Energy Management of Autonomous Electric Vehicles by Reinforcement Learning Techniques

The increase in e-mobility poses new challenges to power grid operators who must cope with the variability and uncertainty of renewable energy sources and customer demand and the electric vehicle integration into the grid. In this paper a Reinforcement Learning algorithm based on Principal Policy optimization is proposed for energy management of electric vehicles and PV storage units. The RL algorithm considers the vehicle battery constraints, range anxiety and battery aging constraints. Moreover, the algorithm controls the charging of the photovoltaic storage unit to minimize the PV energy curtailment. Results show the improvement of the proposed algorithm compared to Business-as-usual and value-iteration solutions.