Dear editor, In recent years,data-driven car-following models have been developed based on their ability to drill down to information in driving data and their flexibility.According to a study by Fleming et al.[1],the main limitation of existing methods is an insufficient amount of natural driving data.In addition,there are many situations during actual driving processes,such as those under extreme conditions[2]and in the early stages of car-following behavior modeling.In these cases,sparse data learning algorithms are indispensable.

The prediction of energy consumption is the primary goal of an intelligent energy management system (IEMS). Based on the actual road-traffic conditions, the vehicle energy consumption on the whole planned path can be predicted online by road condition recognition or speed sequence prediction. Because the speed sequence prediction required by the latter cannot accurately reflect the real dynamic characteristics of vehicle speed such as acceleration and deceleration changes due to the random factors of traffic or human beings, which will greatly affect the predicting accuracy, especially on the urban road with complex working conditions. Therefore, based on the analysis of the cumulative relationship between vehicle speed characteristics and energy consumption, this study proposes a prediction method of vehicle driving energy consumption based on the statistical characteristics of vehicle speed, regardless of the accuracy of the prediction of vehicle speed sequence, including the establishment of a long-term vehicle speed feature prediction model and energy consumption prediction model by BP and SVM algorithms. Finally, its rationality is validated based on the authentic data with an accuracy of about 95%, significantly improved compared with that based on long-term vehicle speed prediction.

Car-following model is indispensable to evaluate the characteristics of car-following behaviors. Through an analysis and comparison of data-driven and theoretically driven car-following models, it shows that the data-driven model has poor interpretability and high quality of training data set is required, while for the theoretical-driven model, it is unable to describe the individualized features and models of the driver so as to a low model accuracy. To solve the problem, a novel modelling method is proposed using adaptive Kalman filter algorithm to integrate the long-short-time memory neural network (LSTM) data-driven model with the IDM theoretical-driven model to build the car-following model. Test results of real driving data from a single driver prove that the fusion car-following model has higher accuracy than a single model, while at the same time highlighting the driver's personality compared to the IDM model. Besides, it improves the generalization ability of the traditional data model, which is reflected by better fitting in the extreme case (for example, the stable state when the acceleration, velocity is zero). Finally, the trajectory simulation results show that the proposed integrated data-driven car-following model can better simulate the micro-traffic behavior of car following.

The accurate prediction of vehicle speed plays an important role in vehicle’s real-time energy management and online optimization control. However, the current forecast methods are mostly based on traffic conditions to predict the speed, while ignoring the impact of the driver-vehicle-road system on the actual speed profile. In this paper, the correlation of velocity and its effect factors under various driving conditions were firstly analyzed based on driver-vehicle-road-traffic data records for a more accurate prediction model. With the modeling time and prediction time considered separately, the effectiveness and accuracy of several typical artificial-intelligence speed prediction algorithms were analyzed. The results show that the combination of niche immunegenetic algorithm-support vector machine (NIGA-SVM) prediction algorithm on the city roads with genetic algorithm-support vector machine (GA-SVM) prediction algorithm on the suburb roads and on the freeway can sharply improve the accuracy and timeliness of vehicle speed forecasting. Afterwards, the optimized GA-SVM vehicle speed prediction model was established in accordance with the optimized GA-SVM prediction algorithm at different times. And the test results verified its validity and rationality of the prediction algorithm.