Dynamically Integrated Manufacturing Systems (DIMS)—A Multiagent Approach

D. Z. Zhang,A. Anosike,Ming Kim Lim
DOI: https://doi.org/10.1109/TSMCA.2007.897710
2007-01-01
Abstract:Manufacturing businesses in today's market are facing immense pressures to react rapidly to dynamic variations in demand distributions across products and changing product mixes. To cope with the pressures requires dynamically integrated manufacturing systems (DIMS) that can manage optimal fulfillment of customer orders while simultaneously considering alternative system structures to suit changing conditions. This paper presents a multiagent approach to DIMS, where production planning and control decisions are integrated with systems reconfiguration and restructure. A multiagent framework, referred to as a hierarchical autonomous agent network, is proposed to model complex manufacturing systems, their structures, and constraints. It allows the hierarchical structures of complex systems to be modeled while avoiding centralized control in classical hierarchical/hybrid frameworks. Subsystems interact heterarchically with product orders to carry out optimal planning and scheduling. An agent coordination algorithm, operating iteratively under the control of a genetic algorithm, is developed to enable optimal planning and control decisions for order fulfillment to be made through interactions between agents. This algorithm also allows the structural constraints of systems to be relaxed gradually during agent interaction, so that planning and control are first carried out under existing constraints, but when satisfactory solutions cannot be found, subsystems are allowed to regroup to form new configurations. Frequently used configurations are detected and evaluated for system restructure. The approach also enables Petri-net models of new system structures to be generated dynamically and the structures to be evaluated through agent-based discrete event simulation.
What problem does this paper attempt to address?