考虑人车路耦合的弯道车速预测模型及车辆主动控制研究

The lateral instability accidents such as rollover and sideslip are the main traffic accident forms on curves. Current safety treatments include: 1) set roadside speed limit signs along the curves. It is unable to take dynamic and individual factors into account such as changes in the weather, differences in vehicle parameters and driver behaviors etc; 2) utilize the vehicle electronic stability control system to apply active control to vehicles in lateral instability condition and correct the vehicle motion. In fact, inappropriate speed on curve negotiation is a major cause of lateral instability accidents. Different drivers have different psychological expectations and tolerance to the curve negotiation speed. Therefore, this research introduces driver behavior characteristics to the traditional curve speed model based on vehicle dynamic analysis, and recommends appropriate speed to drivers before entering the curved road based on curve speed prediction model considering driver-vehicle-road interaction, which can prevent vehicles from lateral instability in advance. Meanwhile, a driver model which mimics human driver behaviors is built based on stochastic model predictive control method accounting for uncertainty of some parameters such as road surface adhesion coefficient etc., which will take autonomous decision-making control if drivers ignore warnings. The research lays the theory foundation for the development of curve speed warning and vehicle active control technologies.

公路弯道的主要事故形态为侧翻、侧滑等侧向失稳事件，现有的安全措施主要包括：1）在弯道设置路侧限速标志，该措施难以充分考虑天气变化、车辆参数差异、驾驶员行为特性等动态与个体因素；2）利用车辆电子稳定控制系统，对已处于侧向失稳状态的车辆施加主动控制，纠正车辆运动姿态。事实上，过弯车速的合理选择，是预防侧向失稳的有效措施之一，而不同行为特性的驾驶员对过弯车速的心理预期和承受度不尽相同。因此，本项目将驾驶员行为特性引入传统基于车辆动力学分析的弯道车速模型中，建立考虑人车路耦合的弯道车速预测模型，即在车辆进入弯道前，为驾驶员推荐合理的过弯车速，提前主动预防车辆侧向失稳的危险态势；同时，考虑路面附着系数等参数的随机性，基于随机模型预测控制方法，建立仿人类驾驶行为特性的驾驶员模型，以实现对驾驶员忽略预警后的车辆进行仿人自主决策控制。本研究可为弯道车速预警及车辆主动控制技术的发展奠定理论方法基础。

本研究针对弯道行车最易发生的侧翻、侧滑等侧向失稳事件，探究人车路耦合因素对弯道行车安全的影响机理，通过对驾驶员行为、行车状态、道路环境等多源信息进行获取，在线辨识弯道行车时发生侧向失稳事件的危险态势。特别地，将过弯车速的合理选择作为降低侧向失稳风险的有效手段，提出了在传统以车辆、道路信息为主进行弯道车速建模的基础上，建立了考虑驾驶员行为特性的弯道车速预测模型。同时，面向弯道行车的主动控制，建立了模仿人类驾驶行为特性的驾驶员模型，实现了弯道行车的仿人自主决策控制。此外，研究团队进一步探究了多车协同驾驶控制技术，充分利用道路条件，实现了若干单车之间的协同控制与驾驶，使其不仅具有组队行驶速度快，间距小等特点，还能增加交通的安全性和可组织性。研究成果对于发展和推广智能网联汽车的应用，提高交通安全，缓解交通拥堵具有重要意义。.研究期间，共发表学术期刊论文18篇，其中SCI收录论文11篇，会议论文8篇，授权发明专利11项，获天津市科技进步二等奖，毕业研究生共13名，其中博士生2名，硕士生11名，完成任务要求。.