【作者】 钟明；

【导师】 缪立新； 周受钦；

【作者基本信息】 清华大学 ， 物流工程（专业学位）， 2013， 硕士

【摘要】 集装箱运输是国际贸易的重要组成部分。随着集装箱吞吐量的增长，集装箱码头的作业效率和作业能力面临着越来越严峻的考验。对作业流程进行优化调度和合理配置，是提升码头竞争力的重要手段。集装箱码头的作业流程包含多种作业计划。本文主要针对出口集装箱堆场内的贝内箱位分配问题及预倒箱问题进行研究。针对出口集装箱贝内箱位分配问题，本文改进了一个动态规划模型，使其更加符合实际情况。数值结果表明改进后的模型能更好的设计堆放策略，从而有效的减少翻倒箱次数。本文还针对贝内箱位分配问题提出了一个两阶段的启发式算法：第一阶段基于邻域搜索的启发式算法，设计出各重量类型的集装箱的在不同堆垛形态的优先堆放次序；第二阶段针对集装箱集港过程的特点，提出基于翻滚策略的箱位堆放局部优化算法，使得集装箱在优先堆放次序之外拥有更多选择空间，从而有效的减少翻倒箱次数。数值结果表明，本文提出的算法比实际生产中采用的方法能大幅度的改进堆垛质量，减少翻倒箱操作。本文还针对集装箱集港完成后，装船操作前的贝位内预倒箱问题进行了研究，提出了一套基于规则的启发式算法。该算法规则简单，操作方便。数值结果表明，该算法对解质量的改进显著。更多还原

【Abstract】 Container shipping is an important part of the international trade.The increasing ofthe container throughputpose a challenge to the terminal operator in providing efficientand effective service. Therefore, it is needed to optimize the workflow and schedulingof the container operations to enhance the competitiveness of container terminals.Container terminal operations include a variety of decision-makings. In this thesis,we restricted our attention to the location assignment within a yard bay for outboundcontainers and the container pre-marshalling problem.This thesisimproves a dynamic programming model to assign yard location toarriving containers by adjusting the probability of the remaining containers which havenot yet arrived at the terminal, so that the improved model is more consistent with thepractical setting. The numerical experiments show that the improved model provides abetter stacking strategy which cansignificantly reduce the number of theexpectedrelocation movements.In order to solve the large-scale problem, a two-stage heuristic algorithm isproposed. In the first stage, based on the neighborhood search, a heuristic is developedto generate the precedence of the stacking patternsfor each container weight group. Inthe second stage, a heuristic algorithm based on rolling strategy is proposed, so that abetter stacking location can be chosen to reduce the number of the expected relocationmovements. The performance of the two-stage heuristic algorithm is evaluated bycomparing itwith the method used in practice. The numerical experiments show thattheproposed algorithm can significantlyreduce the number of the expected relocationmovements.As to the pre-marshalling problem, a rule-based heuristic is proposed. Thenumerical experiments show that the proposed algorithm is competitive.更多还原