Abstract

In the area manufacturing area, the simulation is a essential tool for the validation of methods and architectures before apply them in
a manufacturing environment. Current Simulation Tools conduct simulation of manufacturing environments based on static models that
make use of the traditional sequential and centralized programming for manufacturing processes, where the mechanisms of planning and
control offer insufficient flexibility to respond to the manufacturing styles that change continually and manufacturing
environments highly mixed and low volume. Consequently, conventional simulation tools limit the scalability and reconfigurability for
manufacturing systems modeling that allow to adapt models to the changing needs of the client. It is difficult to find a simulation
tool that can ``intelligently'' execute simulation tasks more and more complex. The difficulty lies in the inclusion of the knowledge
managed by the original system in the tool and that the tool acts as a wizard that provides advices and guides the User during the
simulation. Therefore, the need for new simulation tools for factories emerges in order to cover features such as: a) flexibility
and adaptability, to model complex behaviors of a manufacturing system, b) scalability for the inclusion of additional
functionality, c) proactive and reactivity to the automatical adaptation faced with the environment changes, and d) learning
features (intelligence) based on the experience gained during the simulation.

The artificial intelligence techniques have been used for the intelligent manufacturing for more than two decades. The techniques
of the artificial intelligence area allow the definition of manufacturing entities that are distributed, autonomous,
intelligent, flexible, fault tolerant and reusable, which operate as a set of entities that work together. In addition, the recent
developments in the Multi-agent Systems area have brought with them new and interesting possibilities. Some researchers have applied
agent technology  in the integration of the business manufacturing, collaboration, planning of manufacturing processes, the scheduling
for shop floor control, material handling and inventory management, as well as the implementation of new types of Manufacturing Systems
such as Holonic Manufacturing Systems. Taking into account these successful applications of the Multi-agent Systems in Intelligent
Manufacturing, we are convinced that this technology can also improve the performance of the Intelligent Manufacturing Systems
Simulation.

In this thesis we propose the definition of an architecture for a Simulation Environment for Manufacturing Systems supported by
agents. This architecture integrates the functionality of a traditional simulation tool, it also enables the simulation of
complex behaviors linked to Intelligent Manufacturing Systems and provides solutions and improvements that are adapted to the
requirements of the new era of manufacturing. Therefore, the architecture focuses on improving tasks of the global simulation
process that include: a) assistance during the representation and programming models of Manufacturing Systems, b) provide flexibility
to set out the scenarios (hypothesis) and the experiment designs, c) the model simulation, d) provides metric set for the evaluation of
models, shop floor configuration and production data in order to help the User during decision-making process, and e) the validation
of the results against the hypothesis. Also, it is proposed a Metamodel supported by agents that supports to the Architecture during 
the design and the programming of Models of Intelligent Manufacturing Systems. The Metamodel offers flexibility to the definition of models 
with a wide range of variants that allows the User to design experiments that take into account the requirements of the new manufacturing era. 
Through the individual design of the system entities it is possible to define complex behaviors of real Manufacturing Systems. The Metamodel
offers the possibility to include patterns to define the interactions among the Production Orders and Factory Resources
during the allocation of tasks. In this way, the architecture uses the Metamodel to facilitate the design of Simulation Models of
Manufacturing Systems supported by agents and the graphical animation of these models. Finally, to validate the architecture and
the Metamodel has been implemented a prototype that provides easy to use interfaces that provide support to the creation and simulation
models. The prototype has been used for the simulation of a study case: a manufacturing system for silos (a product container).