Maritime decarbonization through machine learning: A critical systematic review of fuel and power prediction models

Abstract

A vital component of decarbonization and operational optimization in the maritime industry is predicting ship propulsion power requirements and fuel consumption rates. This study systematically and critically reviews the application of machine learning (ML) in fuel and power estimation and prediction (FEP) in the last decade (2013–2024) regarding the two cores of ML models, including aspects of data and the applied learning algorithms. This study revealed the urgent need of the field in data-centricity and standardization of model performance benchmarking that covers more than just accuracy. Research directions were recommended, focusing on reliable and applicable FEP, objective-specific development, and model trustworthiness and maintenance policies. This paper advocates a practical application of ML and other AI applications in real-world settings to support their certifiability and the development of related policies and regulations, thus enhancing the transition toward robust data-driven decarbonization and operational efficiency.