Reinforcement Energy Management Strategy for a Plug-in Hybrid Electric Vehicle Considering State-of-Charge Constraint

Energy management strategy is a crucial factor impacting the fuel efficiency in hybrid electric vehicles. In this paper, an energy management strategy settled by reinforcement learning is proposed for a plug-in hybrid electric vehicle (PHEV). First, the architecture of the target PHEV is modeled and analyzed in detail. As one of the effective state-action value estimation method in temporal-difference learning, Q-learning is then employed to solve the energy management problem. The main mechanism of Q-learning and specified training process considering battery state of charge planning is articulated, and the corresponding simulation validations are implemented sequentially. Comparing to global optimal control strategies, the proposed algorithm performs near-optimal power allocation with the state of charge planning and constraint.