Real-vehicle experimental validation of a predictive energy management strategy for fuel cell vehicles

Abstract

Predictive information is highly valuable for energy management strategies (EMSs) of fuel cell vehicles. In particular, long-term predictions can significantly improve the fuel efficiency because they allow for an optimization of the energy management before departure. This potential has been demonstrated in numerous simulation studies. This work extends the literature with an extensive experimental validation of a predictive EMS that exploits route-based long-term predictions in the form of optimized reference trajectories for the battery state of charge. The experimental validation is performed with a real passenger fuel cell vehicle and strongly focuses on the real-world application where random influences such as traffic cause considerable disturbances with respect to the long-term prediction. The validation comprises two stages: First, real driving tests are repeatedly conducted on public roads, analyzing the robustness of the predictive EMS and assessing fuel efficiency gains over a nonpredictive EMS. Second, chassis dynamometer tests are performed where a selected real driving cycle is reproduced to compare the two EMSs directly. The chassis dynamometer tests confirm a significant reduction in the fuel consumption by $\text{6.4}\text{\%}$ compared to the nonpredictive EMS. The experimental results are analyzed quantitatively and qualitatively in detail.