A transformer-based approach for deep feature extraction and energy consumption prediction of electric buses based on driving distances

Changyin Dong, Zhuozhi Xiong, Chu Zhang, Ni Li, Ye Li, Ning Xie, Jiarui Zhang and Hao Wang

Applied Energy, 2025, vol. 380, issue C, No S0306261924013242

Abstract: Electric buses (EBs) are widely recognized as an environmentally friendly and energy-efficient alternative to traditional diesel buses. However, the issue of driving range anxiety hinders their further adoption and popularity. This study proposes a transformer-based approach to predict the energy consumption rate (ECR) of EBs based on driving distances by extracting implicit features from dynamic characteristics. First, high-resolution bus data is acquired and divided into trip segments, which are then fused with meteorological and road network data. Second, sliding windows are employed for different travel ranges to construct the dataset. Each sample in the dataset contains 12 historical dynamic feature time series and 12 predetermined features within the predicted range. The historical dynamic features are deeply extracted using the Transformer encoder module to obtain implicit features like driver behavior and bus kinematic characteristics. They are then fused with other factors, and the prediction of the ECR is performed using a fully connected neural network. Finally, a sensitivity analysis is conducted for each feature to demonstrate its impact and variations under different implicit features and travel distances. The results indicate that the prediction accuracy improves as the driving range increases. The mean absolute error (MAE) decreases from 0.28kWh/km for 250 m to 0.09kWh/km for 10 km. Compared to seven methods in the literature, this approach reduces the MAE by 1% to 39% in various conditions. This research contributes to accurate prediction of energy consumption for future distances, and supports the real-time calculation of remaining battery-supported distance.

Keywords: Electric bus; Energy consumption prediction; Transformer; Deep learning; Sensitivity analysis (search for similar items in EconPapers)
Date: 2025
References: View references in EconPapers View complete reference list from CitEc
Citations:

Downloads: (external link)
http://www.sciencedirect.com/science/article/pii/S0306261924013242
Full text for ScienceDirect subscribers only

Related works:
This item may be available elsewhere in EconPapers: Search for items with the same title.

Export reference: BibTeX RIS (EndNote, ProCite, RefMan) HTML/Text

Persistent link: https://EconPapers.repec.org/RePEc:eee:appene:v:380:y:2025:i:c:s0306261924013242

Ordering information: This journal article can be ordered from
http://www.elsevier.com/wps/find/journaldescription.cws_home/405891/bibliographic
http://www.elsevier. ... 405891/bibliographic

DOI: 10.1016/j.apenergy.2024.123941

Access Statistics for this article

Applied Energy is currently edited by J. Yan

More articles in Applied Energy from Elsevier
Bibliographic data for series maintained by Catherine Liu ().