The accuracy and interpretability of ship energy consumption prediction play a critical role in energy conservation and emission reduction in the shipping industry. Traditional methods often suffer from issues of low accuracy or acceptability. To address this, an explainable machine learning based ship energy consumption prediction model is proposed. Firstly, sensors were used to collect data related to ship operational energy consumption. Secondly, four different machine learning methods (multiple linear regression (ML), artificial neural networks (ANN), support vector machines (SVM), and extremely randomized trees (ET)) were employed to model ship energy consumption. Finally, the best model was combined with the SHapley Additive exPlanations (SHAP) to perform interpretability analysis on the prediction results. Using an ocean-going container ship as an example, experimental results showed that the ET model improved RMSE and MAPE by 38.08%-69.85% and 47.13%-75.79% compared to other machine learning models. SHAP-based global and local interpretability analysis revealed that ship GPS speed, mean draft, and trim significantly impact energy consumption, with GPS speed being the key factor. The proposed explainable machine learning method not only ensures high accuracy in energy consumption prediction but also provides interpretability, supporting decision-makers in optimizing ship operations and achieving green shipping goals.


    Access

    Check access

    Check availability in my library

    Order at Subito €


    Export, share and cite



    Title :

    Data-Driven Interpretable Machine Learning Methods for the Prediction of Ship Energy Consumption


    Additional title:

    Lect. Notes Electrical Eng.


    Contributors:
    Jia, Limin (editor) / Ou, Dongxiu (editor) / Liu, Hui (editor) / Zong, Fang (editor) / Wang, Pangwei (editor) / Zhang, Mingfang (editor) / Hu, Zhihui (author) / Fan, Ailong (author) / Li, Jiale (author) / Lin, Zhiheng (author)

    Conference:

    International Conference on Artificial Intelligence and Autonomous Transportation ; 2024 ; Beijing, China December 06, 2024 - December 08, 2024



    Publication date :

    2025-03-31


    Size :

    9 pages





    Type of media :

    Article/Chapter (Book)


    Type of material :

    Electronic Resource


    Language :

    English





    Ship energy consumption prediction method based on data driving

    WANG LIPENG / ZHANG ZHI / ZHU QIDAN et al. | European Patent Office | 2021

    Free access


    Enhancing Power Net Efficiency with Data-Driven Consumption Prediction - A Machine Learning Approach

    Muller, Julian / Schuchter, Florian / Brauneis, Daniel et al. | IEEE | 2024


    Mitigating Overfitting in Interpretable Machine Learning Using Bayesian Methods

    Nolan Strauss / Hongsup Oh / Geoffrey Bomarito et al. | NTRS