To reduce anthropogenic greenhouse gas emissions, novel and clean propulsion technologies are being developed in the maritime sector. Hybrid powertrains consisting of batteries and fuel cells represent a good alternative to conventional propulsion systems due to their potentially high energy and power density and low emissions. By employing prediction algorithms, which forecast upcoming load demands and effectively transfer power amongst the propulsion components, intelligent energy management may enhance fuel efficiency even further. A load-prediction engine based on machine learning is proposed to improve the fuel efficiency of a cruise ship’s hybrid propulsion system consisting of a high-voltage lithium-ion battery and a high-temperature solid oxide fuel cell (SOFC).
To achieve an improved fuel economy with prediction parameters, different machine learning algorithms were used in a training phase and implemented in the energy management system to predict load patterns online. The prediction parameters were implemented within the energy management system to achieve a more efficient power distribution between the propulsion components. In total five prediction algorithms were compared to each other. It was found out that a gradual horizontal shift in the training and testing phase gave rise to high RMSE. By using the prediction algorithms, the fuel costs of a cruise could be reduced by up to 5.6 %.