A SUMO-Simulated Reinforcement Learning Framework for Bidirectional Charging of Electric Vehicles

Abstract

This paper presents a reinforcement learning-based approach to optimize bidirectional charging strategies for electric vehicles. The proposed algorithm effectively manages the charging and discharging processes, considering dynamic energy prices and varying driving patterns while ensuring minimal disruption to the user experience. The algorithm is evaluated through simulation using the SUMO framework, demonstrating its ability to reduce charging costs and enhance the overall efficiency of bidirectional charging. Key findings include the algorithm's effectiveness in adapting to diverse driving scenarios and its potential to contribute to a more sustainable and efficient energy grid. The algorithm is connected to the SUMO framework with simple interfaces to easily replace either simulation or the reinforcement learning algorithm in further research.