Upload
btu-cottbus
View
2
Download
0
Embed Size (px)
Citation preview
Preprints of the 2013 IFAC Conference on ManufacturingModelling, Management, and Control, Saint PetersburgState University and Saint Petersburg National ResearchUniversity of Information Technologies, Mechanics, andOptics, Saint Petersburg, Russia, June 19-21, 2013
ThB3.6
478
Preprints of the 2013 IFAC Conference on Manufacturing Modelling, Management, and Control, Saint Petersburg State University and SaintPetersburg National Research University of Information Technologies, Mechanics, and Optics, Saint Petersburg, Russia, June 19-21, 2013
479
knowledge as well as the resources specifications used to
process the product at a node in consideration. The Query
Generator takes each node one by one and generates relevant
query. The query engine takes each query generated against
each node and infers potential environmental impact of the
node in consideration. In case the sought information is
found in the knowledgebase, the query engine delivers
information to Evaluation and Decision Making Module. If
no result is found, that would infer that the case is completely
new and therefore the node has to be sent to the LCA
simulation tool for environmental impact assessment. The
model synthesizer takes information concerning the selected
node and generates appropriate query. This query scans the
knowledgebase and generates the possible potential
environmental impact as well as the relevant indicators and
nodal input to environmental indictor output model. This
model is actually input-output simulation model which can be
discovered or inferred from the application specific
knowledgebase. The planner can visualize the model and can
reject or accept the model. Upon approval, the input-output
model information is sent to the simulation tool for
simulation of potential environment impact. The calculated
results are delivered to both the Evaluation and Decision
Making Module and the Knowledgebase. The system works
in connection with the existing legacy software tools for
planning through web-service over the web. This allows
planners to share as well as store information over the web
about simulated cases and make the inference capabilities of
the tool powerful in assisting the choice of optimal
production schemes based on environmental impact.
The implicit knowledge of the planners is collected through
formal means by providing a web-based Delphi survey
conducted among the experts of the particular domain. Delphi
method has been used which is a well establish method and
acknowledged in the field of information science (Okoli and
Pawlowski 2004, Cegielski 2001, Hayne and Pollard
2000,Holsapple and Joshi 2002, Lai and Chung 2002,
Mulligan 2002, Nambisan et al. 1999, Schmidt et al. 2001).
The advantages of Delphi surveys are that the experts can
describe and supplement their cases over several rounds and
can post comments to the cases of the other experts
anonymously. This ensures that each case is viewed and
reviewed several times. Lastly, there is an opportunity to
evaluate the cases and make statements regarding their
quality. It is also possible to conduct the Delphi survey via
web-based application as a location independent solution.
The survey is supported by a monitoring team. The team is
responsible for identifying experts and for organizing the
survey. During the survey, they are supporting the experts,
summarizing interim results and may obtain information for
the next round. In addition, they carry out preparatory and
follow-up steps of Delphi survey. A generic process of a
Delphi survey has been described in (Okoli and Pawlowski
2004, Weimer and Seuring 2008, Skulmoski et al. 2007); the
survey for this research activity will be structured in the same
manner. During the Delphi survey, the illustrated process will
be conducted in each round. Initially, the questionnaire is
developed. After each round, the monitoring team conducts
evaluation and prepares for next rounds. After four rounds,
the cases developed are stored in the case base. Subsequently,
the whole case base is evaluated. Based on the evaluation, the
individual cases are entered into the case base of the CBR
system. The task of the ontology-based case-based reasoning
(CBR) for automotive ramp-ups is intended for storing
practical knowledge of the employees. It is particularly useful
to be reused in case a knowledge holder leaves the company.
Hence, it is used pragmatically to solve problems that would
occur in future automotive ramp-ups. Employees can use
experience from previous ramp-ups to help solve current
problems and therefore make the automotive ramp-up process
more efficient (Puppe et al. 2003). Case-based reasoning, as a
technique of artificial intelligence, compares a new problem
(case) with problem cases from previous ramp-ups already
stored in the system. When an inquiry is made, a new case is
compared with the cases already stored in the case base
(Aamodt and Plaza 1994). The programmed mathematical
algorithm calculates the similarity of the cases stored in the
case base with the new case. The experience and problem-
solving knowledge contained in the most similar cases can be
adapted or used as an aid for the new case in current
automotive ramp-ups and projects. By conducting repetitions,
the user ascertains whether the solution was successful or
unsatisfactory for the new case. However, the new case is
saved in the case base regardless of the result of this revision.
