Optimal Electrical Vehicle Charging Planning and Routing Using Real-Time Trained Energy Prediction With Physics-Based Powertrain Model

Electric vehicles (EVs) are rising in popularity due to technological developments and the nation's decarbonization efforts, which are enticing drivers to move away from gasoline-dependent transportation. However, a major challenge in EV charging is optimal day-ahead charging planning under time-varying conditions, such as fluctuating charging prices and traffic conditions. Additionally, these conditions can lead to uncertain energy consumption while driving an EV. Predicting the energy consumption of an electric vehicle is complicated because it is influenced by more factors than a typical battery, such as driving habits and traffic conditions. This paper proposes an optimal EV charging planning strategy using an artificial intelligence-based energy consumption prediction model to address this problem. The energy consumption prediction model, based on long short-term memory (LSTM), was developed and validated against other artificial intelligence models. The LSTM-based model uses real-time training, making it suitable for uncertain scenarios encountered by EVs. The predicted results and traffic information are considered as weight factors in solving the EV charging planning and routing optimization problem. The optimization problem is formulated as mixed integer linear programming. Simulation studies under various driving and traffic conditions validate the proposed optimal charging planning and routing method. Since travel time can vary according to driving characteristics and traffic conditions over the same driving distance, the results confirm that the proposed method can predict energy consumption more accurately than other methods. It also selects routes with lower energy consumption relative to total driving time and provides stable operation.