Aplikcie a prpadov tdie rozhodovanie a predikcia

1
Aplikácie a prípadové túdie (rozhodovanie a
predikcia)

• 4. Prednáka

2
Selected Case Studies

• Made in our Lab
• Organize with Our Lab
• Known in the literature

3
The basic Principle of modern Intelligent
technology approach
Historical Data Analysis ( Statistical Neural
Approach )

Expert Knowledge

Synergy building Incremental Knowledge base
Continuous process
Continuous process
Multi-Agent Systems approach
4
Load Prediction using InT

• Load forecast http//neuron.tuke.sk/competition
• Company behind East-Slovakia Power
  Distribution Company
• Responsible EUNITE node
• Center for Intelligent Technology
• TU Kosice, Slovakia
• http//www.ai-cit.sk

5
Defined Problem
Data From 2 years Electricity Load every 30
minutes Add. Data tempreture
Predictor
31 values (max. load for january next year)
6
Loads of the Years 1997 and 1998
Half an hour load for some day
7
Temperature of the Years 1997 and 1998
8
January 1999 - Original Set Expected to Be
Predicted
9
The Best and the Worst Results
10
Results - MAPE and MAX Errors
11
56 Registered Competitors From 21 Countries
12
Conclusion Research Related

• Results prove the ability of these technologies
  to make prediction with reasonable accuracy.
• The results could be improved if some closer
  discussion with company would be during the
  competition but could not be done during
  competition.
• Various technologies including SVM, Neural ....
• Submitted reports could be research value and
  should be published in some EUNITE publication.

13
Metalurgy
14
Intelligent Control of Furnace with electric arc

• Basic principle
• Modeling the environment and adapt the
  controller

20-25 saving of electrodes 5-8 saving of
energy
Productivity
10 - 12
15
Optimization of Roll Profiles in the Hot Rolling
of Steel using InT
The best possible control to get precise
parameters for the output
16
What to improve ???
17
Results are better then conventional control
OUTPUT
INPUT
"Intelligent Roll Profile Optimisation System"
18
Mechanical Engineering
19
Fuzzy Logic Diagnostics Anticipated Machine
Malfunctions (Sneck)

• Vibration problem
• Many applications
• IF X (temperature) is high, AND Y (speed) is low,
  THEN Z (stop the machine and/or activate an
  alarm). Diagnosis a bearing is in trouble.

20
Example Continously Adapting Gear Shift Schedule
in VW New Beetle
21
EBERLE Fuzzy Thermostat

• My personal experience
• On my house heating system once reaching
  target temperature keeping it very good
  control .
• Reaching target tempetarure Depends on
  ekvithermal heating
• (E1-E9)

22
Various application ....
23
Home appliances
Company BPL Product washing machine ABS
50F Fuzzy system decides the type of Program
amount of water and washing ingredients
Company BPL Product washing machine ABS
60 NF Neuro-fuzzy system detects a type of
material in the machine and decides the type of
the program and amunt of water and washing
ingredients.
24
Home Appliances
Company Videocon-international Product
washing machine V-NA- 45 FDX The same as
before just 996 different cycle to choose from
. Which on is decided By neuro-fuzzy system
Company Videocon-internetional Product
Washing machine Fuzzy control of the machine
25
Home appliance
Company Sanyo Product washing
machine ASW-F60T The same concept made by
company
Company LG Product Refrigerator Neural
fuzzy system controls the freezing procedures in
the refrigirator
26
Home appliance
Company Sanyo Cook , owen cooker
ECJ-5205SN According to the senszors of infra,
thermal senzor a huminity senzor it estimate a
meal quality and determine A time of cooking.
27
Electronics
Company Sharp Product microwave owen Accoding
to the analysis of the inside air the lenght of
the cooking is controlled. The analysis of the
Food smell during cooking is matter of interest.
Company Videocon Product air-conditioner Neuro
-fuzzy control of air-conditioner to keep equal
temperature within the room
28
Electronics
Company Cannon Product videocamera Canon uses
fuzzy system with 13 rules to focus the
objectives based on the information in the image
characteristics
Company Mitsubishi Product TV set Make a
neural controller to adjust the image contrast
according to the broadcast image. This adaptive
approach produce a very good User feeling while
seeing TV program.
29
Electronics
Company Samsung Product Blod pressure
measurement Fuzzy system controls the overall
process of Blood measurement
Company Samsung Product Camera Fuzzy control
of image focusing sharpening
30
Electronics
Company JVC porduct car-radio Using neural
networks it is able to control car radio with
high reliability and adapt to the voice of the
speaker.
Company IntelaVoice Product switcher controled
by voice Using neural networks it is able to
control the switch with high reliability and
adapt to the voice of the speaker.
31
Copy machines
Company Canon Product Copy Machine Series of
CLC700 a CLC800 have a fuzzy control of the toner
to achieve the best results
Company Panasonic Product Copy machine In the
series FP-1680 up to FP-4080 is implemented a
neuro-fuzzy system to control various parameters
to get the best copy results as possible
32
Car industry
Companies Mercedes Hyundai Mercedes in
model CLK use Automatics transmission based on
Highly adaptive technology to adapt to the style
of the driver. Similar approach is in XG Hyundai
model.
Company BMW BMW uses long time a fuzzy
approach in ABS brake system which adapts the
braking process with the aim to avoid blocking
phase. Also in other advance systems these
technologies are used.
33
Car Industry

• Company Siemens AG
• Product Smart Airbag
• Smart airbags for persons safety uses some
  parts of intelligent technologies including
  adapting
• safety measures to the people.

