Autonomous Fixed-Time Model Predictive Control for Decentralized Coordination and Stabilization of Hybrid Fuel Cell/Battery/Supercapacitor Systems

For the fuel cell/battery/supercapacitor hybrid energy storage system (HESS) with constant power loads (CPLs), its controller needs to meet desired control goals, including decentralized dynamic power sharing, fast voltage regulation, large signal stabilization, reliability, and scalability. To this end, this paper proposes an autonomous fixed-time model predictive control (FxTMPC), which aims to achieve the above objectives with less system prior knowledge and sampling cost. First, the fixed-time disturbance observer (FxTDO) is constructed to estimate and feed-forward compensate for unknown and time-varying source-load side information, which not only reduces controller design complexity but also significantly improves robustness against disturbances. Second, the receding horizon optimization problem is solved offline based on source-load estimation variables, and FxTMPC is obtained by embedding fixed-time algorithms into the framework to correct the reference trajectory. Then, a decentralized hybrid droop control (DHDC) based on FxTMPC is developed, which achieves virtual impedance/resistance/capacitance at output terminals by modifying voltage reference values, not only enables different types of sources to respond to low/mid/high-frequency components of the load respectively, but also to reduce sensor usage with the assistance of FxTDO, thus achieving above the desired objectives with low cost. Finally, simulation and experimental results under multiple scenarios demonstrate that the proposed control strategy achieves HESS objectives with less prior knowledge and sensors, while comparative studies confirm its superiority.