A multi-agent based rescheduling framework for mixed-model assembly line balancing

In general, company responds to customers´ diversified needs, the necessity to make production system more versatile and flexible. The assembly line must change from fixed assembly lines to mixed-model assembly lines (MMAL). Because an assembly line is dynamic, and when system disturbance occur, rescheduling is necessary to adjust or regenerate an effective feasible schedule. For maximizing the productivity of the dynamical MMAL, the paper develops a decentralized framework using multi-agent system to reschedule MMAL balancing problem when unexpected disturbance occurred. A three-level multi-agent system is constructive, and integrates dispatching rule and meta-heuristic techniques. In scheduling stage, two criteria are simultaneously considered for optimization: to minimize the number of workstation for a given cycle time, and to maximize the workload smoothing. In each rescheduling stage, schedule is generated using a meta-heuristic that allows two objectives, efficiency and stability, to be balanced.