Distributed intelligent control of exceptions in reconfigurable manufacturing systems |
| |
Authors: | Manfredi Bruccoleri Michele Amico Giovanni Perrone |
| |
Affiliation: | 1. Dipartimento di Tecnologia Meccanica, Produzione e Ingegneria Gestionale, Faculty of Engineering , University of Palermo , Viale delle Scienze, Palermo, 90128, Italy;2. Dipartimento di Ingegneria e Fisica Ambientale (DIFA) , University of Basilicata , C. da Macchia Romana, Potenza, 85100, Italy |
| |
Abstract: | In order to react to the continuous and unpredictable changes in product demand, in product variety, and in process technologies, reconfigurable manufacturing systems allow quick adjustment of production capacity and functionality by rearranging or changing their modular components. In this kind of system, operation management issues, such as exception handling policies, become more complex since correct reconfiguration strategies have to be selected. This paper explores the potential of the reconfigurability feature to be a basis for the development of new strategies to handle out-of-the-ordinary events in the production process; in particular, maintaining production flow when machine breakdowns occur. Decisions regarding how to deal with exceptions to the production process are complex and depend on the manufacturing system configuration and on many performance and economic variables. The authors propose agent-based manufacturing control for exception handling because of its ability to be very agile, as well as being reactive and efficient. Manufacturing agents, while working to pursue their specific goals, achieve the global target of the system. Complex decisions can be made due to the synergy arising from the agents' internal reasoning and the negotiation process among these agents. The adopted negotiation mechanism is based on the contract-net protocol, while different strategies have been designed for the internal reasoning. The authors demonstrate that, under certain conditions, an agent's internal strategies based on fuzzy reasoning improve the global performance of the system. The proposed control model has been tested on a discrete event simulation test-bed. |
| |
Keywords: | |
|
|