An adaptive PMP-based model predictive energy management strategy for fuel cell hybrid railway vehicles

In order to achieve ecological driving without pollutant emissions on non-electrified rail tracks, the focus is on the development of railway vehicles powered by fuel cell and battery systems. A key issue is to reduce hydrogen consumption while maintaining the battery's state of charge (SoC). In that context, this paper proposes a new casual energy management strategy based on Pontryagin's Minimum Principle (PMP) within the framework of Model Predictive Control (MPC). The entire energy management problem is formulated by solving a cost function in the prediction horizon, different from the typical one found in other work using model predictive control to realize energy management. The main advantage of this strategy is that by introducing the PMP's co-state, which is adaptively evaluated using average power estimation and actual SoC, into the cost function, a predicted optimal SoC trajectory is no longer required within the framework of MPC. Therefore, the strategy does not require complete information on the rail track and behaves casually. In addition, the dynamic of the fuel cell power is considered by introducing a tuning factor into the cost function, which benefits the service life of fuel cells. The proposed strategy is tested and validated in a realistic driving cycle by a hardware-in-the-loop (HiL) test bench at the Center for Mobile Propulsion (CMP) of RWTH Aachen University. The simulation results show an optimum close to the offline PMP results and up to 12.1% hydrogen savings compared to a typical MPC strategy using constant SoC reference. The HiL results demonstrate the real-time capability of the proposed strategy and show the excellent fuel economy with merely 2.7% more hydrogen consumption than the global optimum result. (C) 2020 Elsevier B.V. All rights reserved.