Scheduling just-in-time part replenishment of the automobile assembly line with unrelated parallel machines

With increasing product customization, just-in-time part replenishment has become a significant scheduling problem in the automobile assembly system. This paper investigates a new unrelated parallel machine scheduling problem of an assembly line, where machines are employed to deliver material boxes from an in-house warehouse to workstations. The schedule is to appropriately specify the assignment and sequence of material boxes on each machine for minimizing line-side inventories under no stock-out constraints. By taking advantages of domain properties, an exact algorithm is developed to cope up with small-scale instances. In terms of real-world scale instances, a hybrid teaching-learning-based optimization metaheuristic is established by integrating teaching-learning-based optimization with a beam search technique. Experimental results indicate that the scheduling algorithms are effective and efficient in solving the proposed unrelated parallel machine scheduling.