Machine Learning-Based Energy Demand Prediction for Electric Buses

Abstract : This study explores the use of machine learning (ML) to predict the energy demand of Battery Electric Buses (BEBs), aiming to improve the efficiency and sustainability of urban transportation systems. Electrification of city buses plays a critical role in achieving greener, more sustainable public transport. However, predicting energy demand accurately is a complex task that requires efficient methods for vehicle design and fleet management. Traditional physics-based models are often cumbersome and time-consuming, while data-driven machine learning models offer scalable and faster alternatives. This research proposes a novel ML-based methodology that predicts the energy economy of BEBs using real-world driving data, incorporating key features such as vehicle speed and passenger load. By leveraging these simplified input variables, the proposed model achieves over 94% accuracy in predicting energy demand, making it highly practical for real-world applications. The ability to accurately predict energy consumption helps optimize fleet operations, reduce operational costs, and enhance energy efficiency, benefiting manufacturers, fleet operators, and urban communities. Additionally, the simplicity of the input variables makes the system robust and adaptable, requiring minimal computational resources. This study demonstrates the potential of machine learning in the electrification of transportation systems, providing a pathway for cost-effective and energy-efficient urban mobility solutions that can contribute to the wider adoption of electric buses in cities globally. The proposed methodology presents a significant step toward transforming public transportation systems into more sustainable, environmentally friendly models of mobility.

Keyword Battery Electric Buses (BEBs), Energy Economy Prediction, Machine Learning, Data-Driven Models, Speed Profile Features, Sustainable Transportation, Fleet Optimization, Energy Consumption Modeling.