Energy Consumption Prediction of Electrical Vehicles Through Transformation of Time Series Data

In recent years, global greenhouse gas emissions have become an important issue. Electrical Vehicles (EVs) are a promising solution to reduce greenhouse gas emissions. However, widespread adoption of EVs is hindered in part by “range anxiety” wherein the remaining drivable distance is unknown. Models that can predict the energy consumption of EVs can not only address this issue, but also assist in optimizing charging patterns. This paper proposes a model for EV energy consumption prediction that adopts relevant factors (temperature, road gradient, vehicle loading, etc.) to characterize a driver's real-life vehicle usage. By using a novel transformation of the input data, the proposed model can output the consumption prediction of multiple users simultaneously. Additionally, the model can process multiple types of car models that effectively use information specific to that group. The results show that the system can parallelize the prediction task with desired accuracy when compared to the single-output system and offers the ability to model different EV models at the same time.