A Machine Learning Approach to Modeling Small Vessel Performance

Publication Details

This paper appeared at the 2024 OCEANS conference in Halifax, Nova Scotia.

Paper Link: https://ieeexplore.ieee.org/abstract/document/10754239/

Summary

This paper an LSTM based method for predicting fuel consumption aboard small fishing vessels. Our paper provides an overview of our technique, including data collection, pre-processing, model training and evaluation techniques. This work provides a solid starting point for continuing to develop Glas Ocean Electric’s ML platform with the goal of increasing sustainibility in the maratime sector.

Authors

Matthew Peachey, Karansingh Chauhan, Juliano Franz, Sue Molloy, Christopher Whidden

Abstract

The maritime industry is a critical component of global trade. However, it is also responsible for a significant amount of Greenhouse Gas emissions. While small vessels are often overlooked as main contributors to emissions, when analyzed as a group, they collectively have a meaningful impact. In order to help promote more sustainable operation of these vessels, it is important to to model fuel consumption based on different weather and sea conditions. To this end, we present a machine learning approach to modelling fuel consumption aboard small vessels using an LSTM architecture. In this paper, we discuss the process of collecting data aboard one such small vessel, the end-to-end model training & evaluation process, and discuss plans for future work.