Deep Reinforcement Learning for Distributed Flow Shop Scheduling with Flexible Maintenance

A common situation arising in flow shops is that the job processing order must be same on each machine; this referred to as a permutation shop scheduling problem (PFSSP). Although many algorithms have been designed solve PFSSPs, machine availability typically ignored. Healthy conditions are essential for production process, which can ensure productivity and quality; thus, deteriorating effects periodic preventive maintenance (PM) activities considered paper. Moreover, distributed networks, manufacture products quickly, of increasing interest factories. To end, paper investigates an integrated optimization PFSSP with flexible PM. With introduction constraints multi-factory scheduling, complexity computation time solving increases substantially large-scale arithmetic cases. In it, deep Q network-based solution framework diminishing greedy rate The proposed compared DQN fixed rate, addition two well-known metaheuristic algorithms, including genetic algorithm iterated algorithm. Numerical studies show application approach studied production-maintenance joint exhibits strong performance generalization abilities. suitable interval also obtained, some managerial insights.