Full metadata record
| DC Field | Value | Language |
|---|---|---|
| dc.contributor | Department of Logistics and Maritime Studies | en_US |
| dc.contributor.advisor | Wang, Shuaian (LMS) | en_US |
| dc.contributor.advisor | Liu, Yan (LMS) | en_US |
| dc.creator | Wang, Haoqing | - |
| dc.identifier.uri | https://theses.lib.polyu.edu.hk/handle/200/13782 | - |
| dc.language | English | en_US |
| dc.publisher | Hong Kong Polytechnic University | en_US |
| dc.rights | All rights reserved | en_US |
| dc.title | Data-driven approaches for intelligent ship fuel consumption management : machine learning models, optimization techniques, and domain knowledge | en_US |
| dcterms.abstract | Maritime transport serves as the backbone of the global supply chain. Ship fuel consumption constitutes a significant portion of maritime transport costs, while its emissions pose substantial environmental risks. Predicting and optimizing ship fuel consumption under varying scenarios is a pivotal procedure in enhancing ship fuel efficiency and sustainable maritime transport. This thesis leverages real-world data, machine learning models, optimization techniques, and domain knowledge in the shipping industry to prompt intelligent fuel consumption management. | en_US |
| dcterms.abstract | The first study designs an innovative post hoc algorithm to correct the predictions of fuel consumption obtained by the off-of-shelf machine learning model—classification and regression trees (CART). We call this algorithm “PH-CART”, which serves as a semi-domain knowledge-aware decision tree to combine domain knowledge in maritime transport with CART using a linear optimization model. Based on a real-world dataset, the experiment demonstrates that PH-CART outperforms CART in terms of both accuracy and robustness. Furthermore, the PH-CART model exhibits a high level of interpretability because it generates a univariate function representing the relationship between ship sailing speed and fuel consumption. This function is developed by incorporating all contextual information from other variables. The study contributes to sustainable maritime transport by offering more accurate and robust predictions of ship fuel consumption. Moreover, this study provides a new perspective by applying domain knowledge to an industry-specific issue in the transportation domain. | en_US |
| dcterms.abstract | The second study uses domain knowledge to develop two innovative methods for predicting ship fuel consumption--the first is a physics-informed neural network (PI-NN) model that improves the interpretability of the black-box model while maintaining accuracy and the second is a mixed-integer quadratic optimization (MIQO) model that considers more forms of feature variable expressions in an additive white-box model. The proposed approaches address the tradeoff between model interpretability and model accuracy in ship fuel consumption prediction. The experiment results demonstrate that the PI-NN model improves the interpretability of the black-box model while preserving accuracy. The MIQO model considers alternative variable expressions, leading to the flexibility of the white-box model. Finally, SHapley Additive exPlanations (SHAP) is used to explain how each feature value contributes to the predictions of the black-box model, thereby providing insights into how each value of feature variables affects fuel consumption. This study provides a solution to the tradeoff between model interpretability and model accuracy and can promote the application of data-driven models in ship fuel consumption prediction. Moreover, this study gives implications for the application of explainable machine learning models in practice. | en_US |
| dcterms.abstract | Many shipping companies are unwilling to share their raw data because of data privacy concerns. However, certain problems in the maritime industry become much more solvable or manageable if data are shared. In the third study, we develop a two-stage method based on federated learning (FL) and optimization techniques to predict ship fuel consumption and optimize ship sailing speed. Because FL only requires parameters rather than raw data to be shared during model training, it can achieve both information sharing and data privacy protection. Our experiments show that FL develops a more accurate ship fuel consumption prediction model in the first stage and thus helps obtain the optimal ship sailing speed setting in the second stage. The proposed two-stage method can reduce ship fuel consumption by 2.5%-7.5% compared to models using the initial individual data. Moreover, the proposed FL framework protects the data privacy of shipping companies while facilitating the sharing of information among shipping companies. | en_US |
| dcterms.extent | xiii, 140 pages : color illustrations | en_US |
| dcterms.isPartOf | PolyU Electronic Theses | en_US |
| dcterms.issued | 2025 | en_US |
| dcterms.educationalLevel | Ph.D. | en_US |
| dcterms.educationalLevel | All Doctorate | en_US |
| dcterms.accessRights | open access | en_US |
Copyright Undertaking
As a bona fide Library user, I declare that:
- I will abide by the rules and legal ordinances governing copyright regarding the use of the Database.
- I will use the Database for the purpose of my research or private study only and not for circulation or further reproduction or any other purpose.
- I agree to indemnify and hold the University harmless from and against any loss, damage, cost, liability or expenses arising from copyright infringement or unauthorized usage.
By downloading any item(s) listed above, you acknowledge that you have read and understood the copyright undertaking as stated above, and agree to be bound by all of its terms.
Please use this identifier to cite or link to this item:
https://theses.lib.polyu.edu.hk/handle/200/13782