Abstract: In conventional adaptive cruise control system based on model predictive control algorithm, the comfort constraint was in the form of fixed acceleration. The problem of range of acceleration extremes at different speeds was not considered in depth, which would lead to excessive changes in system acceleration and impact in the pursuit of safety and speed following, thus reduced the comfort of driving and riding. The NGSIM natural driving data set was used to obtain the speed-acceleration relationship of the vehicle, which was used as the constraint object of acceleration. The constant constraint of the conventional model predictive control was changed to the variable constraint following the change of speed, to improve the following comfort based on natural driving data. The rationality of the proposed method was verified by simulation test. The research results show that the proposed method can make the acceleration and impact fluctuation of ACC system smaller while ensuring safe following and speed following. Compared with the model predictive control with constant acceleration constraint, the maximum impact decreases by 14.96%, which effectively improves the comfort of ACC system.

Key words: traffic engineering; adaptive cruise system; model predictive control; variable acceleration constraint; NGSIM; comfort

摘要： 在常规的基于模型预测控制自适应巡航系统的算法中，舒适性约束均是采用固定加速度的形式，没有深入考虑不同速度下的加速度极值范围问题，这会导致在追求安全性和速度跟随性情况下，系统加速度和冲击度变化过大，从而降低了驾乘的舒适性。根据NGSIM自然驾驶数据集所得到了车辆的速度-加速度关系，并以此作为加速度的约束目标，将常规模型预测控制的定约束改为随速度变化的变约束，从而达到根据自然驾驶数据提高跟车舒适性的目的，并通过仿真试验验证了该方法的合理性。研究结果表明：该方法可使ACC系统在保证安全跟车和速度跟随的同时，使得加速度和冲击度波动变小；最大冲击度相比定加速度约束的模型预测控制减小了14.96%，可有效改善ACC系统的舒适性。

关键词: 车辆工程；自适应巡航系统；模型预测控制；变加速度约束；NGSIM；舒适性