Advancements in Electric Vehicle Charging Optimization: A Survey of Reinforcement Learning Approaches

Amid increasing concerns about climate change and energy scarcity, significant attention has been directed toward incorporating renewable energy sources, energy storage solutions, and electric vehicles (EVs) into contemporary power grids. Incorporating EVs into smart grids presents a viable solution for lowering greenhouse gas emissions. However, effectively managing their charging and discharging as distributed energy resources introduces significant challenges. The unpredictable nature of renewable energy output, uncertainties in EV characteristics, variable electricity pricing, and shifting load demands further undermine the stability of power systems. Efficient EV charging management is essential to optimize these processes while ensuring system reliability, security, and efficiency. Reinforcement learning (RL), especially when paired with deep learning techniques, has attracted significant attention due to its ability to function without a predefined model and to make real-time decisions for optimizing EV charging. This paper reviews existing research on RL-based EV charging strategies, categorizing them into centralized and decentralized approaches. Furthermore, it highlights key challenges and proposes future research directions to enhance RL-driven EV charging coordination in power systems.