The global push toward sustainable transportation has made electric buses (E-buses) a key solution for reducing greenhouse gas emissions. However, their widespread adoption faces challenges, particularly in cold climates where battery efficiency drops significantly. Canada, with its harsh winters, presents a unique case study for optimizing E-bus operations, yet existing energy consumption models perform poorly in such conditions. A new study addresses this gap by developing advanced machine learning models tailored to Canadian data, offering precise predictions for E-bus energy use under varying climates and heating systems.
The research compares multiple data-driven approaches, revealing that tree-based models—particularly Random Forest—deliver the highest accuracy in predicting energy consumption, with a mean absolute error of just 0.09–0.1 kWh/km. Notably, the study highlights stark differences in performance between bus types: while some models rely solely on battery-powered heating (leading to higher winter consumption), others use auxiliary diesel heaters, which stabilize energy use across seasons. SHAP analysis further identifies key influencing variables—weather conditions dominate for battery-heated buses, whereas operational factors (e.g., speed, acceleration) are critical for diesel-assisted systems.
These findings empower transit agencies to optimize fleet deployment and charging strategies, reducing operational costs and minimizing range anxiety. For instance, cities can prioritize diesel-assisted buses for extreme cold or adjust schedules based on seasonal consumption patterns. The models also avoid reliance on hard-to-measure variables (e.g., driver behavior), making them practical for real-world planning.
This research paves the way for smarter, climate-resilient E-bus networks. Potential applications include: (1) Regional Customization: Adapting the framework to other cold climates like Scandinavia or Northern Europe. (2) Policy Support: Helping governments set emission targets by quantifying energy savings from different heating technologies. (3) Technology Integration: Guiding manufacturers in designing next-gen batteries or hybrid heating systems. Further studies could explore AI-driven real-time adjustments for routes and charging, enhancing efficiency dynamically.
By bridging the gap between theoretical models and real-world challenges, this study marks a significant leap toward sustainable urban transit. Its interpretable machine learning approach not only boosts predictive accuracy but also provides actionable insights for policymakers and operators. As cities worldwide accelerate their shift to electric mobility, such innovations ensure that cold climates are no longer a barrier—but a manageable factor—in the journey to zero-emission transportation.
Reference
Author: Kareem Othman a b, Diego Da Silva a, Amer Shalaby a, Baher Abdulhai a
Title of original paper: Interpretable machine learning models for predicting Ebus battery consumption rates in cold climates with and without diesel auxiliary heatingc
Article link: https://doi.org/10.1016/j.geits.2024.100250
Journal: Green Energy and Intelligent Transportation
https://www.sciencedirect.com/science/article/pii/S2773153724001026
DOI: 10.1016/j.geits.2024.100250
Affiliations:
a Civil Engineering Department, University of Toronto, Toronto, Canada
b Public Works Department, Faculty of Engineering, Cairo University, Giza, Egypt
