In recent years, fully electric vehicles (EVs) have accounted for a higher proportion of urban buses. On account of their relatively short cruising range, many technologies were used to improve the driving range, such as improving energy management strategies and promoting the performance of the battery and engine. The purpose of this study was to develop an energy-efficient driving strategy to save energy. The strategy consists of two parts: determining a velocity interval for lower energy and establishing an energy-saving acceleration mode. First, 30 velocity datasets were collected from an actual bus line. An electric bus model and energy consumption equation were established in the AVL CRUISE software to analyze the energy consumption. Next, the velocity interval was determined based on the actual data with the objective of maximizing engine efficiency and minimizing energy consumption. By considering uniform motion and traffic conditions, 30-40 km/h was determined as the velocity interval of lower energy for the electric bus mentioned in this paper. The acceleration characteristic parameter β represents the curve of velocity versus time in different acceleration processes, which was chosen to describe the acceleration mode; we found that when β is greater than 0.2, the lower the β value, the lower the energy consumption per kilometer was. Finally, the energy-saving driving strategy was verified by conducting a simulation. It was determined that the reduction in energy consumption per kilometer after implementing the energy-saving driving strategy was between 12.32% and 18.7% for short sections of acceleration and 2.47% for the entire bus trip.

Speed has an important impact on driving safety, however, this factor is not included in existing safety warning algorithms. This study uses lane change systems to study the influence of vehicle speed on safety warning algorithms, aiming to determine lane change warning rules for different speeds (DS-LCW). Thirty-five drivers are recruited to carry out an extreme trial and naturalistic driving experiment. The vehicle speed, relative speed, relative distance, and minimum safety deceleration (MSD) related to lane change characteristics are then analyzed and calculated as warning rule characterization parameters. Lane change warning rules for a rear vehicle in the target lane under four-speed levels of 60 ≤ v < 70 km/h, 70 ≤ v < 80 km/h, 80 ≤ v < 90 km/h, and v ≥ 90 km/h are established. The accuracy of lane change warning rules not considering speed level (NDS-LCW) and ISO 17387 are found to be 87.5% and 79.8%, respectively. Comparatively, the accuracy rate of DS-LCW under four-speed levels is 94.6%, 93.8%, 90.0%, and 92.6%, respectively, which is significantly superior. The algorithm proposed in this paper provides warning in the lane change process with a smaller relative distance, and the accuracy rate of DS-LCW is significantly superior to NDS-LCW and ISO 17387.

When passing through an intersection section, the relationship between driving behavior and energy consumption of pure Electric Buses (E-Bus) is unclear. In this study, natural driving data on two Bus Rapid Transit (BRT) routes were collected to quantify and analyze the implied relationship between driving behavior and energy consumption when entering an intersection point. Furthermore, it is proposed that predicting energy consumption on the basis of distinguishing whether an intersection is stopping or non-stopping would be a more accurate scenario. Concerning the working method, firstly, statistical analysis is used to observe the difference between the energy consumption of the stopping and non-stopping samples; secondly, correlation analysis and linear regression are used to analyze the significant parameters related to whether to stop or not, and energy consumption; finally, machine learning method is used to establish the classification model of whether to stop or not at an intersection as well as a prediction model the energy consumption of the intersection.The results show that the model accuracy of XGBoost-KNN is higher than that of KNN and XGBoost in predicting whether to stop or not, which is 84.4%. For predicting energy consumption, the GBDT has the lowest prediction accuracy; as for XGBoost and SVM, which have a higher prediction accuracy, distinguishing whether to stop or not helps to enhance the model’s prediction accuracy. Furthermore, after distinguishing whether to stop or not, SVM outperformed XGBoost in R², MAE, and RMSE. Research results provide a new perspective for studying the relationship between the driving behavior and energy consumption of pure electric buses at intersections. Meanwhile, they also offer the possibility for further research on the applicability of energy consumption when expanding from the BRT to more complex mixed traffic environments.

This paper evaluates the ecological level of driving behavior of electric buses when entering and leaving stops. A dataset of entering and leaving stops is first created based on the natural driving data of electric buses. The representative parameters of driving behaviors for entering and leaving stops are then selected through correlation analysis and multiple stepwise linear regression analysis. Afterwards, the threshold value for defining the eco-driving behavior is determined by analyzing the energy consumption characteristics of entering and leaving stops. Finally, the Random Forest (RF), Gradient-Boosted Decision Trees (GBDT), and Light Gradient Boosting Machine (LightGBM) algorithms are applied to develop the evaluation models of eco-driving level for entering and leaving stops. The obtained results show that the accuracies of the LightGBM model for the evaluation of the eco-driving level in entering and leaving stops are 89% and 86.7%, respectively. These values are better than those of the RF and GBDT algorithms, and thus they demonstrate that the LightGBM model can more accurately evaluate the eco-driving in entering and leaving stops.