| This paper presents Chargym, a Python-based openai-gym compatible environment, that simulates the charging dynamics of a grid connected Electrical Vehicle (EV) charging station. Chargym transforms the classic EV charging problem into a Reinforcement Learning setup that can be used for benchmarking of various and off-the-shelf control and optimization algorithms enabling both single and multiple agent formulations. The incorporated charging station dynamics are presented with a brief explanation of the system parameters and function of the technical equipment. Moreover, we describe the structure of the used framework, highlighting the key features and data models that provide the necessary inputs for optimal control decisions. Finally, an experimental performance analysis is provided using two different state-of-the-art Reinforcement Learning (RL) algorithms validating the operation of the provided environment. |
*** Title, author list and abstract as seen in the Camera-Ready version of the paper that was provided to Conference Committee. Small changes that may have occurred during processing by Springer may not appear in this window.