ELECTRIC VEHICLE SHIFT STRATEGY BASED ON MODEL PREDICTIVE CONTROL

In order to satisfy high torque output and high speed driving demand, electric vehicles need a gearbox to adjust the gear ratio. The shift schedule is popular in gear shift research. The most widely used schedule, the two-parameter shift schedule, ignores the influence of dynamic conditions, resulting in that it is hard to suit the road and it causes energy waste. In this paper, a strategy based on model predictive control is proposed. A Recurrent neural network is used to predict velocity sequences in the 5-second horizon. Dynamic programming is adopted to construct a benchmark strategy and also to act as the rolling optimization part of the MPC shift schedule. Simulation results show that this shift strategy can reduce the shift frequency while saving energy consumption.