In tramp shipping, a preliminary route is required for voyage planning at the pre-fixture stage (before a chartering contract is agreed). Such routes are conventionally designated by using pilot charts or software considering long-term statistical weather. However, it has been experienced by tramp operators that such route solutions often poorly estimated sailing distances for long journeys and thereby cause inappropriate cost estimation and bad voyage plan. To fill this gap, a data-driven methodology is proposed in this paper to establish a practical route library with the consideration of ship sizes, load conditions and seasonality. In this method, it first requires a dividing of ship trajectories into local sea passage and open sea passage. The voyage trajectories made of AIS points are then simplified to pattern nodes based on a speed-weighted geolocation method. Afterwards, the KMeans algorithm is deployed to properly classify these pattern nodes, identifying the most representative nodes (routes) in open sea passages. Simultaneously, the connection points are identified by DBSCAN algorithm, representing local sea passages. Combining the representative routes in open sea passages and the connection points in local sea passages, the most navigated routes between two ports are obtained. Finally, case studies are conducted for the Pacific Ocean and the Atlantic Ocean respectively using global AIS data from tanker vessels to demonstrate the feasibility and effectiveness of this methodology. The proposed route library is capable of providing reliable route references to support the decision-making at the pre-fixture stage.

在杂货运输中，需要在装船前阶段（在商定租船合同之前）进行航次规划的初步路线。考虑到长期统计天气，这些路线通常通过使用导航图或软件来指定。然而，流浪汉运营商的经验是，此类航线解决方案往往无法准确估算长途旅行的航行距离，从而导致成本估算不当和航程计划不佳。为了填补这一空白，本文提出了一种数据驱动的方法，以建立一个考虑船舶尺寸、负载条件和季节性的实用航线库。在该方法中，首先需要将船舶轨迹划分为局部海上通道和公海通道。然后，由 AIS 点组成的航行轨迹被简化为基于速度加权地理定位方法的模式节点。然后，部署 KMeans 算法对这些模式节点进行正确分类，确定公海通道中最具代表性的节点（路线）。同时通过DBSCAN算法识别连接点，代表当地的海上通道。结合公海航道的代表性航线和局部航道的连接点，得到两个港口之间通航最多的航线。最后，分别使用来自油轮的全球 AIS 数据对太平洋和大西洋进行了案例研究，以证明该方法的可行性和有效性。