随着全球经济的持续增长，海运作为国际贸易的主要运输方式，需求持续攀升。港口泊位的有限容量和航道的固定通航能力无法满足不断增长的船队运载需求，船舶滞港和港口拥堵问题持续加重。面对港口服务效率的挑战和拥堵问题的加剧，港口运作管理领域迫切需要通过科学的方法来提升泊位和航道服务效率。为此，针对当前港口运营面临的具体问题，本文研究考虑航道受限的船舶航道交通和泊位调度问题，围绕港口航道资源对泊位、内锚地和岸桥资源集成调度，旨在通过优化资源调度解决方案，助力提升港口整体运营效能。首先，本文研究了一个考虑通用航道布局的船舶交通调度和泊位分配问题，以最大化资源利用效率和船舶服务水平。文章特别关注通用港口航道布局，其中航道宽度因段而异，从而导致航道中复杂多变的船舶交通。本文首先提出了整数规划模型，综合考虑船舶通行潮汐窗口、内锚地有限容量、航道分段宽度以及航道行驶安全间隔等。为求解该复杂问题，文章提出了一种机器学习增强的列生成算法，其中子问题旨在为船舶生成港口服务计划。为便于求解子问题，本文又将其构建成在时空网络的最短路径问题，并使用机器学习方法缩减网络规模。基于上海港实际数据，文章进行了数值实验来评估所提出模型和方法的有效性和优越性。其次，为应对机器故障和天气变化等突发事件对船舶到港影响，本文研究了船舶到达时间不确定的航道交通调度和泊位分配问题。航道受限上重点考虑了船舶的吃水深度叠加航道潮汐因素的可通行潮汐时间窗、船舶在航道行驶安全距离等。文章使用机会约束方法考虑船舶到港时间不确定，并建立了分布鲁棒机会约束模型，然后使用切比雪夫不等式将其转换为可求解的线性模型。文章使用机器学习增强的列生成求解该模型，并设计了启发式算法生成初始可行列。数值实验验证了算法的有效性，以及模型与确定性模型、随机模型相比的保守度和鲁棒性。最后，在考虑船舶交通调度和泊位分配基础上，本文研究了一个综合考虑航道潮汐影响以及港池内锚地区域有限容纳能力的船舶交通调度、泊位和岸桥分配问题。文章提出了一个混合整数规划模型，并基于所提模型在Dantzig-Wolfe分解框架下设计了一种分支定价切割算法，同时使用两种有效不等式进行增强。文章使用天津港的实际数据，进行了数值实验，验证了提出算法的优越性。同时，进行了航道潮汐影响和港池内锚地区域容纳能力的定量分析，以获取管理学启示。

With the continuous growth of the global economy, seaborne transport, as a main mode of international trade, has seen a rapid increase. The limited handling capacity of port berths and navigation channels cannot meet the ever-growing needs of the shipping fleet, leading to persistent issues of vessel delays and port congestion. Faced with challenges in port service efficiency and escalating congestion, there is an urgent need in the field of port operations management to enhance the efficiency of berth and channel services. Therefore, this paper tries to address those particular challenges related to the intricacies of vessel traffic scheduling and berth allocation considering the limited channel capacities, with focus on the integrated scheduling of port resources—including navigation channels, berths, inner anchorages, and quay cranes—to enhance operational efficiency of ports.Firstly, this paper investigates an integrated vessel traffic scheduling and berth allocation problem considering a generic channel layout, optimizing both the traffic plan of vessels entering and leaving the port and their berthing plans to maximize resource efficiency and service levels for vessels. Special attention is given to the generic port channel layout, where channel may divide into several segments with different widths, leading to complex traffic rules within the channel. An integer programming model is proposed, taking into account various factors including tidal windows for channel passage, limited capacity at inner anchorages, segmented channel widths, and safety intervals for vessels in the channel. To solve this complex model, a machine learning-enhanced column generation (MLCG) algorithm is introduced. The MLCG algorithm decomposes the model into a master problem and multiple vessel-dependent pricing subproblems, with each aiming to generate a service plan for a vessel. To facilitate the solution of the pricing subproblems, a time-space network is constructed so that they can be formulated as a shortest path problem. Machine learning methods are adopted to reduce the size of the time-space network to facilitate the computation efficiency. Numerical experiments based on real data from the Port of Shanghai are conducted to evaluate the effectiveness and practicality of the proposed solution method.Secondly, to manage the effects of unpredictable elements such as harsh weather conditions and facilities breakdown on vessel arrival times, this paper explores the vessel traffic scheduling and berth allocation problem under uncertain conditions, paying attention to the tidal windows that may limit the vessel passage due to the required draft depths and the safe distances of passing vessels. A chance-constrained approach is used to account for the uncertainty of vessel arrival times, and a distributionally robust chance-constrained model is established, which is then converted into an integer programming model using Chebyshev‘s inequality. The model is solved using a machine learning-enhanced column generation approach, and heuristic algorithms are designed to generate initial feasible columns. Numerical experiments demonstrate the effectiveness of the proposed algorithm, as well as the conservatism and robustness of the proposed model compared to deterministic and stochastic models.Lastly, besides vessel traffic scheduling and berth allocation, this paper moves further hinterland and investigates a comprehensive problem that considers the tidal impact on channel passage, and the limited capacity of inner anchorages to address the challenges of vessel traffic scheduling, berth allocation, and quay crane assignment. A mixed-integer model is proposed, and a branch-and-price-and-cut (B&P&C) algorithm is developed within the Dantzig-Wolfe decomposition framework, enhanced by two effective inequalities. Numerical experiments using real data from the Port of Tianjin confirm the superiority of the proposed B&P&C algorithm. Quantitative analyses of the tidal impacts and the capacity of inner anchorages within the port basin are conducted to derive management insights.