With the recent tightening of greenhouse gas (GHG) reduction targets by the International Maritime Organization (IMO), there is a growing need for data-driven energy management and operation planning for ships. Electric propulsion hybrid ships, in particular, can operate in multiple propulsion modes depending on the power sharing between generators and batteries, and their fuel consumption strongly depends on sea states and route conditions. This study proposes an integrated, data-driven framework for mode-aware optimal operation of a direct-current (DC)-based serial electric propulsion hybrid ship. In-service operational data from a electric propulsion hybrid demonstration ship were combined with reanalysis-based metocean data to construct a comprehensive dataset that couples propulsion, power-system, and environmental variables. Using this dataset, mode-specific fuel-consumption and fuel-efficiency characteristics were derived, and Light Gradient Boosting Machine (LightGBM) models were trained to predict generator load or fuel efficiency for each propulsion mode as functions of ship speed and metocean conditions. The prediction models were validated against independent voyage records, and the sensitivity of route-level fuel-consumption error to segment length was analyzed to select an appropriate spatial discretization. The validated models were then embedded into a two-stage optimal operation framework. A coastal route was first generated using an A*-based path planner with quadtree and Voronoi representations of coastal geometry. Subsequently, a genetic algorithm (GA) was used to determine the speed and propulsion mode in each route segment so as to minimize total fuel consumption under battery state-of-charge (SOC) and arrival-time constraints. Applications to coastal and offshore routes demonstrated substantial reductions in total fuel consumption relative to actual operations, indicating that the proposed framework can serve as a practical basis for developing energy-efficient, low-emission operational strategies for next-generation electric propulsion hybrid ships. • This study analyzes the operational characteristics of an electric-propulsion hybrid ship and embeds them into route planning. • Mode-dependent fuel-economy characteristics are extracted from in-service operational data and visualized using heatmaps. • A machine-learning-based model is developed and validated to predict generator load and fuel-economy performance under operating conditions. • Operation optimization is conducted in two stages: coastal-feature-aware route generation and mode-aware optimization to minimize fuel consumption along the route. • Case studies for coastal and offshore routes demonstrate meaningful reductions in total fuel consumption through the proposed framework.
Jang et al. (Wed,) studied this question.