A Machine Learning Approach to Predicting Estimated Time of Arrival on the Inland Waterways

Predicting the Estimated Time of Arrival (ETA) during waterway transits is a critical capability within maritime commerce that provides maritime stakeholders improved abilities to optimize vessel utilization, improve supply chain logistics, and enhance economic efficiencies. Currently, the ability to predict Estimated Time of Arrival is less than 50% accurate, with mariners reluctantly accepting the eventual arrival times that result from unforeseen or unacknowledged factors. The objective of this research was to determine the causes for ETA delays and introduce a machine learning approach to improving ETA predictions. Specifically, the research evaluated the effects of three categories of factors that contribute to ETA delays: maritime traffic conditions, navigational anomalies (obstructions or hazards), as well as environmental conditions. Maritime traffic conditions include traffic density on AIS-equipped vessels for both upstream and downstream transits. Potential obstructions include the presence of bridges, locks, and both natural and man-made hazards to navigation. Environmental factors include elements such as weather, specifically temperature, river water level, and precipitation. Data analysis was performed using two years of data to determine the feature sets with greatest influence on ETA prediction. Data were preprocessed for use in machine learning models that use an ensemble of daily averaging and regression techniques to determine the effects on speed of transit. The new predictive model was tested on new data and achieved at least 87% accuracy in predicting expected speeds, as well as 71%, 78%, and 70% accuracy in predicting speeds for weather effects including minimum temperature, high water level, and precipitation, respectively.

Author

• Salindong, Arthur R

Language

• en

Keyword

• estimated time of arrival
• vessel transit speed
• machine learning
• predictive analytics
• inland waterways
• maritime

Date created

• 2022

Type of Work

• Dissertation

License

• All rights reserved

Rights statement

• In Copyright

GW Unit

• Engineering Management

Degree

• D.Engr.

Advisor

• Sarkani, Shahryar
• Fossaceca, John M

Committee Member(s)

• Blackford, Joseph P
• Fossaceca, John M
• Sarkani, Shahryar

Persistent URL

•  https://scholarspace.library.gwu.edu/etd/1544bp94g