Energy Management Strategy of Fuel Cell Electric Vehicles Using Learning-Based EMPC

A reliable energy management strategy (EMS) incorporating vehicle connectivity technology is of great importance for the performance enhancement of fuel cell electric vehicles (FCEVs). Cloud computing and learning-based methods create new opportunities and challenges for EMS by offering powerful computation and storage capabilities. To fully utilize connected driving information and maintain closed-loop economic performance, while reducing the computational burden, this paper proposes a novel cloud-assisted economic model predictive control with learning-based terminal cost (EMPC-LTC), which is tailored to the EMS of FCEVs. The economic model predictive control (EMPC) optimizations are performed both on the cloud and on the local processor. On the cloud side, a long-horizon EMPC with future driving condition is constructed. Then, the cloud can support a learning process of a parametric terminal cost acting as a good approximation of the cost-to-go performance of the long-horizon EMPC. On the local agent side, the onboard controller runs a short-horizon EMPC with the learning-based terminal cost obtained from the cloud, which can reduce online computing. The effectiveness of this EMPC-LTC is verified through the probabilistic validation technique, considering the errors caused by transmission and learning. An application of this proposed controller on the EMSs of FCEVs demonstrates its effectiveness.