Minimize makespan of permutation flowshop using pointer network

ABSTRACT During the shipbuilding process, a block assembly line suffers bottleneck when largest amount of material is processed. Therefore, scheduling optimization important for productivity. Currently, sequence inbound products controlled by determining input using heuristic or metaheuristic approach. However, algorithm has limitations in that computation time increases exponentially as number objects increases, and separate calculations are required every problem. Also, such dispatching limitation exploring problem domain. this study tries reinforcement learning based on pointer network to overcome these limitations. Reinforcement with found be suitable permutation flowshop problem, including input-order optimization. A trained neural applied without re-learning, even if inputs changed. The model shows meaningful results compared algorithms makespan time. outperforms within limited range