FBS-enhanced agent-based dynamic scheduling in FMS

Flexible manufacturing systems (FMSs) are highly automated and require effective scheduling approaches to improve the system performance and integrate various control decision-making activities. In this paper, a multi-agent approach integrated with a filtered-beam-search (FBS)-based heuristic algorithm is proposed to study the dynamic scheduling problem in a FMS shop floor consisting of multiple manufacturing cells. The approach is based on a hybrid architecture which is composed of a set of distributed agents each using local information to generate real-time schedules. The FBS-based algorithm is embedded into agents as the scheduling engine, which makes improvement in the generation procedure of branches to obtain optimal or sub-optimal schedules quickly. A FIPA CNIP-based negotiation mechanism is utilized to realize coordination of agents to achieve a globally good schedule. To demonstrate the feasibility and effectiveness of the proposed approach, a proof-of-concept prototype system has been implemented in the JADE™ platform. Computational experiments of the approach are also presented in comparison with those of other methods.