Vessel arrival time to port prediction via a stacked ensemble approach: Fusing port call records and AIS data

Zhong Chu, Ran Yan*, Shuaian Wang

*Corresponding author for this work

Research output: Contribution to journalArticlepeer-review

Abstract

Accurate prediction of vessel arrival time (VAT) to port is essential for optimizing port operations, particularly given the common discrepancies between the vessel-reported estimated time of arrival (ETA) and its actual time of arrival (ATA). Traditional VAT prediction models predominantly rely on either static port call data (e.g., ETA and ATA) or dynamic automatic identification system (AIS) data, with limited integration of both sources to comprehensively address forecasting needs and biased forecasting results. To address these limitations, this study introduces a framework that, for the first time, integrates static port call data with dynamic vessel AIS data using a time-based comparative interpolation method to enhance VAT prediction accuracy. By synchronizing scheduled operations with real-time vessel movements, our approach captures nuanced temporal variations, significantly enhancing VAT prediction accuracy. Based on a tree-based stacking model and real-world vessel arrival data from Hong Kong Port (HKP), the proposed framework leverages the strengths of tree-based methods in handling tabular data and demonstrates substantial improvements in VAT prediction accuracy. Our results show an 54.53% reduction in mean absolute error (MAE) (from 6.84 to 3.11 h) and an 50.14% reduction in root mean squared error (RMSE) (from 10.61 to 5.29 h) compared to vessel-reported ETAs. Key features such as vessel-reported ETA, vessel sailing speed, vessel physical features, and spatiotemporal AIS data contribute to these improvements. This research addresses a critical gap by providing a unified approach that leverages both static and dynamic data sources, offering port authorities a more reliable and robust tool for vessel arrival forecasting and the subsequent informed port resource planning.

Original languageEnglish
Article number105128
JournalTransportation Research Part C: Emerging Technologies
Volume176
DOIs
Publication statusPublished - Jul 2025
Externally publishedYes

Bibliographical note

Publisher Copyright:
© 2025 Elsevier Ltd

ASJC Scopus Subject Areas

  • Civil and Structural Engineering
  • Automotive Engineering
  • Transportation
  • Management Science and Operations Research

Keywords

  • AIS data
  • Machine learning
  • Maritime transportation
  • Port operations
  • Vessel arrival time prediction
  • Vessel trajectory analysis

Cite this