A hybrid on-line approach for predicting the energy consumption of electric buses based on vehicle dynamics and system identification

Energy consumption modeling for electric vehicles is conducive to achieving eco-driving optimization and alleviating driver 'range anxiety'. However, it remains challenging due to the complexity of influencing factors. In this study, we propose a novel hybrid cumulative energy consumption (CEC) prediction approach by a combination of vehicle dynamics and the forgetting factor recursive least squares (FFRLS) to achieve real-time online forecasting accurately. The approach is developed with real-world data collected from over 983,000 frames across six electric buses in Guangzhou, China. To enhance the accuracy and practical application of the approach, we divide the data into segments as a unit of calculation for CEC. Subsequently, we constructed the model in three working conditions based on the vehicle longitudinal dynamics, and determined the parameters to be estimated for the model. With the help of the online learning capabilities of FFRLS algorithm, the estimated parameters can be constantly adjusted and updated according to the state of the buses. The results demonstrate that the mean absolute percentage error (MAPE) of the proposed method is 7.05 %. This performance surpasses that of existing relevant studies, indicating that the model provides a more accurate method for eco-driving and city-scale electric bus operation systems.