Experimental data and the executable to test the performance of situation-adaptive policy for container stacking in an automated container terminal

Determining where to stack the containers at the storage yard of a container terminal is an important problem because that decision critically affects the efficiency of container handling in the yard and eventually the efficiency of the vessel operations that is considered the most important for the productivity of the whole terminal. One limitation of the stacking policies previously proposed is that they are static in nature. Although good locations for stacking may change as the workload of vessel operation changes, the previous policies are insensitive to such changes. Failure to recommend good locations leads to elongated operations of yard cranes and thus makes them hard to keep up with the workload of vessel operation. In this paper, we propose a method for deriving a dynamic policy that can adapt to the workload of vessel operation that changes over time. Our method derives two boundary polices: one for very high workload and the other for very low. Then, a policy appropriate for any intermediate workload can be synthesized from the two boundary policies through an interpolation. Simulation experiments showed that the proposed policy significantly reduced overall container handling time compared to the previous static policy. When measured in terms of the time the transportation vehicles wait for container handling services, the improvement was about 19%.在集装箱码头（container terminal）的堆场（storage yard）中规划集装箱堆存位置是一项关键问题，该决策直接影响堆场的集装箱装卸效率，并最终决定船舶作业效率——而船舶作业效率被视为影响整个码头生产效能的核心指标。 现有已提出的堆存策略（stacking policies）存在一项核心局限：其本质为静态策略。尽管最优堆存位置会随船舶作业作业负荷（workload）发生动态变化，但传统静态策略无法适配此类工况变动。若无法推荐适配当前场景的堆存位置，会导致堆场起重机（yard cranes）作业时长被拉长，进而难以匹配船舶作业的作业负荷需求。 本文提出一种可适配随时间动态变化的船舶作业负荷的动态堆存策略生成方法。该方法首先生成两类边界策略（boundary policies）：分别对应极高作业负荷与极低作业负荷场景，随后通过插值（interpolation）算法从这两类边界策略中合成适用于任意中间作业负荷的堆存策略。 仿真实验结果表明，相较于传统静态策略，本文所提动态策略可显著降低整体集装箱装卸时长。以运输车辆等待集装箱装卸服务的时长为评估指标时，该策略的性能提升幅度约为19%。