This ensures that unsuccessful solutions are not attempted
repeatedly. Thus the case base continues to grow through this
automatic learning, following a circular process (Cocea
2011). With the help of this methodology, the practical
knowledge of the employees is stored systematically and
used pragmatically in future automotive projects.
5. IMPLEMENTATION AND VALIDATION
Two individual software solutions were implemented based
on the unified concept and validated based on two different
pilot cases. The first case (Case I) represents potential
environmental impact of manufacturing of customized hood
assembly in decentralized manufacturing environment. The
other case (Case II) relates to ramp-up management in
screwing of automotive components. The Case I is shown in
Fig. 1. at the left side of the partition line XY. The web client
application of Evaluation and Decision Making Module
(EDMM) loads the production scheme file (xml file format)
from the web server. The Scheme Parser dissolves the
production scheme into several production nodes. The
information is sent nodewise to the Query Generator which
later on creates query relevant to the selected node in a
SPARQL format. A shark fish search based algorithms are
implemented to enable search in the knowledgebase based on
several ontologies. The results are generated by the Query
Engine and stored back to the knowledgebase as well as sent
to EDMM in an xml format over the web. The
knowledgebase is modelled using ontologies. A semantic
web reasoner is developed based on Jena Engine and
enriched with semantic web-rules concerning pragmatic cases
of the potential environmental impact of the possible
materials, resources and process technologies to consolidate
inference capabilities of knowledgebase. Fig.2. illustrates the
ontology modelled for this purpose with a specific case of
manufacturing customized hood assembly in a decentralized
manufacturing network.
Preprints of the 2013 IFAC Conference on Manufacturing Modelling, Management, and Control, Saint Petersburg State University and SaintPetersburg National Research University of Information Technologies, Mechanics, and Optics, Saint Petersburg, Russia, June 19-21, 2013
480
Fig. 2. Ontology for customized hood assembly
manufacturing process environment
For planners to search knowledge through the intelligent
assistance system, the client applications allows users
connected over web to customize their search more concrete
instead of searching it using natural language query where
precision of the searched information is a not ensured to a
great extent and often fetch high number of results.
Therefore, a graphical user interface has been developed that
guides a planner for customizing query for more concrete
search of information. This graphical user interface can
customize the specifications of product, process technology
and resource to support easy generation of SPARQL based
query which is easy to navigate information through the RDF
structure. Similarity algorithms are implemented along with
heuristic searching rules to navigate the information inside
the knowledgebase fast. The searched information is then
compiled in an xml file format and finally uploaded on
server. The software module connects the several modules
including databases, the ontologies and the graphical user
interface to the web server using freely available Apache
Tomcat Server.
For the second case (Case II), the software prototype has
been developed that provides planners and experts to conduct
several rounds of Delphi surveys in production ramp-up of
screwing processes. The automatic generation of
questionnaires for Delphi surveys is supported using
ontologies. The cases are generated after completion of
Delphi surveys. This ontology relates to ramp-up of screwing
processes in automotive sector and serves as a starting point
for gathering expert knowledge. On one hand, it is used to
represent the topics and terms from ramp-up management
area, for the experts in a hierarchical and connected manner.
On the other hand, the ontology structures the gathered expert
knowledge so that it can then be used by the CBR System.
The ramp-up management of screwing processes is validated
based on the concept with software programming of three
selected, refined, and linked methods (Ontology, Delphi
method, and CBR) in practice, This prototype is able to
model and represent the ontology as well as to conduct
Delphi surveys with its individual rounds efficiently. In
addition, the CBR system and the similarity algorithm are
also integrated. Since these three methods are to be linked to
each other, a common platform is created for their
intercommunications. The illustration (see Fig. 4. ) shows the
IT structure of the prototype.
Fig. 3. Graphical user interface for customizing user queries
The prototype offers an opportunity to implement ontologies
modeled with the standardized software like Protégé into the
system. In addition, the cases from the database can be linked
through the web server. The web server functions as an
interface for the direct exchange of information with the
database. A freely available Apache Tomcat server is used
for this purpose; the application for the Delphi survey is
installed on it.
The module of the Delphi survey implemented in the
software prototype is constructed flexibly. This means that
the layout or the user interface is made adjustable to different
requirements (company divisions). It is also important that
the different ontologies from the respectively relevant areas
of ramp-up management can be integrated. The structure of
the individual survey rounds is adjustable to the
corresponding ontologies. Through this way, the usability of
the software across divisions has been guaranteed. The
programmed CBR application is also implemented. The
database system used is PostgrSQL, which is also freely
available. It can be used to create user profiles and
passwords, for example, which allow the assignment of the
cases to the respective experts and govern the individual
rights of use. The expert knowledge gathered by the Delphi
surveys and the CBR System is further saved in the database.
