Predicting Electric Bus Energy Consumption Using Explainable Machine Learning

Battery electric buses (BEBs) must finish every trip on a fixed store of energy, so a poor trip-level forecast costs either a stranded bus or a needless charging stop. This paper presents a reproducible pipeline that predicts trip energy intensity (kWh/km) from seven pre-trip variables, explains each prediction with SHAP (SHapley Additive exPlanations), and passes the forecast to a pre-dispatch range-assurance check and a time-of-use depot-charging scheduler, all served through a 13-endpoint REST interface. Real fleet telemetry was not available, so the study uses two physics-based synthetic benchmarks: a 10,459-trip RouteZero-style set (RZ-Syn) and an 8,500-trip Indian-conditions set (IN-Syn). Nineteen physically motivated features feed four regressors. On six held-out buses (1,864 trips), histogram gradient boosting reaches MAE 0.0374 kWh/km, RMSE 0.0541 kWh/km, R² 0.9916 and MAPE 3.45%, against a generator noise ceiling of R² 0.9949; over 20 re-drawn bus splits MAPE is 3.44 ± 0.08%, and the Indian benchmark gives R² 0.9829. Stress tests temper these figures: withholding the hillside regime raises MAPE to 6.43%, transfer between benchmarks raises it above 20% with systematic under-prediction, and a per-variable input-range check does not reliably detect either. The nominal 95% interval covers only 86.1% of test trips; split-conformal recalibration restores 94.5%. In a low-battery dispatch test the worst-case range check missed none of 735 infeasible trips, whereas a static average-consumption rule missed 10.3%. Smart charging cut simulated electricity cost by 32–71%. Validation on measured fleet data is the necessary next step

battery electric bus, energy consumption prediction, gradient boosting, SHAP, explainable machine learning, range assurance, smart charging, synthetic data

1. B. Sturmberg, J. N. Hendriks, et al., “An integrated model of electric bus energy consumption and optimised depot charging,” npj Sustain. Mobil. Transp., 2024, doi: 10.1038/s44333-024-00008-2. [Google Scholar] [Crossref]

2. H. Basma, C. Mansour, M. Haddad, M. Nemer, and P. Stabat, “Comprehensive energy modeling methodology for battery electric buses,” Energy, vol. 207, Art. no. 118241, 2020. [Google Scholar] [Crossref]

3. H. Abdelaty, A. Al-Obaidi, M. Mohamed, and H. E. Z. Farag, “Machine learning prediction models for battery-electric bus energy consumption in transit,” Transp. Res. D, Transp. Environ., vol. 96, Art. no. 102868, 2021. [Google Scholar] [Crossref]

4. H. Abdelaty and M. Mohamed, “A prediction model for battery electric bus energy consumption in transit,” Energies, vol. 14, no. 10, Art. no. 2824, 2021. [Google Scholar] [Crossref]

5. S. Nan, R. Tu, T. Li, J. Sun, and H. Chen, “From driving behavior to energy consumption: A novel method to predict the energy consumption of electric bus,” Energy, vol. 261, Art. no. 125188, 2022. [Google Scholar] [Crossref]

6. S. Nan, F. Liao, T. Li, H. Chen, and J. Sun, “Identifying the electricity-saving driving behaviors of electric bus based on trip-level electricity consumption: A machine learning approach,” Environ. Sci. Pollut. Res., vol. 30, no. 34, pp. 82743–82759, 2023. [Google Scholar] [Crossref]

7. W. Wang, M. Mohamed, B. Zhang, Q. Zhang, and H. Abdelaty, “Electric bus energy prediction and factors interactions using explainable machine learning models,” Eng. Appl. Artif. Intell., vol. 163, Art. no. 112877, 2026. [Google Scholar] [Crossref]

8. J. Vepsäläinen, A. Ritari, A. Lajunen, K. Kivekäs, and K. Tammi, “Energy uncertainty analysis of electric buses,” Energies, vol. 11, no. 12, Art. no. 3267, 2018. [Google Scholar] [Crossref]

9. M. Gallet, T. Massier, and T. Hamacher, “Estimation of the energy demand of electric buses based on real-world data for large-scale public transport networks,” Appl. Energy, vol. 230, pp. 344–356, 2018. [Google Scholar] [Crossref]

10. M. Almajed, M. Hassan, T. D. Turjo, M. A. Islam, M. E. Haque, and M. M. H. Rahman, “Explainable machine learning for electric bus energy prediction under tropical urban conditions: A physics-informed parametric framework,” Energy Eng., vol. 123, no. 10, 2026, doi: 10.32604/ee.2026.084024. [Google Scholar] [Crossref]

11. S. M. Lundberg and S.-I. Lee, “A unified approach to interpreting model predictions,” in Proc. Adv. Neural Inf. Process. Syst., vol. 30, 2017, pp. 4765–4774. [Google Scholar] [Crossref]

12. T. Chen and C. Guestrin, “XGBoost: A scalable tree boosting system,” in Proc. 22nd ACM SIGKDD Int. Conf. Knowl. Discovery Data Mining, 2016, pp. 785–794. [Google Scholar] [Crossref]

13. G. Ke et al., “LightGBM: A highly efficient gradient boosting decision tree,” in Proc. Adv. Neural Inf. Process. Syst., vol. 30, 2017, pp. 3149–3157. [Google Scholar] [Crossref]

14. J. H. Friedman, “Greedy function approximation: A gradient boosting machine,” Ann. Statist., vol. 29, no. 5, pp. 1189–1232, 2001. [Google Scholar] [Crossref]

15. L. Breiman, “Random forests,” Mach. Learn., vol. 45, no. 1, pp. 5–32, 2001. [Google Scholar] [Crossref]

16. F. Pedregosa et al., “Scikit-learn: Machine learning in Python,” J. Mach. Learn. Res., vol. 12, pp. 2825–2830, 2011. [Google Scholar] [Crossref]

17. D. R. Roberts et al., “Cross-validation strategies for data with temporal, spatial, hierarchical, or phylogenetic structure,” Ecography, vol. 40, no. 8, pp. 913–929, 2017. [Google Scholar] [Crossref]

18. S. Kaufman, S. Rosset, C. Perlich, and O. Stitelman, “Leakage in data mining: Formulation, detection, and avoidance,” ACM Trans. Knowl. Discovery Data, vol. 6, no. 4, Art. no. 15, 2012. [Google Scholar] [Crossref]

19. A. Lajunen, “Energy consumption and cost-benefit analysis of hybrid and electric city buses,” Transp. Res. C, Emerg. Technol., vol. 38, pp. 1–15, 2014. [Google Scholar] [Crossref]

20. K. R. Kambly and T. H. Bradley, “Estimating the HVAC energy consumption of plug-in electric vehicles,” J. Power Sources, vol. 259, pp. 117–124, 2014. [Google Scholar] [Crossref]

21. Road Vehicles—Vehicle to Grid Communication Interface—Part 1: General Information and Use-Case Definition, ISO Standard 15118-1, 2019. [Google Scholar] [Crossref]

• Classifying News Articles on Nigerian Newspapers Using Machine Learning
• Approaching Early Dyslexia Prediction with Ensemble Machine Learning: A Secondary Empirical Analysis of the Rello Et Al. Public Behavioural Dataset
• Machine Learning Modelling of Geostationary SBAS Satellite Orbit Errors Observed from an Equatorial West African Station
• A Comparative Study of CNN and Vision Transformer Models in Alzheimer's Detection
• A Stacking Ensemble Machine Learning Framework for Predicting Academic Performance and Identifying At-Risk Students in Rural Secondary Schools: Evidence from Aurangabad District, Maharashtra