Physics-guided Neural Network-based Shaft Power Prediction for Vessels

TL;DR

<3-5 sentence high-level summary>This work tackles shaft power prediction for vessels to support fuel-efficiency and decarbonization in maritime operations. It introduces a physics-guided neural network (PGNN) that integrates empirical resistance formulas for calm water, wind, and waves into a loss function, and leverages a polynomial RPM predictor as a feature. Across four similar-sized cargo vessels and challenging sea conditions, PGNN achieves lower MAE, RMSE, and MAPE than both empirical-formula baselines and a baseline neural predictor, demonstrating improved physical consistency and predictive accuracy. The approach offers a practical pathway to more accurate voyage optimization, while highlighting avenues for future enhancements such as explicitly modeling fouling in the physics loss.

Abstract

Optimizing maritime operations, particularly fuel consumption for vessels, is crucial, considering its significant share in global trade. As fuel consumption is closely related to the shaft power of a vessel, predicting shaft power accurately is a crucial problem that requires careful consideration to minimize costs and emissions. Traditional approaches, which incorporate empirical formulas, often struggle to model dynamic conditions, such as sea conditions or fouling on vessels. In this paper, we present a hybrid, physics-guided neural network-based approach that utilizes empirical formulas within the network to combine the advantages of both neural networks and traditional techniques. We evaluate the presented method using data obtained from four similar-sized cargo vessels and compare the results with those of a baseline neural network and a traditional approach that employs empirical formulas. The experimental results demonstrate that the physics-guided neural network approach achieves lower mean absolute error, root mean square error, and mean absolute percentage error for all tested vessels compared to both the empirical formula-based method and the base neural network.

Figures (3)

• Figure 1: The presented two-stage shaft power prediction model.
• Figure 2: Physics-guided Neural Network Pipeline for Shaft Power Prediction.
• Figure 3: Predicted vs. Actual Shaft Power for All Vessels in the Test Set for PGNN.