【作者】 于敏；

【导师】 王诺；

【作者基本信息】 大连海事大学 ， 交通工程， 2010， 硕士

【摘要】 泊位作为集装箱码头的稀缺资源,其调度工作质量的高低直接影响码头装卸作业的效率,影响集装箱班轮的船期和港口企业的声誉,同时,也体现着码头的管理水平和综合竞争实力。实践中,泊位分配的同时也进行相应作业岸桥数目的安排,即码头调度人员根据一定的优化策略及泊位、岸桥是否空闲等约束条件,为一系列到港船舶安排靠泊顺序、靠泊泊位及确定作业岸桥数目。与单纯的泊位分配和岸桥分配相比,泊位-岸桥分配把两者当成一个整体,能有效反映其相互影响相互制约的关系,更符合码头生产组织的实际情况。集装箱码头泊位、岸桥分配问题是典型的NP难题,一直以来很难用精确算法求解。目前,大多数研究以缩短船舶在港时间或者码头生产成本最低为目标,采用启发式算法或全局搜索进化算法进行求解。但这些研究倾向于单目标求解算法或者将多目标转化为单目标进行求解,而集装箱码头泊位优化分配本应对多个目标同时进行优化,这对算法实施的结果影响较大。因此,在多目标模型求解过程中显然需要加以考虑。考虑到泊位-岸桥多目标优化分配模型求解的复杂性,本文在前人研究的基础上,提出了一种改进多目标遗传算法,用以求解集装箱码头泊位-岸桥优化分配问题。模型部分,从船舶服务需求和码头综合效益两方面出发,主要选取以船舶平均在港时间和码头生产成本为目标的数学模型。模型求解采用改进多目标遗传算法,首先采用染色体组的方式表示个体即模型的可行解,为保证Pareto最优解集的分散性和均匀性,采用随机方法生成初始种群；其次,给出了多个约束条件下的交叉算子运算规则,同时引入岸桥分配启发式算法确定模型从属变量,以求得个体的目标函数值,并应用Pareto分级方法和个体拥挤距离计算进行适应度值评价；最后,针对算法求出的Pareto最优解集,给出了一种泊位-岸桥分配最终实施方案的选择策略。经过算例表明,与传统单目标优化相比,本文提出的优化方法能获得使码头综合效益较大的满意解,同时也证明了该优化模型和算法的有效性。更多还原

【Abstract】 As rare resources of container terminals, the berths’scheduling levels directly affect the efficiency of cargo-handling in terminal, sailing schedule of container liner and reputation of port enterprise, and reflect management level and comprehensive competitive power of terminal, In practice, berths scheduling together with the allocation of quay cranes, that is, according to a certain optimizing tactics and some constraints such as the berth, quay cranes are free or not, the terminal scheduling worker arranges berthing order, berth and the quality of loading-uploading quay cranes for the container vessels arrived. Comparing with the separated berth allocation or quay crane allocation, the berth-quay crane allocation makes the two as a whole system, which can efficiently reflects the relationship of each influence and each restriction, it corresponds with the actual terminal production organization better.The berth-quay crane allocation problem of container terminal is typical NP-hard problem, which is hardly solved by exact algorithm. At moment, most researches make the reduction of vessels’service time or the terminal’s production cost as the goals, then sovle the problem by heuristic algorithm or global search evolutionary algorithm. But this kind of researches prefer to the single optimization or changing MOP into single optimization, the berth allocation optimization should solve the goals together, which has confluences on results of the algorithm. So it is apparent to consider the factors in the berth scheduling optimization.Considering the complexity of berth-quay crane allocation MOP’s solution, the multi-objective genetic algorithm is adopted in this essay, and improved based on the study of predecessors, in order to solve the berth-quay crane allocation problem of container terminals. In model, starting from vessels’service demand and container terminal’s comprehensive benefits, this paper chooses a multi-objective optimization model, simultaneously considering the stay time of vessels and the production cost as optimization objectives. To solve the model, improved multi-objective genetic algorithm is adopted. Firstly, the individuals, which are the feasible solutions of model, are expressed by genome. In order to ensure the dispersion and uniformity of Pareto optimum set, the initial population is produced by the way of random method. Secondly, a cross rule is given to deal with the restriction conditions as well as model’s dependent variables are determined by a quay crane heuristic, to calculate the individuals values, and the individuals’fitness is evaluated by the Pareto ranking method. Further more, a strategy is suggested to choose a valid berth-quay crane allocation schedule from the Pareto optimum solutions. Finally, experiments are given to verify the efficiency of the model and method, while the results show that a satisfied solution can be obtained by the approach proposed in this thesis.更多还原