基于机器学习的集装箱船舶运行轨迹与到港时间预测

Research on Container Ship Travel Trajectory and ETA Prediction Based on Machine Learning

随着全球贸易总量的增加,港口运营效率的重要性日益凸显。准确预测海运集装箱船舶的预计到达时间(Estimated Time of Arrival,ETA)可有效提升物流效率。然而,在实践中,大多数海运存在转运和中途停留;同时航行路线也面临各种因素的影响,这使得船舶ETA充满着不确定性与复杂性。误差较大的船舶ETA不仅阻碍了码头其他利益相关者的有效规划和执行,也导致抵达码头的货物类型和数量存在巨大的波动,增加了多式联运风险。基于机器学习方法,结合AIS历史数据和其他已知因素建立人工神经网络模型,对于长距离海运中集装箱船舶的运行轨迹和ETA进行预测。结果表明,所建立的模型能够有效预测船舶的运行轨迹和ETA。

In this paper,based on the machine learning method,in connection with AIS historical data and other known factors,we established an artificial neural network model to predict the travel trajectory and ETA of container ships in long-distance shipping.By comparing the prediction results and the actual data,we showed that the established model could effectively predict the travel trajectory and ETA of long-distance container ships.