Battery Thermal-conscious Energy Management for Hybrid Electric Bus Based on Fully-continuous Control with Deep Reinforcement Learning

This paper proposes a knowledge-based, thermal-conscious strategy for the energy management of hybrid electric bus (HEB). The deep deterministic policy gradient (DDPG) algorithm with priority experience replay (PER) is exploited to distribute the power smartly among energy components. The fully-continuous separate speed- and torque-control mechanism is further devised to excavate the upper optimization potential of PER-DDPG strategy. Moreover, in the PER-DDPG framework, the penalties to over-temperature are embedded for thermal safety enforcement. Comparative results also disclose the superiority of the proposed strategy in terms of the over-temperature protection and overall optimization performance in the energy management of HEB.