Embed Size (px)
Citation preview
Copyright © 2007 ontoprise GmbH, Karlsruhe
www.ontoprise.de
How Ontologies and Rules Help to Advance Automobile Development
RuleML 2007Juergen Angele, ontoprise GmbH
www.ontoprise.de
© 2007 ontoprise GmbH- 2 -
know how to use Know-how!
Founded: 1999 Team: approx. 50 EmployeesHeadquarter: Karlsruhe
Market: 9 out of the 20 largest German companies are our customers
Strategic Partner for Oracle and Software AG
Ontoprise is a leading semantic software company. Our goal is to make a company‘s know-how visible
and reusable
www.ontoprise.de
© 2007 ontoprise GmbH- 3 -
Karlsruhe: Location for Semantic Technologies
Application-orientedResearch
Know-how TransferRealizing new Scenarios
Application-orientedResearch
Product DevelopmentInnovative Solutions
Basic ResearchApplication-
orientedResearch
AIFB
www.ontoprise.de
© 2007 ontoprise GmbH- 4 -
„If I knew that we have already
implemented this component from
this vendor somewhere else, I had
calculated differently.“
„With this error message, I always
exchange the complete PC. Is there
another way?“
„While I am still busy collecting all the
information from all systems most of
the customers already ran out of
patience“
Support users in quickly finding information
Single View on Multiple Sources
Reuse the know-how of your engineers
SemanticMiner®
SemanticGuide®
SemanticIntegrator®
Field of Applications for ontoprise Technology
www.ontoprise.de
© 2007 ontoprise GmbH- 5 -
Selected References
www.ontoprise.de
© 2007 ontoprise GmbH- 6 -
Semantics = All About Meaning
social semantics (meaning)
On
tolo
gy
Railroad Object
Train
Steam Train
Track
Mogul
Mogul Ski Race
Mogul Emperor
Mogul Train
„An Ontology is ashared conceptualization ofa domain [Tom Gruber].“
www.ontoprise.de
© 2007 ontoprise GmbH- 7 -
Parts of an Ontology
Mogul Mallet
gage width:1435 mm
gage width:1067 mm
narrow-gage standard gage
suitable for
suitable for
suitable for
Semantics (Meaning)
Relations
Rules
On
tolo
gy
„An Ontology is a formal and defined System of Concepts and Relations between these Conceptsused to describe complex domains of knowledge.“
IF gage width = 1435 THEN suitable for standard gage ELSE narrow gage
Railroad Object
Train
Steam Train
Track
www.ontoprise.de
© 2007 ontoprise GmbH- 9 -
Semantic Web Layer
OntoBroker OWL
OntoBroker
Tim Berners-Lee, ISWC November 2005, http://www.w3.org/2005/Talks/1110-iswc-tbl/#(12)
www.ontoprise.de
© 2007 ontoprise GmbH- 11 -
Challenges
Majority of innovations is in electronical equipment
Increasing complexity in development and integration
shorter development cycles increasing quality measures
Automatic Analysis of Control Units for Audi
Solution Extraction of rules from requirement specification and functional frames Collection of expert knowledge from engineers in free text Carve out of logic by means of an ontology Automatic validation of test results by means of specifications
Goals Introduction of efficient testing
methods to reduce manual work Manage complexity Increase transparency
www.ontoprise.de
© 2007 ontoprise GmbH- 12 -
• Audi Valvelift System (AVS)
• 2 different Cam Contours for small and large Valve Lift
• Increases Engine Efficiency (more Power, lower Fuel Consumption)
• Controlled by Engine ManagementSystem• Deterministic Finite Automaton• S1, S4 - small, large Valve Lift• Transition Functions
Use Case
www.ontoprise.de
© 2007 ontoprise GmbH- 13 -
• Observable Variables during HiL Tests
• Snapshots at different Times
• Ontology Reflects Data Structure recorded during HiL Tests
• Introduces Terms as known to Experts (e.g., engineSpeed)
Use Case
www.ontoprise.de
© 2007 ontoprise GmbH- 14 -
Collection of rules
Documentation Requirement Specification, Descriptions, Functional Framework(Word, Excel, PDF)
Expert knowledge
Reverse Engineering Interviews
Collection of rules(natural language)
www.ontoprise.de
© 2007 ontoprise GmbH- 15 -
Ontology- and rule development
rules
rule 1
rule 2
rule 3
rule n
Ontology
rules(NL)
advantages:
• Set of rules is extendable in an incremental way
• Rules are not hidden in program code.
