Vehicle dangerous behaviour warning plays an important role to improve road traffic safety and efficiency, so a safety assessment method of vehicle behaviour based on the improved Dempster–Shafer (D–S) evidence theory is proposed. Firstly, through analysis of vehicle collision accident mechanism, some factors closely related to vehicle safety are extracted. Also, multiple sensors are synthetically utilised to collect information, which realises the awareness of vehicle state, road attribute, driving environment etc. Then vehicle behaviour identification is accomplished according to the parameter information of the vehicle-mounted sensors, as well as the related data of adjacent vehicles in vehicular ad hoc networks (VANET). Finally, a sequential type of weighted correction method based on evidence variance is used to integrate different levels of multi-source heterogeneous information and to achieve safety assessment of vehicle behaviour. The experimental results show that the improved D–S evidence theory reduces the evidence conflict, increasing the accuracy and reliability of vehicle behaviour safety assessment. The study solves the fundamental core problem of active safety warning in VANET and provides a new means of traffic accident warning for the road traffic management department.

中文翻译：

---

基于改进的D–S证据理论的车辆行为安全性评估

车辆危险行为预警在提高道路交通安全性和效率方面起着重要作用，因此提出了一种基于改进的Dempster-Shafer（DS）证据理论的车辆行为安全性评估方法。首先，通过对车辆碰撞事故机理的分析，提取出与车辆安全性密切相关的一些因素。另外，通过综合利用多个传感器来收集信息，从而实现对车辆状态，道路属性，驾驶环境等的认识。然后根据车载传感器的参数信息以及相关数据完成车辆行为识别。车辆中相邻车辆的数量特别指定网络（VANET）。最后，基于证据方差的序列类型加权校正方法用于整合不同级别的多源异构信息并实现车辆行为的安全性评估。实验结果表明，改进的D–S证据理论减少了证据冲突，提高了车辆行为安全评估的准确性和可靠性。该研究解决了VANET中主动安全预警的根本核心问题，为道路交通管理部门提供了一种新的交通事故预警手段。