In order to enhance the real-time performance of model predictive control (MPC) based energy management strategy (EMS), this paper proposes a real-time automotive energy management strategy based on backpropagation neural network (BPNN) approximation of MPC controller. Firstly, establish a pure electric vehicle (PEV) simulation model, then design a BP neural network, and the joint simulation model is established to predict the speed, the energy management is finally realized by combining the energy consumption evaluation model. Compared with the EMS based on MPC controller, the energy-saving control effect is marginally diminished, and the real-time performance of vehicle speed track prediction is greatly improved. The proposed EMS can optimize the driver’s speed trajectory and control vehicle movement in real-time, resulting in better comprehensive energy-saving control effects compared to traditional MPC.