Deep Reinforcement Learning-based Electric Vehicle Charging Optimization with Renewable Generations

With the increasing penetrations of electric vehicles (EV) and distributed renewable generations in power distribution systems, the peak load of EVs are overlapped with the conventional peak load of the distribution systems and creates the peak curve, which challenges the system operations. In this work, a deep reinforcement learning (DRL)-based EV charging optimization strategy is proposed. Firstly, the problem is formulated as a Markov Decision Process (MDP) for DRL algorithms. Secondly, the Deep Deterministic Policy Gradient (DDPG) algorithm is implemented to optimize the charging decisions of EVs. Lastly, distributed solar generation and location energy storage systems are utilized to further improve the economy and reduce carbon emissions. The proposed DRL-based framework is tested with models and the performance is validated with real-world data.