• Rules are automatically explainable.
www.ontoprise.de
© 2007 ontoprise GmbH- 16 -
Collecting rules
www.ontoprise.de
© 2007 ontoprise GmbH- 17 -
Ontology- and rule development
ECU Specification: „If the engine speed is greaterthan 4000, the valve lift system must switch to S4 if itis in S1.”
?S[nextState->?S4] <-?S:Situation[state->?S1, engineSpeed->?V] and?V > 4000.
Experts: „At idle speed the small valve lift must beused.“
ERROR(?S) <-?S:Situation[state->?S4, idle->1].
www.ontoprise.de
© 2007 ontoprise GmbH- 18 -
Testing measurements
measurement
OntoBroker
Ontology and rules
Analysis result
www.ontoprise.de
© 2007 ontoprise GmbH- 19 -
Analyzing data
www.ontoprise.de
© 2007 ontoprise GmbH- 20 -
Found errors
www.ontoprise.de
© 2007 ontoprise GmbH- 21 -
Explaining inferences
In situation s2 still state 4 holds. In the situation before state 4 hold as well. The rotary speed
was larger than the threshold. Therefore a state transition should
have happened.
www.ontoprise.de
© 2007 ontoprise GmbH- 22 -
Ontology-based Testcar Configurator at Audi
Functionality
Goal
„There is no other technology to both describe this level of complexity and being flexible enough to adapt to changes. ontoprise‘s technology enables us to describe and make executable our complex domain in a flexible and maintainable manner.“
Support of internal order processing for building and rebuilding testcars (AVx)
Integrate cross department dependencies into AVx
The knowledge about functional, gemoterical and processual dependencies is spread over many engineers
Reduce time for testcar lifecycle and therefore for the whole development cycle
Utilize expertise of engineers to improve testcar process
prevent time-lags in testcar process
Sample Ontology (Source ontoprise)
I
I
I
I
I
340 kW
Part_ID
340 kW
has_Power
designed_for_powerPart_ID
has_Part
has_Part
Has_Part
has_Part
CAR
Engine, Motor
Chassis,Under-carriage
Electronics
Body FW 4x4-587
M V8-340
SE 32-566
controls
Part_ID
has_Power(Engine) < designed_for_power(Chassis)
Otherwise error
www.ontoprise.de
© 2007 ontoprise GmbH- 23 -
Ontology combines rules, structures and information
Ontology
Mapping of existing information
StructuresDependencies, rules
www.ontoprise.de
© 2007 ontoprise GmbH- 24 -
Engineer
namephone
document
Object
Person Part Approval
Employee External Filter
Developer
needsneedsresponsibleresponsible
KAT
DPF
Rules: Parts must be approved before you can test them embedded
Only the Person responsible for a part can approve its testing
Rules: Parts must be approved before you can test them embedded
Only the Person responsible for a part can approve its testing
part_ofpart_of
synonymsynonym
part_ofpart_of
Is_a Is_a
Is_a
Is_a
Testing
subtopicsubtopic
Ontologies represent the meaning of information
www.ontoprise.de
© 2007 ontoprise GmbH- 25 -
Relationships/Constraints
Example Rule: The maximum power of the motor must not exceed the one of the brakes: Pmotor < Pbrakes
Menu
www.ontoprise.de
© 2007 ontoprise GmbH- 26 -
“Since researchers still disagree about the best way of combining (OWL) ontologies and rules, we have selected the representation language that comes with the best professional support.
F–Logic turned out to be most suitable for our demands.” [Audi]
www.ontoprise.de
© 2007 ontoprise GmbH- 27 -
What our customer likes on Flogic
F-Logic declarative (logic-based) clear semantics (well-founded semantics)
powerful (rules, functions, negations) ontology based structuring (frame) schema reasoning simple human readable syntax homogenious rule and query syntax logical model of a domain
modelling environment close integration into databases IT infrastructure handling large amounts of data fast engine
www.ontoprise.de
© 2007 ontoprise GmbH- 28 -
OntoStudio / NEON toolkit
www.ontoprise.de
© 2007 ontoprise GmbH- 29 -
OntoStudio / NEON toolkit
www.ontoprise.de
© 2007 ontoprise GmbH- 30 -
What our customer dislikes on Flogic
F-Logic
it is not a standard !
www.ontoprise.de
© 2007 ontoprise GmbH- 31 -
Conclusion
Ontologies and Rules …
increase the transparency by carving out logics from applications and data because all results are explained in natural language
make complexity manageable because informal and distributed knowledge is formalized and therefore made
machine processable because knowledge can be structured and re-used
help to build flexible systems that can adapt to changes quickly
but we need a standard!
