means also has a great
applications.










Study
Report

2010


-

22
-


PART 4.


Additional
Gain
s


Every
T
uesday,
T
hursday and
W
ednesday we

had

an
English

meeting. Ph.D.
candidates of our lab show

us their recent
researches
.
I

had
learnt
Cellular

automata
simulation,
Optimization for
nonlinear

programming, Subside system for bus, Fuzzy
logistics
, VRP problem
with time window and School bus routing problem.

I had listened The 2
nd

International Symposium on IT Convergence Engineering at
POSTECH
. I feel that nowadays a lot of logistics
technologies are

used in humans


daily life. For instance, they try to use RFI
D into
a hospital computer system. Every
patient

will have

a bracelet which RFID information are included. In this way, doctors
could know patients very well and very fast, nurse could distribute medicines by
check
ing

the patients


RFID information. It

will

increase efficiency

of the

hospital

and
significantly reduce

mistakes which could be a vital influence on our life.

The life in logistics lab is very precious
to

me
!

It b
rings me a lot of improvements,
simulation, iteration, optimization

and also
Kor
ean

and
Korean

culture!

I have read
the One to One papers together with my professor and learnt something about the God
and Jesus Christ. I think no matter
one

believe it or not the stories in the Bible will
give
you

more courage and a more

wide sight which could lead to
a brilliant

mind.

My advisor Prof. Kim

s gentle and smile,
Junhyuk

s

cool
,
Jeongin

s easygoing,
Jooyoung

s careful,
Yeowoon’
s enthusiastic,
Jongsung’
s mature

and
Hyunchul’
s vigor

are my great gifts! Thank all of you
for

tak
ing

care of me
!









Study
Report

2010


-

23
-


PART 5.


References


[1]
James C B
.
, Grah
am B
.

and Mark C
.,

(2009).
TRANSYT 13
user guide
.

[2] NHN

corp
, Map of
N
aver, 2008 from
http://map.naver.com

[
3
]
Graham B
., (2008).
TRANSYT
-
VISSIM LINK
user guide
.

[
4
] China International marine Net, 2003
-
2010 from
http://www.shippingchina.com

[
5
]
Department of
Commerce
, People

s Republic of China, 2010 from

http://countryreport.mofcom.gov.cn/record/view.asp?news_id=2042
5

[
6
]

Department of
Commerce
, People

s Republic of China, 2010 from

http://countryreport.mofcom.gov.cn/record/view.asp?news_id=2042
6

[
7
]

Department of
Commerce
, People

s Republic of China, 2010 from

http://countryreport.mofcom.gov.cn/record/view.asp?news_id=2042
7

[8] P
eople Net, Dec. 11th, 2009
, Guangzhou

Daily

from
http://ccnews.people.com.cn/GB/10559494.html

http://ccnews.people.com.cn/GB/12469907.html

[9]
China news,
Sep. 21
st

, 2009 from
http://www.chinanews.com.cn/cj/cj
-
gjcj/news/2009/09
-
21/1877095.shtml

[
10
] China International marine Net, 2003
-
2010 from
http://port.shippingchina.com/portdata/index/index.html

[11]

China International marine Net, 2003
-
2010 from

http://info.shippingchina.com/hmcs/index/index/oneday//type/hmcs/page/6.html

[
1
2
]
S
hanghai

I
nternational

P
ort

(G
roup
)
, 2010 fr
om
http://www.portshanghai.com.cn

[
1
3
]

Brendan J. F
.
,
& Delbert D.,

(2007).
Clustering by Passing Messages
between

Data

Points
,
Science
,

315
,

972
-
976

[1
4
]
Brendan J. F
.
,
(2009).
A Binary Variable Model for Affinity Propagation
,
Neural
Computation
,

21, 1589

1600