集装箱码头作业系统层次化、并行、异构与可重构计算模型

基于计算思维和计算透镜, 分析了集装箱码头的装卸作业与调度决策, 基于“并行计算”、“异构计算”和“可重构计算”提出了计算物流视角下的集装箱码头作业层次化、并行、异构与可重构计算模型; 将计算机科学领域中多种典型计算体系结构的设计思想和运作机制, 泛化、迁移、修正、融合和定制到集装箱码头作业系统中, 设计了面向此计算模型的混合调度策略, 提出了集装箱码头调度新的抽象计算模型与工程解决路径; 以某大型集装箱码头为实例, 基于集装箱码头作业层次化、并行、异构与可重构计算模型, 进行了物流广义计算自动化的设计与性能评估。研究结果表明: 采用计算模型能确定码头的集装箱吞吐量上限, 实例中约为码头年设计能力的2.75倍; 在满负荷情况下, 基于等待作业集装箱队列的负载均衡调度策略和基于等待作业船型的负载均衡调度策略均能将大型集装箱干线船舶物流广义计算任务延迟缩短约17 h; 在明显作业过载时, 前者能将物流广义计算任务延迟减少100~110 h, 后者能减少约120 h; 在满负荷和作业过载情况下, 2种策略均能缩短大型集装箱干线船舶物流广义计算访问存储时间1~2 h, 后者在作业过载情况下表现更佳; 2种策略都能很好地优先服务重点班轮集合, 且有各自对应的适用状况和调度重点, 码头管理者可根据具体情况选择适用。 

Hierarchical,parallel, heterogeneous and reconfigurable computation model of container terminal handling system

Based on computational thinking and computational lens, the loading and unloading operations and the scheduling decisions of container terminals were analyzed. Based on the parallel computation, heterogeneous computation and reconfigurable computation, a hierarchical, parallel, heterogeneous, and reconfigurable computation model of container terminal handling (HPHRCM-CTH) from the perspective of computation logistics was proposed. The design philosophies and operational mechanisms of typical computing architectures in the computer science were generalized, migrated, modified, fused, and customized to the container terminal handling system (CTHS), and the hybrid scheduling strategy for the HPHRCM-CTH was presented. A new abstract computation model and engineering solution to the container terminal scheduling were put forward. Taking a large container terminal as an example, the design and performance evaluation of logistics generalized computation automation were carried out based on the HPHRCM-CTH. Analysis result shows that the HPHRCM-CTH can determine the upper limit of container throughput that is about 2.75 times of the annual design capacity of the container terminal in the example. At the condition of full load, the scheduling strategies of load balancing for the pending queues of containers (LB-PQC) and ship types (LB-PQS) can shorten the logistics generalized computation task latency (LGC-TL) of large container mainline ships by about 17 h. At the condition of obvious job overload, the LB-PQC can reduce the LGC-TL by 100-110 h, while the LB-PQS can reduce the LGC-TL by about 120 h. At the conditions of full load and job overload, the LB-PQC and LB-PQS can reduce the logistics generalized computation memory access time (LGC-MAT) for large container mainline ships by 1-2 h, and the LB-PQS performs better under the conditions of job overload. The LB-PQC and LB-PQS both can give well priority to the key service liners, and have the respective applicable condition and scheduling emphasis, and the terminal manager can choose the right one according to the specific situation.