Upload
others
View
12
Download
0
Embed Size (px)
Citation preview
亚太示范电子口岸网络建设汇报
Dr. Xu Kai (Kernel)
8-21-2015
Shipping Big Data Applications Solutions
Research
Contents
New Requirements
How To Deal With
Some Initial Results
Not Only For Shipping
BACKGROUND
Today, international trade and logistics is undergoing profound changes. On the one hand,
e-commerce is changing the traditional business model, a large number of transaction
information has been generated through the network; on the other hand, with the
IOT(Internet of things), sensors distributed in all aspects of logistics and transportation are
also keep generating large amounts of data. Therefore, government departments and
enterprises have access to more data than in the past, people urgently need new methods
to maximize the value of these data.
Shanghai International Shipping Institute, the place where I work, provides consulting and
information services for the government and enterprises. We know the value and
significance of the data in the strategic decision of the government and enterprises, so we
have done a lot of basic work in these areas.
Part 1
China Shipping Database
More than 400 statistical indicators;
More than 2.3 million statistics;
Five sections:
Port &Terminal, Shipping Market, Shipping Service, Economic Indicator, Prosperity Index
February 2012:
Start construction
September 2013:
Go live
CSDB is very good. But the data it provides are from the traditional way.
When we search for "Shanghai", the 62
kinds of statistical indicators associated
with Shanghai will focus displayed on a
page, such as : Cargo throughput of China’s
coastal ports, Oil natuaral gas and their
products Traffic, etc.
From the traditional point of view, the CSDB has reached a advanced level.
When we search for "iron ore", the 8
kinds of Statistical indicators about it
will be displayed in one webpage. such
as : Iron Ore Production Traffic, China
Iron Ore Shipping Volume for Import,
etc.
Acquisition Integration Storage Processing Analysis
Past Investigation/
Manual
Paper/
Digital info Isolate
Low speed/
Low capacity/ Localization
Processing batch
Statistics/
Causal
analysis/ Sampling analysis
Now Internet of
things/ Automatic
Enterprise Service bus(ESB)
Real time/
Distributed/
Cloud
computing
Parallel Computing
Correlation
analysis/ Full sample analysis
new problems
Information Explosion
Standardization Data Center Super computing
Data Mining
What is changing in Big Data Era?
Big data research is going to solve these new problems.
The
Internet of
Things
Computer
Interaction
Big Data
How Big Data is generated?
Some data collection techniques we have.
OBD汽车数据总线
a11
a22
3a3
4a4
b1
b2
b3
b4
5
6
7
8
Vcc1
0
GND
0
MCU控制器SPI接口
尾气传感器
Vin
GND
Vref
B1
B8
Sign
ENB
A/D模数转换器
北斗定位模块/通信模块
I2C总线
串口通信
北斗网络
Human Animal
Cargo Container
Ship
Car Truck Machine
Wearable Device
Smart bracelet Logistics black box Terminal system for ship Car network suite
Geographic
Data
Electronic chart
See floor map
Shoreline berth
resources
Monitoring
Data
AIS
Beidou
satellite
positioning
Shipping lane
RFID
identification
VDR
Statistics
Transport
volume
Transport
capacity
Port throughput
Shipping route
congestion
Dock operation
efficiency
Freight rate
Business
Data
Ship visa
Ship registration
data
Rescue request
B/L data
Customs
clearance data
Shipping order
Textual
Data
Shipping
Regulations
Maritime case
Shipping
contract
Media Data
Port surveillance
video
Ship photos
What shipping data do we have today?
"4V" Features of Big Data
•Freight has strong
timeliness, it changes
every day
•20 million AIS data can
be compressed to 7
million
•Format heterogeneity,
including video,
images, other
unstructured data
•More than 20 million
AIS data being
produced every day
throughout the world
Volume Variety
Velocity Value
Part 2
Data is the specific form of information that can be used to save, copy and
spread.
So the "dead" data is useless, only to analyze the valuable part of the
information can be called "big data“; otherwise, it’s just "a pile of data".
And the purpose of data analysis is to help company make the right decision
in the business.
Data analysis serves business decision
Big Data Driven Model
Demand Transaction
DataService
Demands orientation Data driven
Quote from the book "Big Data Driven Shipping Informatization“
Shannon(C.E.Sannon), the founder of information theory proposed the
concept of "information entropy“ in 1948 to measure the number of
“information“ a data contains.