The database recognizes whether a case entered within the
Delphi survey is completely concluded or must still be
improved by the experts in further survey rounds. Also, the
Preprints of the 2013 IFAC Conference on Manufacturing Modelling, Management, and Control, Saint Petersburg State University and SaintPetersburg National Research University of Information Technologies, Mechanics, and Optics, Saint Petersburg, Russia, June 19-21, 2013
481
detailed information from the CBR survey for determining
similar cases from the past is saved in the database. This data
is used again finally as an input for further development of
the case base at the later stage. After having been shown the
results of similar cases from past ramp-ups via the web
server, the user can ascertain from them new solution
approaches for confronted ramp-up problem. The newly
gained knowledge during the solution should be input back to
the CBR case base so that it is available to other users for
global access. For this reason, the database makes available
the data that was previously entered into the system while
searching, in order to be able to save new cases in the system
built upon it.
Fig. 4. Structure of the prototype for screw process ramp-up
management
With the IT structure shown, it is possible to access the
programmed applications from any workstation globally
through the internet. For example, experts residing on
different continents may participate in Delphi surveys. It is
also possible to use the CBR system from anywhere in the
world. A global knowledge management system is thus
created making it possible for the globally available know-
how in the area of ramp-up management to be identified
(Delphi survey), stored in an intelligent structure (ontology),
and utilized pragmatically by all users (CBR system). This
global usage of the system is of particular importance. By this
means a comprehensive knowledge base for the ontology-
supported CBR System can be built and used to master the
challenges for a knowledge management system for vehicle
ramp-ups described at the beginning.
6. CONCLUSIONS AND OUTLOOK
A unified concept for integrating the two software solutions,
for extracting and reusing the implicit knowledge from the
planners as well as the planning related softwares have been
developed and validated through pilot cases. The first pilot
case relates to the environmental impact assessment of
customized hood manufacturing in the decentralized
manufacturing network. Web-based software solution
comprised of customized user query interface, ontology
based knowledge base and Scheme parser is implemented
and validated for the pilot case. The software takes input
through two means i.e. direct from users through customized
user queries for searching the information from the
knowledgebase and deliver results relevant to the query. This
query is generated after the user has customized his input
information. In case of data file as input to the intelligent
assistance system, the information contained in the data file is
first parsed through Scheme Parser and Query is generated in
a sequential manner against the parsed information. The
query engine is used to search information from ontologies.
The information is delivered to the decision module for
processing further in decision making process. The second
pilot cases relates to the extraction, storage and automatic
learning from personal implicit knowledge through formal
means. The knowledge is discovered through web-based
multi-round Delphi surveys and later stored in ontology part
of Case Base Reasoning module. The software is tested for
ramp-up of screwing process.
In future, the software will be extended to implement and use
ontologies for inferring generic input-output environment
simulation model of material, process and resource based on
the nature of manufacturing process. This model will be
directly used inside the LCA simulation tool. The generated
results will be used to enrich the knowledgebase for better
inference capabilities. The validation scenario will be
expanded to assist decision making for manufacturing of
other products such as customized orthotics in a decentralized
manufacturing network. For enhancing the software module
dedicated for screwing process ramp-up, the same approach
will be used for rolling out further topic areas in the ramp-up
management. For this purpose ontologies that structure the
knowledge domains of the respective areas must first be
created with the relevant experts. It will be followed by
Delphi surveys, so that the case base can be filled with the
valuable empirical knowledge of ramp-up management
experts.
For both modules of the software representing the
corresponding pilot cases, the unified software will be tested
by employees and further refinement will be made based on
the feedback from the end users.
7. ACKNOWLEDGEMENTS
The work reported is this paper is partially supported by the
European Commission Project e-Custom FP7-2010-NMP-
ICT-FoF 260067.
REFERENCES
Aamodt, A., Plaza, E. (1994). Case-Based Reasoning:
Foundational Issues, Methodological Var-iations, and
System Approaches. Artificial Intelligence, 7/1, pp. 39-
52.
Beissel, S. (2011). Ontology-based Cased-based Reasoning.
Betriebswirtschaftlicher Verlag Gabler. 1. edition.
Wiesbaden. ISBN-10: 3834930644. ISBN-13: 978-
3834930644.
