Optimal Design of Flexible Job Shop Scheduling Under Resource Preemption Based on Deep Reinforcement Learning

With the popularization of multi-variety and small-batch production patterns, flexible job shop scheduling problem (FJSSP) has been widely studied. The sharing processing resources by multiple machines frequently occurs due to space constraints in a shop, which results resource preemption for workpieces. Resource complicates problems that are otherwise difficult solve. In this paper, under process scenario is modeled, two-layer rule algorithm based on deep reinforcement learning proposed achieve goal minimum time. simulation experiments compare our with two traditional metaheuristic optimization algorithms among different distribution scenarios static environment. suggest more effective than meta-heuristic application scenarios. Ablation experiments, generalization, dynamic performed demonstrate excellent performance method FJSSP preemption.