An agent-based methodology for manufacturing decision making: a textile case study

Abstract

Decentralised decision making and real-time response to the unforeseen changes, often taking place in both the manufacturing and market environments, are two important factors that affect the flexibility, required for a production chain to follow demand. This article describes an agent-based methodology, for addressing both real-time and decentralised manufacturing decision-making challenges. At the manufacturing levels of production and processes, the agents employ a mechanism to generate local alternatives as well as a message exchange procedure to build decision trees, which are traversed and evaluated via user-defined objective functions. The generic structure used in the approach allows in a wide range the application of production system configurations. The real-time information, required for monitoring the system status and for generating valid alternatives, is obtained through integration with the existing information systems. An application to the textile industry is presented and discussed.

Keywords:

• agent-based systems
• decision support systems
• scheduling

Acknowledgements

We thank the people of Biokarpet S.A. (the carpet producer in northern Greece) and Mr Fanikos for their contribution. This research was partially supported by grants from RIDER, an ESPRIT IV funded project, and MyCar IP, an FP6 funded project.