Adaptive model prediction based real-time energy management for marine fuel cell hybrid power systems

Objectives To address the model mismatch problem caused by time-varying battery parameters under complex variable load conditions in ship hybrid power systems, which consequently degrades control accuracy and energy efficiency, an adaptive model predictive control (AMPC) energy management strategy based on a linear parameter-varying (LPV) model is proposed. Methods First, based on a second-order RC equivalent circuit model of the lithium-ion battery, the variation characteristics of model parameters with the state of charge (SOC) are analyzed to construct an LPV prediction model capable of capturing the real-time dynamic characteristics of the system. A multi-objective optimization function is formulated to simultaneously consider hydrogen consumption, battery energy loss, and fuel cell current smoothness. The entropy weight method is adopted to determine weighting coefficients. The nonlinear constrained optimization problem is then transformed into a quadratic programming (QP) problem for online solution. Comparative simulations are conducted under typical ship operating conditions. Results Simulation results show that, compared with the linear time-invariant model predictive control (LTI-MPC) and rule-based strategies, the proposed strategy mitigates model mismatch effects. Over the entire operating cycle, the fuel cell current change rate is limited to 25.80 A/s, representing a 34.5% reduction compared with the rule-based strategy, while the total system energy consumption is reduced by 24.1%. Conclusions The proposed strategy achieves coordinated optimization of economic performance and durability for ship hybrid power systems while maintaining computational real-time feasibility.