Endurance-Oriented Model Predictive Energy Management for a Proton Exchange Membrane Fuel Cell–Battery Hybrid Quadcopter Under Dynamic Mission Conditions

Proton exchange membrane fuel cell–battery hybrid power systems provide an effective solution to overcome the limited endurance of battery-powered multirotor unmanned aerial vehicles. However, the highly transient power demands of quadcopter platforms, combined with balance-of-plant losses and operational constraints, create significant challenges for reliable energy management. This study proposes a degradation-aware stress-mitigation model predictive control-based energy management framework to maximize mission endurance under realistic conditions. A control-oriented, physics-consistent model is developed using manufacturer polarization data from a 500 W Aerostak proton exchange membrane fuel cell. The model captures polarization behavior, balance-of-plant loads, battery dynamics, and direct current-bus power balance. The model predictive control strategy optimally allocates power by maintaining direct current-bus stability, regulating battery state-of-charge within safe limits, and constraining fuel cell power ramp rates to mitigate degradation. High-fidelity simulations are conducted under stochastic wind disturbances and mission-dependent load profiles, including takeoff, climb, cruise, and maneuvering phases. The results show continuous power delivery without unmet load demand. The hybrid system achieves a flight endurance of 220–224 min, consuming a total of 89.99 g of hydrogen at an average rate of 0.398–0.412 g/min, indicating a notable reduction under the considered operating conditions. Additionally, long-term analysis indicates that over 97% of initial endurance is preserved after 100 cycles, demonstrating robustness against fuel cell aging. An analytical real-time feasibility assessment further indicates that the control-oriented formulation is compatible with the computational resources of typical unmanned aerial vehicle-class onboard processors, while the integration of adaptive and robust predictive control techniques is identified as a direction for future work.