34
Application in Mining
35
Reservoir Characterization
Fractured reservoir modeling

36
Intelligent Technologiesin terms of data mining
and data fusion
I Conventional interpretation
II Conventional integration
III
III Intelligent characterization
Data Fusion
II
I
Data Mining
37
Intelligent Reservoir Characterization
Data Segmentation Real wells gtPseudo wells gt
Synthetic
38
Intelligent Reservoir Assessment
Data Segmentation
centres
match
segments
39
Neural network configuration
Rotliegend case study
Input from Reflectivity cube
From porosity trace of pseudo-wells in training
set, from porosity trace of real wells in test
set.
Input from AI cube
40
Intelligent Reservoir Characterization

• Visualization of 3D bodies with similar seismic
  character
• Neural network learned to segment 12 seismic
  attributes into 10 segments
• No physical significance of different clusters yet

41
Model Probability Information gain

• Stochastic inversion
• real wells
• simulated wells
• Scoring
• seismic character
• geostatistical probability
• Output
• expectation grids
• standard deviation grids

42
Satellite Image data classification
MF ARTMAP classification
Original image. Highlighted areas were
classified by an expert (A urban area, B
barren fields, C bushes, D agricultural
fields, E meadows, F forests, G water)
43
Results
(Classification by Parallel MF-ARTMAP)
44
Engineering application

• Intelligent Technologies only with
• Collaboration with experts from engineering
• Field collaboration is necessary!!!

45
Can you call your product intelligent ?? Why not
???Will it have impact on demand and sales
???????Can we measure Intelligence of Machine
or Software ????Machine InT Q.
46
Application in Financial Domains
47
False Bank Transactions Identification
O.K
InT Decision Ability to Learn Incrementally
Bank Transaction
COULD BE FALSE
48
Transaction Features
49
step1 - clustering
Symbolic feature Filter association
How many subspaces Subspaces ??????
Input for empirical knowledge
50
step2 ident. of unknown input
Entropy of Points ?????
Yes
Unknown Input
Is the input belongs to the identified cluster
? (determined in step 1)
NO
No false
51
Financial fraud transaction classification on
test data
52
Feedback analysis for an Expert
53
BASEL II EU condition
54
BASEL II Upgrade Banks Risk Management

• High AAA to A-
• Medium BBB to B-
• Low under B-

Intelligent Rating Management System (IRMS)
Advanced Approach
Client Data
55
Workflow of Credit Approval
Rating Assessment (Intelligent RMS)
Bank Management Board
Client
Credit Manager
Monitoring Client (i.e. on quarterly basis)
56
Credit Manager team approach to Intelligent RMS
WEB oriented interface to I-RMS server
57
Why is Intelligent RMS progressive
Historical Data Analysis
Intelligent Rating Management
Expert Knowledge

Synergy building Incremental Knowledge base
Continuous process
Continuous process
Credit Rating Assessment
58
ATM Machine Cash Management
59
ATM Machine Cash Prediction
Historical Data Analysis ( Statistical Neural
Approach )
Intelligent iCMS
Expert Knowledge

Synergy building Incremental Knowledge base
Continuous process
Continuous process
ATM Client behaviour
60
Adaptive Workflow
example
1.
70 000
Intelligent CMS (iCMS)
ATM Manager
2.
data
85 000
expert
3.
45 000
Feedback
Selection of a group ATM (e.g. 10 machines and
replenishment time T 8 days
10.
135000
Analysis of leftover
Gate-X
61
WHY DO IT ???

• With a good forecast the socket amount of an ATM
  machine can be reduced.
• A preliminary estimate shows a possible
  reduction of up to 100.000 DM for the provided
  data set of an ATM machine in Düren/NRW.

Background
Benefit
Concept
Conclusion
Ecanse
62
Many Other applications

• These applications are not public many
  Companies does not declare InT application It
  is commercial secret.
• Trippi Neural Networks in Finance and Investing
  (ISBN 1-55738-919-5)

63
How far is the level of Automation and Machine
Intelligence ???
QUO VADIS Information technology and
Automation???????
64
ASIMO the ROBOT
22.8.2003
arrived with the Japanese delegation and,
although not human, is receiving more attention
than Premier Junichiro Koizumi
the smallest member of the delegation a mere
120 centimeters tall is named ASIMO and is a
third-generation humanoid robot made by Honda