Work station Web server
Ontology
Database
Production
equipment
Screw
surface
Screw
parts
handling
Screw
material
Loose part
Screw surface
«�
«
«��
«
«�
Retrieval
Thing
Screwing
toolPeriphery
Screwing
program
Parts
surface
Parts
handling
Parts
material
Screw connections
Production
equipment
Relation screw
Relation part
Screw Parts
«�
«
«
«�«
Preprints of the 2013 IFAC Conference on Manufacturing Modelling, Management, and Control, Saint Petersburg State University and SaintPetersburg National Research University of Information Technologies, Mechanics, and Optics, Saint Petersburg, Russia, June 19-21, 2013
482
Cegielski, C.G. (2001) A Model of the Factors that Affect the
Integration of Emerging Information Technology Into
Corporate Strategy. Unpublished Doctoral Dissertation,
University of Mississippi.
Cocea, M. (2011). User Modelling and Adaption in
Exploratory Learning. Doctoral Dissertation, University
of London.
F. Puppe, H. Stoyan und R. Studer (2003). Knowledge
Engineering. In Görz, G., Rollinger, C.-R.,
Schneeberger, J. (eds.) Handbuch der Künstlichen
Intelligenz, 4rth edition, pp. 599±641. Oldenbourg,
München.
Hayne, S., Pollard, C. (2000) A comparative analysis of
critical issues facing Canadian information systems
personnel: a national and global perspective. Information
& Management, 38 (2), pp. 73±86.
Hefke, M. (2009). Technology-based Introduction of
Knowledge Management. Südwestdeutscher Verlag für
Hochschulschriften. Saarbrücken. ISBN-10:
3838101472, ISBN-13: 978-3838101477.
Heisig, P., Orth, R. (2005). Knowledge Management
Frameworks in research and practice. European Resarch
Center for Knowledge and Innovation. 1. edition. TU
Berlin. ISBN-10: 3000172440. ISBN-13: 978-
3000172441.
Holsapple, P., Joshi, K. (2002) Knowledge manipulation
activities: results of a Delphi study. In-formation &
Management, 39 (6), pp. 477±490.
Kuhn, A., Wiendahl, H.-P., Eversheim, W., Schuh, G.
(2002)³fast ramp-up". Verlag Praxiswissen, Dortmund.
Lai, V., Chung, W. (2002) Managing international data
communications. Information & Man-agement, 45 (3),
pp. 89±93.
Minhas, S., Juzek, C., Berger, U. (2012)a. Ontology based
intelligent assistance system to support manufacturing
activities in a distributed manufacturing environment. In
Proceedings of 45th CIRP Conference on Manufacturing
Systems (CIRP CMS), Athens, Greece.
Minhas, S., Juzek, C., Berger, U. (2012)b. Elicitation of
requirements for a knowledge management concept in
decentralized production planning. In International
Conference on Mechanical, Industrial and Manufacturing
Technology (WASET2012), Berlin, Germany.
Mulligan, P. (2002) Specification of a capability-based IT
classification framework. Information & Management,
39 (8), pp. 647±658.
Nambisan, S., Agarwal, R., Tanniru, M.: Organizational
mechanisms for enhancing user innovation in
information technology. MIS Quarterly, 23 (3), pp. 365±
395.
Okoli, C., Pawlowski, S. D. (2004). The Delphi method as a
research tool: an example, design considerations and
applications. Information & Management, 42, pp. 15-29.
Brancheau, J.C., Janz, B.D., Wetherbe, J.C. (1996). Key
issues in information systems manage-ment, 1994±95
SIM Delphi results. MIS Quarterly, 20 (2), pp. 225±242.
Schmidt, R.C., Lyytinen, K., Keil, M., Cule, P. (2001).
Identifying software project risks: an in-ternational
Delphi study. Journal of Management Information
Systems, 17 (4), pp. 5±36.
Skulmoski, G. J., Hartmann, F. T., Krahn, J. (2007) .The
Delphi Method for Graduate Research. Journal of
Information Technology Education, Volume 6, pp. 1-21.
Weimer, G., Seuring, S. (2008). Information needs in the
outsourcing lifecycle. In Industrial Management & Data
Systems, Vol. 108 No. 1, pp. 107-121.
Preprints of the 2013 IFAC Conference on Manufacturing Modelling, Management, and Control, Saint Petersburg State University and SaintPetersburg National Research University of Information Technologies, Mechanics, and Optics, Saint Petersburg, Russia, June 19-21, 2013
483