In this thesis, we propose a framework for modeling requirements of manufacturing
systems (MS). This framework is intended to help MS engineers to
perform their modeling tasks in today's ever-changing environment. It relies
on two major ideas: a multi-formalism approach, combining several
languages into a coherent formalism, and a component-based modeling approach,
arguing that models of manufacturing systems could be build partly
by assembling and customizing reusable model components.
A multi-formalism approach consists in using several formalisms to
create a model of a single system. The motivation for the multi-formalism
approach is that each modeling languages used is partially different from
the others and has its own flaws and strengths.
The modeling framework proposed combines the Albert II, i* and CIMOSA
languages. Albert II is a formal agent-oriented language for modeling the
requirements of real-time distributed systems. CIMOSA is a language dedicated
to manufacturing systems and covers many aspects of these systems.
i* is a framework aimed at the modeling of rationales underlying organizational
structures. The combination of languages yields a more expressive, more
natural and more precise formalism.
The combination is achieved through meta-modeling and the definition
of a set of mapping rules that establish a correspondence among some of
the concepts of the three formalisms. Guidelines on how to apply these
rules in order to enhance models are provided.
The framework is also component-based since it provides a number of
patterns for modeling the requirements of MS. Each of them is structured
as a set of classes that have relationships with each other and with classes
from other patterns. They can be reused and customized for the application
at hand.
Patterns are first expressed independently of Albert II, i* and CIMOSA.
A set of mapping rules similar to those between languages are defined to
obtain a corresponding model in Albert II, i* and CIMOSA.
Patterns proposed include a pattern for Parts, for special classes
of manufacturing Functions, for very simple Control strategies, for Manufacturing
Goals and for manufacturing active Entities (agents).