Adaptive agent-based manufacturing control and its application to flow shop routing control

Abstract

The manufacturing industry is currently facing unprecedented challenges from changes and disturbances. The sources of these changes and disturbances are of different scope and magnitude. They can be of a commercial nature, or linked to fast product development and design, or purely operational (e.g. rush order, machine breakdown, material shortage etc.). In order to meet these requirements it is increasingly important that a production operation be flexible and is able to adapt to new and more suitable ways of operating. This paper focuses on a new strategy for enabling manufacturing control systems to adapt to changing conditions both in terms of product variation and production system upgrades. The approach proposed is based on two key concepts: (1) An autonomous and distributed approach to manufacturing control based on multi-agent methods in which so called operational agents represent the key physical and logical elements in the production environment to be controlled – for example, products and machines and the control strategies that drive them and (2) An adaptation mechanism based around the evolutionary concept of replicator dynamics which updates the behaviour of newly formed operational agents based on historical performance records in order to be better suited to the production environment. An application of this approach for route selection of similar products in manufacturing flow shops is developed and is illustrated in this paper using an example based on the control of an automobile paint shop.

Keywords:

• adaptive manufacturing control
• operational agent
• agent-based manufacturing control
• flow shop routing

Acknowledgments

Yi Peng is a PhD student in the Manufacturing Automation and Control Systems Research Group with Dr. Duncan McFarlane. Before joining the Manufacturing Engineering Group in October 1996, Yi Peng has studied and lectured in Department of Automation, Tsinghua University, Beijing, China, for 11 years. He obtained his BE (Automatic Control) and ME (System Simulation) in 1990 and 1992. Between 1992–1993, he also worked as an engineer in a state-owned aero-electronics factory in Shaanxi Province. In Tsinghua University, he worked in The State CIMS (Computer Integrated Manufacturing System) Engineering Research Centre, which is a winner of the 1994 CASA/SME University Lead Award recipient. He was a member of the steering committee of the Concurrent Engineering Project, sponsored by the State High-Tech Programme (863 Programme).

Duncan McFarlane is a Senior Lecturer in Manufacturing Engineering in the Cambridge University Engineering Department, and head of the Centre for Distributed Automation and Control within the Institute for Manufacturing. He has been involved in the design and operation of manufacturing and control systems for over years. He completed a BEng degree at Melbourne University in 1984, a PhD in the design of robust control systems at Cambridge in 1988, and worked industrially with BHP Australia in engineering and research positions between 1980 and 1994. Dr McFarlane joined the Department of Engineering at Cambridge in 1995 where his work is focused in the areas of response and agility strategies for manufacturing businesses, distributed (holonic) factory automation and control, and integration of manufacturing information systems. He is particularly interested in the interface between production automation systems and manufacturing business processes. Dr McFarlane is involved in UK, EU and international research grants investigating the responsiveness of manufacturing production systems and the development of distributed, reconfigurable (‘holonic’) manufacturing control systems and has led two overseas missions in the same areas. He is an editor of Springer Verlag's Industrial Control series and in 1999 was co-editor of a Special Issue in International Journal of Production Management on Manufacturing Responsiveness. He is a member of several EU and UK committees examining future manufacturing systems. In 2000 became Chairman of the Subcommittee on Manufacturing for the IEEE Systems, Man and Cybernetics Technical Committee on Robotics and Manufacturing Systems. He is the European Research Director of the Auto-ID Center which is a collaborative project involving MIT, University of Adelaide, Fudan University, M Lab and Keio University and 85 industrial partners (www.autoidcenter.org). In 2001 he also became a co-investigator in the EPSRC funded Innovative Manufacturing Research Centre based in the Institute for Manufacturing.