A data-driven decision support system for ship energy efficiency using MIMO artificial neural networks

This paper introduces a decision support system aimed at improving ship energy efficiency by integrating an engine optimisation model with a multiple-input multiple-output artificial neural network (MIMO ANN). Real-time ship and engine performance data are continuously collected during a specific navigation route, while an engine optimisation model is developed using Ricardo Wave software and validated with high accuracy. The model generates ship-specific parameters, and predictions are made for fuel consumption, thermodynamics, heat transfer properties, combustion behaviour, valve timing, and exhaust emissions using the ANN model. After evaluating various configurations, the optimal MIMO ANN model is identified, featuring 30 hidden layers. The reported R² of 1.0 and RMSE of 1.57 indicate a high level of fit to the training data, relative to the target variable range. To assess the generalisation capability and mitigate overfitting, a 10-fold cross-validation procedure is conducted, confirming consistent predictive performance across folds, enhancing the reliability of the model for real-world applications. The proposed approach provides a cost-effective solution for shipping companies to monitor engine parameters in real-time, eliminating the need for expensive sensors. By leveraging operational data rather than relying solely on manufacturer specifications, this study presents a novel and practical framework for ship energy management.