The integrated berth allocation and quay crane assignment problem is an important issue for the operations management in container terminals. This issue primarily considers the assignment of berthing time, position, and the number of quay cranes in each time segment to ships that must be discharged and loaded at terminals. This study examines such a problem by considering uncertainties in the late arrival of ships and inflation of container quantity. Based on historical data, we first divide the uncertainty set into $K$ non-overlapping full-dimensional clusters via $K$-means clustering, and the weight of each cluster is calculated. Then, we formulate an almost robust model by introducing the weighted max penalty function with the objective of minimizing the total cost, which is caused by the deviations from the expected berthing location and departure time. The concept of robustness index is introduced to investigate the trade-off between the changes in the objective value and the penalty violation. A decomposition method, which contains a deterministic master problem and a stochastic subproblem, is proposed to solve the problem. In each iteration, the subproblem checks the master problem under different realizations, adds scenarios, and cuts into the master problem if needed. Numerical experiments demonstrate that (i) the proposed method can solve the model efficiently, (ii) the robustness index shows that a significant improvement in objective can be achieved at the expense of a small amount of penalty, (iii) the proposed model can handle uncertainties better than the deterministic, fully robust, and worst-case models in terms of total expected cost, total vessel delays, and utilization rates of the berth and quay crane, and (iv) the proposed method becomes more attractive compared with the first come first served approach as the congestion situation or uncertainty degree increase.

泊位分配与码头起重机分配的集成问题是集装箱码头运营管理的重要问题。本期主要考虑到必须在码头卸货和装船的船舶在每个时间段的停泊时间，位置和码头起重机的数量。这项研究通过考虑船舶迟到和集装箱数量膨胀的不确定性来研究这一问题。根据历史数据，我们首先将不确定性集分为$ķ$ 通过以下方式不重叠的全维聚类 $ķ$-表示聚类，并计算每个聚类的权重。然后，我们引入加权最大惩罚函数，以使总成本最小化为目标，建立了一个几乎健壮的模型，该总成本是由与预期停泊位置和出发时间的偏差引起的。引入鲁棒性指数的概念来研究目标值的变化与罚分违规之间的权衡。提出了一种分解方法，该方法包含确定性的主问题和随机子问题，以解决该问题。在每次迭代中，子问题都会以不同的实现检查主要问题，添加方案，并在需要时切入主要问题。数值实验表明：（i）所提出的方法可以有效地求解模型，