Abstract: A new method for the moving safety state recognition based on computational fluid dynamics (CFD) was proposed to make driver assistance system simultaneously suitable for vehicle-road and vehicle-vehicle two traffic conditions. In the driving safety zone enclosed by the lane line and the flow field around the vehicle, two indexes characterizing the vehicle-road safety states, the vehicle location and the time-to-line-collision (TLC), were extracted according to the fitting lane line and the predicted vehicle driving trajectory. Other two indexes characterizing the vehicle-vehicle safety states, the maximums of longitudinal and lateral airflow velocity on the trajectory, were obtained by the offline calculation of typical models and the online real-time retrieval of the database. The safety states characterized by the four indexes of moving vehicles were classified and recognized by the probabilistic neural network (PNN). The results show that this method is effective in recognising vehicle driving states.

Key words: vehicle engineering, driver assistance system, safety state, CFD, PNN

摘要： 为了使驾驶辅助系统同时适用于车-路和车-车两种交通路况，基于计算流体动力学提出了一种车辆行驶安全状态识别的新方法。在车道线和车辆外流场所围成的行驶安全域中，根据拟合的车道线和预测的车辆行驶轨迹，提取车辆的位置和撞线时间两项车-路安全状态的特征；并通过车辆外流场的建模离线计算和数据库的实时在线检索，获取行驶轨迹上纵、横向最大流速两项车-车安全状态的特征。采用概率神经网络对行驶车辆的这4项特征进行安全模式的分类和识别。结果表明:该方法能够比较满意地识别出车辆的行驶状态。

关键词: 车辆工程, 驾驶辅助系统, 安全状态, 计算流体动力学, 概率神经网络