Electric vehicles (EVs) have rapidly developed in recent years and their penetration has also significantly increased, which, however, brings new challenges to power systems. Due to their stochastic behaviors, the improper charging strategies for EVs may violate the voltage security region. To address this problem, an optimal EV charging strategy in a distribution network is proposed to maximize the profit of the distribution system operators while satisfying all the physical constraints. When dealing with the uncertainties from EVs, a Markov decision process model is built to characterize the time series of the uncertainties, and then the deep deterministic policy gradient based reinforcement learning technique is utilized to analyze the impact of uncertainties on the charging strategy. Finally, numerical results verify the effectiveness of the proposed method.

中文翻译：

---

基于马尔可夫决策过程和强化学习技术的最优电动汽车充电策略

近年来，电动汽车（EV）发展迅速，普及率也显着提高，但这也给电力系统带来了新的挑战。由于电动汽车的随机行为，不正确的电动汽车充电策略可能会违反电压安全区域。为了解决这个问题，提出了配电网络中的最佳电动汽车充电策略，以在满足所有物理约束的同时最大化配电系统运营商的利润。在处理电动汽车的不确定性时，建立马尔可夫决策过程模型来表征不确定性的时间序列，然后利用基于深度确定性策略梯度的强化学习技术来分析不确定性对充电策略的影响。最后，