In global trade and transportation, the energy consumption and carbon emissions of ships have drawn significant attention. Accurately predicting the fuel consumption of ships is of crucial importance for optimizing voyage plans and reducing operating costs. Different from the traditional method of using long short-term networks for fuel consumption prediction, this paper proposes a model that combines the K-Means clustering algorithm and the Long Short-Term Memory (LSTM) network. Based on the operation data collected from actual ships, this study employs the K-Means clustering algorithm to classify the ship navigation conditions, thereby identifying different navigation states. Subsequently, for the five clustered navigation conditions, a Long Short-Term Memory (LSTM) network can be constructed and trained to predict fuel consumption. This study demonstrates that the K-Means-LSTM parallel network can achieve the best results in terms of the coefficient of determination ($R^\wedge {2}$) and the root mean square error (RMSE), outperforming the prediction effect of a single LSTM neural network.
The Ship Fuel Consumption Prediction Based on Working Conditions Classification
16.05.2025
5364904 byte
Aufsatz (Konferenz)
Elektronische Ressource
Englisch
How working conditions affects fuel consumption
Engineering Index Backfile | 1934
Ship fuel consumption prediction method based on distributed energy efficiency calculation
Europäisches Patentamt | 2024
|Predicting Ship Fuel Consumption: Update
NTIS | 1996
|