Prediction of Ship Operation Time at Bulk Cargo Terminals Using Stacking Ensemble Learning

Ship operation time is a crucial factor in developing berth plans. While most existing research focuses on container terminals, few scholars have examined bulk cargo terminals due to the unique nature of their cargo and complex operational processes. This work proposes a berth characteristic classification-based prediction method (BCCPM) to predict the operation time of ships in bulk cargo terminals. Firstly, recognizing both the similarities and differences among berths in bulk cargo terminals, this paper introduces a berth clustering method based on K-means to group berths with similar operational traits. Then, to conquer the limitations of individual machine learning models in prediction, a stacking ensemble learning approach is proposed for predicting ship operation time in various types of berths. Experiments are conducted on the real operation data from Qingdao dry bulk cargo terminal, China. The results show that outperforms single machine learning models in terms of prediction accuracy and generalization. Moreover, BCCPM effectively captures operational nuances of different berths, resulting in a 2-hour reduction in MAE and a 6% decrease in MAPE compared to the overall prediction method.