Optimization strategy for large-scale electric vehicle charging and discharging based on deep reinforcement learning

Abstract As the number of electric vehicles continues to grow, uncoordinated charging behaviour significantly increases users’ charging costs and exacerbates the load pressure on distribution networks during peak periods. This paper proposes an EV charging and discharging optimization strategy based on deep reinforcement learning to address these issues. First, the features of EVs entering the charging station are extracted. Second, the problem of coordinated charging and discharging at charging stations is modelled as a Markov Decision Process by fixing the dimensions of the state and action spaces. Third, a Double Deep Q-Network algorithm is employed to achieve optimal charging and discharging strategies for EVs. Finally, simulation results demonstrate the effectiveness and feasibility of the proposed method in reducing user costs and balancing the load on distribution networks.