A Real-time MPC-based Energy Management of Hybrid Energy Storage System in Urban Rail Vehicles

Abstract

The most challenges for the hybrid energy storage system made up of the battery and super capacitor (SC) are the reasonable energy management strategy (EMS) and real-time implementation. Therefore, a variable-step multistep prediction MPC-based energy management strategy is proposed in this paper, which minimizes the system energy losses of the whole operating process and ensures the battery current and SC SOC in a suitable range. In addition, the neural networks (NN) are applied in this paper for real-time implementation, which are trained by using MPC optimization results. To do this, the loss models of the battery, SC and DC/DC converter are built and Simulation is carried out in MATLAB/Simulink, which shows that the proposed EMS can keep the SC SOC in a suitable range. At the same time, the proposed online energy management method can achieve excellent results of MPC optimization.