A Two-Step Energy Management Optimization Model for Hydrogen Powered Ship with Load Forecasting

The inherent volatility and uncertainty of power demand in hydrogen-powered vessels pose significant challenges to real-time energy management. This study proposes a two-stage energy management optimization model (TS-EMOM) for hydrogen-powered ships, incorporating multi-step load forecasting to achieve stepwise optimization of model predictive control (MPC). In the first stage, TS-EMOM prioritizes high-efficiency operation of the proton exchange membrane fuel cell (PEMFC) while guiding the battery state of charge (SOC) toward its reference value. The preliminary optimization results then serve as reference inputs for the second-stage optimization, which minimizes the combined costs of energy consumption and equipment degradation. A novel multi-step load forecasting framework based on time-varying parameter vector autoregression (TVP-VAR) is developed to support power allocation optimization. The proposed method is validated using operational data from an actual hydrogen-powered vessel. Comparative results demonstrate that TS-EMOM outperforms conventional approaches in real-time power allocation, achieving both reduced comprehensive operating costs and extended service life of power supply equipment.