Assessing self-organization and emergence in Evolvable Assembly Systems (EAS)

Dissertação para obtenção do Grau de Mestre em Engenharia Electrotécnica e de ComputadoresThere is a growing interest from industry in the applications of distributed IT. Currently, most modern plants use distributed controllers either to control production processes, monitor them or both. Despite the efforts on the last years to improve the implementation of the new manufacturing paradigms, the industry is still mainly using traditional controllers. Now, more than ever, with an economic crisis the costumers are searching for cheap and customized products, which represents a great opportunity for the new paradigms to claim their space in the market. Most of the research on distributed manufacturing is regarding the control and communication infrastructure. They are key aspects for self-organization and there is a lack of study on the metrics that regulate the self-organization and autonomous response of modern production paradigms. This thesis presents a probabilistic framework that promotes self-organization on a multiagent system based on a new manufacturing concept, the Evolvable Assembly Systems/Evolvable Production Systems. A methodology is proposed to assess the impact of self-organization on the system behavior, by the application of the probabilistic framework that has the dual purpose of controlling and explaining the system dynamics. The probabilistic framework shows the likelihood of some resources being allocated to the production process. This information is constantly updated and exchanged by the agents that compose the system. The emergent effect of this self-organization dynamic is an even load balancing across the system without any centralized controller. The target systems of this work are therefore small systems with small production batches but with a high variability of production conditions and products. The agents that compose the system originated in the agent based architecture of the FP7-IDEAS proejct. This work has extended these agents and the outcome has been tested in the IDEAS demonstrators, as the changes have been incorporated in the latest version of the architecture, and in a simulation and more controlled environment were the proposed metric and its influence were assessed