Charging price learning based on inverse deep reinforcement learning

Electric vehicles (EVs) have attracted widespread attention in recent years due to their energy-saving and environmentally friendly characteristics. With the rapid adoption of EVs, the charging demand has continuously grown. Consequently, an increasing number of EV charging stations from various vendors have emerged to provide charging services to EV users. To maximize the operational profits of a charging station, it is necessary to study the pricing reward functions of competitor charging stations to forecast their pricing actions and optimize the operational control strategy of its own charging station accordingly. However, this problem is non-trivial to solve due to partially-observable competitor pricing states, the dynamic nature of pricing policy, and the difficulty in precisely modelling pricing impacts. Therefore, this paper considers the competitor pricing reward learning in a competitive charging market. First, the pricing reward function learning problem is formulated as a Markov Decision Process (MDP), incorporating the partial observation of competitor information and the learning reward function. Next, a pricing reward function learning method based on inverse deep reinforcement learning is proposed, which can generate the pricing actions of competitors. Finally, through simulation analysis of the learned pricing reward function structure, it demonstrates that the proposed method can effectively learn the pricing reward function of competitor charging station.