Prediction of driving energy consumption for pure electric buses using dynamic driving style recognition and speed forecasting

The accurate prediction of energy consumption for battery electric vehicles (BEVs) is crucial to reduce the anxiety of the drivers and promote the adoption of electric vehicles. Among the various factors, the driving styles and travel speeds are the most significant influencing factors for vehicle energy consumption. This study focuses on pure electric buses and explores the relationship between the driving styles of the drivers, travel speeds, and driving energy consumption from a data mining perspective. This is tailored to meet the needs for energy-efficient driving of these buses. Systematic modeling is then conducted to classify and recognize the driving styles, predict the future vehicle speeds, and predict the vehicle driving energy consumption. A speed-energy consumption prediction method, based on driving style recognition, is proposed. The results obtained by conducted case studies validate the proposed method. They demonstrate that it has high accuracy, it reaches mean square errors of 4.1154 and 0.0756, and coefficient of determination accuracies of 0.7655 and 0.8893. The prediction of driving energy consumption lays a foundation for the targeted establishment of eco-driving strategies for pure electric buses. It also provides bus companies energy-saving driving training for their drivers.