Variable horizon MPC for energy management on dual planetary gear hybrid electric vehicle

This paper aims at studying the energy management for the dual planetary gear hybrid electric vehicle based on the model predictive control structure with the variable horizon velocity prediction. According the identification of characteristic parameters of historical velocity, the optimal predictive horizon is obtained by considering the energy consumption. Different deep neural networks are trained and applied to predict the future variable horizon velocity through the prediction accuracy. The simulation results show that the proposed method can achieve a 1.6% reduction of the energy consumption and a 57.8% computing time saving compared with the model predictive control in a fixed horizon.