A review of reinforcement learning based energy management systems for electrified powertrains: Progress, challenge, and potential solution

The impact of internal combustion engine-powered automobiles on climate change due to emissions and the depletion of fossil fuels has contributed to the progress of electrified powertrains. Energy management strategies (EMS) have shown huge impact on the energy efficiency of the electrified powertrains. In recent years, Reinforcement Learning (RL) based algorithms have been intensely investigated to develop EMS for electrified powertrain with specific depth in hybrid electric vehicles (HEV), battery electric vehicles (BEV) and fuel cell vehicles (FCV) and the research in this area is still acelerating. However, a comprehensive review of RL-based EMS is lacking in literature. This article reviews the recent penetration of RL based EMS like Q-learning, Deep Q Learning, deep deterministic policy gradient in the electrified powertrains domain. Extensive importance is given to the classification of the literature based on powertrain architecture, RL algorithm, and the different features and operation mechanisms of relevant algorithms are highlighted. The use of connected and autonomous vehicles and relevant communication technology to develop RL-based systems have also been discussed. More importantly, the challenges with regards to existing research in the field of RL-based EMS and the potential solutions with scope for future research are also discussed.

Automated summary

This article discusses the impact of internal combustion engine-powered automobiles on climate change and the progress of electrified powertrains, focusing on the application of Reinforcement Learning (RL) based algorithms in the field. It reviews the recent research and potential solutions for challenges in the area of RL-based EMS.