Real-Time Energy Consumption Prediction of Connected Automated Electric Vehicles Based on Temporal Fusion Transformers

Adopting electric vehicles has great potential to address the energy and environmental crisis posed by the current transportation system. Concerning the battery capacity limitation, accurate and real-time energy consumption (EC) predictions are critical to improving driver’s travel experience with electric vehicles. Additionally, although vehicles are developing toward automation and intelligence, the EC characteristics of connected automated electric vehicles (CAEVs) are under-researched. The main objective of this study is to analyze and predict the CAEVs’ EC based on real-time sensing data. First, a drone was used to collect traffic flow parameters, and the simulation scenarios were further built based on the simulation of urban mobility (SUMO). Second, the differences in the EC distribution between manual driving electric vehicles (MEVs) and CAEVs were analyzed using visualization and statistical methods. Then, the factors influencing CAEVs’ EC were analyzed and quantified. Finally, the real-time EC prediction model was proposed based on the temporal fusion transformers (TFTs). The results show that the TFT model can be utilized for EC time-scale and remaining mileage evaluation, and it performs better than the baseline algorithms. This study contributes to the current understanding of CAEVs’ EC characteristics, which are significant for environmental protection and sustainable development.