Looking for a job position in functional modelling

Kamreddine Ouliddren (kouliddr@sckcen.be)
Mon, 2 Mar 1998 21:00:17 +0100 (MET)

This is a multi-part message in MIME format.
--------------1469E4B71FA0B54EEA0F0309
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7bit

Dear all,
I am about to finish my PhD dissertation in functional modelling and I
am
looking for a position in this field. Therefore, I am addressing this
email
to communities that might be interested in the domain.
My PhD dissertation discusses :" An approach for incremental fuzzy
modelling
of dependencies in complex physical systems".
My work concerns essentially an implementation of fuzzy reasoning in a
Functional modelling scheme to model dynamic behaviours of dependent
components in a complex system.
If you want to have more details about me and my work, please feel free
to
contact me or to read my homepage below. The URL includes a link to my
resume.

Looking forward to hearing from you.

Best regards.

-------------------------------------------------------
Kamreddine Ouliddren
Nuclear Research Centre SCK-CEN
Boeretang 200, B-2400 Mol, Belgium
E-mail: kouliddr@sckcen.be
URL: http://www.sckcen.be/dept/lhma/kohpage.htm
Tel: +32 14 33 24 34. Fax: +32 14 32 13 36

--------------1469E4B71FA0B54EEA0F0309
Content-Type: text/html; charset=iso-8859-1; name="kohpage.htm"
Content-Transfer-Encoding: 8bit
Content-Disposition: inline; filename="kohpage.htm"
Content-Base: "http://www.sckcen.be/dept/lhma/kohpage
.htm"

<!DOCTYPE HTML PUBLIC "-//W3C//DTD HTML 3.2//EN">
Functional modeling

An Approach to Incremental Fuzzy Modelling of Dependencies in Complex Systems.

Kamreddine Ouliddren.

E.mail: kouliddr@sckcen.be

 

Keywords:

Fuzzy Sets, Functional Modelling, Clustering.

Summary:

To build a computer human support system to monitor complex plants, one always has to deal with the major concern of knowledge representation. I have adopted a functional centred modelling approach called Goal Tree-Success Tree Master Plant Logic Diagram (GTST_MPLD). The approach models a complex system in hierarchical levels. There exist different types of hierarchies that can be considered such as: functional hierarchy, structural hierarchy, goal hierarchy, event hierarchy and behavioural hierarchy. Some other kind of hierarchies may be found such as procedural hierarchy that describe levels of procedures used to monitor a complex system.

The GTST_MPLD is a logic based scheme that evaluates a complex system's states at discrete intervals of time. In its essence, the GTST_MPLD scheme embodies three types of hierarchies: Functional, Structural and Goal hierarchy. A new diagram based on fuzzy logic has been developed to represent the dynamic behaviour of components in complex systems. This diagram, is called Dynamic Master Logic Diagram (DMLD). Using DMLD allows to represent continuous state functions, as such a more realistic evaluation of the complex system can be done. The DMLD is aimed to embody the behavioural type of hierarchy.

Furthermore, I propose a learning capability to DMLD that allows to learn the unknown characteristics of the behaviour of components. The model uses real data that represent dependencies between components. The learning capability called is based on a cluster analysis approach that assesses the history data of the concerned components. To minimize the rule base of the fuzzy system, the model subdivise the input-output domain into a set of linear hyperplanes. A minimum number of linearily dependent clusters can be identified in a hyperplane. These clusters will form the fuzzy rules. Such as, in a two-dimensions plane, the behavior of the function is approximated by only two, well choosen, rules. The same idea is applied to n-dimensions spaces. The behavior of the function is approximated by n, linearily independent, clusters. Clustering is adopted to group data, to consider data noise and the sensitivity allowed for the ‘approximation' of the dynamic behaviour.

The work is being supported by Nuclear Research Center SCK-CEN and will be submitted to Faculty of Science in the Vrije Universiteit Brussel for a PhD degree.

Related People:

--------------1469E4B71FA0B54EEA0F0309--