A large amount of information is has been lost from the first hand data
to the data we see.
Data and information
Data
entropy
multidimensional structure
Change an angle
Focus on detail and structure
The conclusion has changed.
We can not see the whole picture without big data
226cm
157cm
192cm(AVG)
Distributed and Parallel Computing
Information
System
DataBase
Information
Platform
Big Data
Develop
Integrate
Cloud
Computing
Online Data
Realize
Use
Stage of Shipping Informatization
Evolution of Data Utilization
Quote from the book "Big Data Driven Shipping Informatization“
Data utilization is developed with information technology
Shipping
Big Data
Storage
Processing
Visualization
Analysis
What Shipping Big Data Research?
Storage
A distributed database can store hundreds of billions of mobile target
monitoring data.
There are 350 billion AIS data in the database.
Processing
Key algorithms for data cleaning and compression.
Analysis
Multiple analytical models and key algorithms.
Concurrent computation environment based on Hadoop technology.
Experimental environment of data analysis based on R and Python.
Visualization
Tracks and Scene playback technology.
Various graphics report technology.
Part 3
Typical domestic route analysis results
Linyun River412275000 Qingyun River412148000
Tiansheng River412658000
"Shipping & Port Big Data Laboratory" Research Results
By analyzing AIS data, we can know the real voyage condition of
each liner routes and calculate actual voyage and fuel consumption
of each route. So shipping company can account cost accurately
and tell owner the punctuality rate of every route.
Global shipping route fitting results
"Shipping & Port Big Data Laboratory" Research Results
We can calculate the number of ships and the total capacity of each route, which provides
data support for market analysis and helps to predict the trend of market price.
Statistics of port production index
We have calculated the number of arrived ship and time of berthing to have an
objective understanding of port loading/unloading efficiency and berth utilization
rate, which provides data support for optimization of port management.
Data Sources AIS Analytical Data Statistical Data
Statistics range Solid Cargo Ship Liquid Cargo Ship Total Total
Statistics Berthing
Time
Berthing
Frequency
Berthing
Time
Berthing
Frequency
Berthing
Time
Berthing
Frequency Annual Throughput
Unit Hour Times Hour Times Hour Times Thousand Tons
Busan Port 28203 800 36600.47 420 64803.47 1220 346100
Tianjin Port 54416.9 1356 25788.3 454 80205.2 1810 540000
Singapore Port 37472.6 1012 29562.3 524 67034.9 1536 581270
"Shipping & Port Big Data Laboratory" Research Results 2014-09
Regional emissions and energy consumption analysis
We can measure the concentration spatial distribution of pollutant emissions and
carbon emissions precisely , and the results are being applied to the research of
establishing gas emission control area in Yangtze River Delta and Pearl River Delta.
Singapore 1 Day CO2 Emissions Shanghai 1 Day CO2 Emissions
"Shipping & Port Big Data Laboratory" Research Results
Ships Behavior Recognition Model
Quay
BerthBerth BerthBerth BerthBerth BerthBerth
AnchorageAnchorageAnchorageAnchorage
Port Area
Event A
Event B
Event C
P1
Pt
Pk
P2
PN
Pm
Pn
"Shipping & Port Big Data Laboratory" Research Results
Previous studies are based on the identification of ship behavior.
Big Data technologies help us find all moored events
Within 15 days, on a global scale, all the ship mooring points.
"Shipping & Port Big Data Laboratory" Research Results
According to location we distinguish between anchoring and berthing Events
"Shipping & Port Big Data Laboratory" Research Results
An example from the Xiamen Port
Ship track data will be converted ship into a sailing log.
Part 4
For Multimodal Transport
Generic Target Monitoring System, GTMS
Unified data format: Multi-target Packet, MTP
Our research is scalable
Shipping
Data
Cargo
Data
Truck
Data
Common Database Structure
Common Processing Platform
In the course of our research,
we reserve the room for
future expansion.
Future data analysis system
SQL
Not only for shipping industry, it will face to global trade and logistics.
Thank You!
Aug, 2015
Any comments, suggestions or feedbacks
will be greatly appreciated!
Dr. Xu Kai
E-mail:[email protected]
Phone: +86 13816950360