Electric Vehicle Charge Planning by Deep Reinforcement Learning

In this paper, we study the problem of planning the charging of an electric vehicle (EV) for long distance trips. In particular, a dedicated tool is needed in order to suggest the driver the best traveling and charging plan based on her/his preferences and on the available charge infrastructures along the route. An ad-hoc tool is developed using Matlab and Simulink software in order to model the EV energy charging/discharging, and to optimize the trip plan based on three alternative goals: minimizing the travel time; minimizing the charging cost; optimizing travel time and cost. The proposed model is based on the use of Deep Reinforcement Learning (DRL) algorithm in which an agent is rewarded or penalized while learning the best charging options along the route according to the three above goals. In particular, in a dynamic energy market context, the model is designed to be robust at energy price variations and at cruise speed variations as well. In addition, a dashboard is designed and developed to provide an user-friendly interface to set the EV and trip parameters and to monitor the trip planning results. A case study demonstrates the application of the proposed tool.