Data-driven Macroscopic Energy Consumption Estimation for Electric Vehicles with Different Information Availability

Electric vehicles (EVs) have shown great potential in reducing transportation-related greenhouse gas emission and fossil fuel consumption in recent years. Moreover, eco-routing systems can be used to further improve the energy efficiency of EVs by choosing the most energy-efficient route based on real-world traffic conditions. These eco-routing systems as well as accurate range prediction rely on reliable EV energy consumption models to calculate EV energy consumption of the different route options. This work proposes a set of link-level EV energy consumption models that estimate EV energy consumption on each roadway link considering real-world traffic congestion. To develop the models, EV energy consumption data under real-world traffic congestion is decomposed into positive kinetic energy (PKE) and negative kinetic energy (NKE). Upon this decomposition, two energy consumption models are built and compared based on different sets of available information. The results of numerical analysis show that the proposed the model with PKE and NKE as predicting variables can achieve more than 95% accuracy.