Predicting Remaining Driving Range of Electric Vehicles Using a Lightweight Transformer Model in a Connected Vehicle Environment
Summary
No summary yet. VIRAT can read this paper and write a plain-language summary of its aim, methods, key findings and limitations.
Summarise with VIRAT Free to try with your Science Catalogs account.Abstract
As electric vehicles become more widespread, range anxiety has emerged as a critical issue hindering their development. Traditional remaining driving range (RDR) prediction methods rely solely on vehicle sensor data, neglecting the impact of dynamic environments and driving behavior. Furthermore, these models exhibit high computational complexity, making deployment on in-vehicle devices challenging. This study proposes an RDR prediction method based on a lightweight Transformer model for the vehicle-toeverything (V2X) environment. Experiments using Shenzhen new energy vehicle data demonstrate excellent prediction performance, with an R2of 0.8121, MSE of 0.1617, and MAE of 0.2961. Prediction errors cluster near zero and exhibit a normal distribution. This approach provides an effective solution for high-precision real-time RDR prediction in V2X environments, advancing the evolution of electric vehicles toward integrated traffic-IoT terminals.
Fields of Study & Subjects
Electric Vehicles and Infrastructure
Electric and Hybrid Vehicle Technologies
Vehicle emissions and performance
Publication details
- Journal / source
- IEEE International Conference on Power, Electronics and Green Energy (ICPEGE)
- Pages
- 1767–1770
- Publication date
- 28 January 2026
- Language
- EN