Analysis of energy consumption for electric buses based on low-frequency real-world data

Energy consumption management is a referential approach to alleviate range anxiety. In this paper, the data-driven modeling and analysis of energy consumption based on low-frequency real-world electric bus data is implemented. Specifically, two quantitative metrics of energy consumption are identified and calculated regarding to the characteristics of low-frequency real-world data. Also, the Road-Driver-Vehicle-Ambient factors system is proposed, forming a comprehensive digital description of the driving process. Supervised learning models with different levels of complexity are then established to predict the energy consumption, with the mean absolute percentage error of 9.2%, 7.5% and 7.7% respectively. Last but not least, the relationship between each factor and energy consumption is explained through quantitative, qualitative and statistical analysis, which intuitively shows the influence of Temperature, Driver/Vehicle-related and Road-related factors on energy consumption