Reinforcement Learning–Based Energy Management Strategy for a Hybrid Electric Tracked Vehicle

摘要: This paper presents a reinforcement learning (RL)–based energy management strategy for hybrid electric tracked vehicle. A control-oriented model of the powertrain and vehicle dynamics is first established. According to sample information experimental driving schedule, statistical characteristics at various velocities are determined by extracting transition probability matrix power request. Two RL-based algorithms, namely Q-learning Dyna applied generate optimal control solutions. The two algorithms simulated on same simulation results compared clarify merits demerits these algorithms. Although algorithm faster (3 h) than (7 h), its fuel consumption 1.7% higher that algorithm. Furthermore, registers approximately as dynamic programming–based global solution. computational cost substantially lower stochastic programming.