Reinforcement learning based energy management for fuel cell hybrid electric vehicles: A comprehensive review on decision process reformulation and strategy implementation

—Reinforcement learning (RL) has shown great application prospects from both industry and academia in recent years, due to its success in surpassing human level performance in several applications. Researchers have also been interested in implementing RL solutions into energy management problem of fuel cell hybrid electric vehicle (FCHEV), and their effort has reached considerable achievements. The existing overviews simply classified and summarized the research findings, without in-depth study on how to reformulate the energy management strategy (EMS) into Markov decision process (MDP). Therefore, to fill this gap, this study attempts to provide a comprehensive review of this topic. This study begins with an introduction to the structural features of FCHEV and an overview of energy management issues and the existing EMS literature. Then, for the first time, the reformulation process of the EMS issue into RL framework is explored. Afterwards, a compendious categorization of widely applied RL algorithms is introduced, and the details of several widely applied RL algorithms are presented, recent successes of RL-based EMS issues is summarized. Finally, this study summarizes the problems and prospects of RL-based EMS.