Abstract: The moving characteristics and rule of vehicle group are studied in the virtual condition，which includes the framework of the software and hardware，the experiment process，the data collection，analysis and calculation，as well as the data reliability analysis. In the experimentation，15 drivers simulate driving in one scene，meanwhile，the moving parameters of vehicles are collected respectively，including the running track，all kinds of velocity and acceleration，and the use of brake and steering wheel of the vehicles etc. Fuzzy neural network and risk probability matrix are used to analyze the data mentioned，and the risk probability of the vehicle group is calculated; furthermore，the method to study behavior safety of vehicle group is proposed.

Key words: virtual driving, vehicle group risk, fuzzy neural network, risk probability matrix

摘要： 利用虚拟化平台研究车群的运动特征规律。内容包括软硬件平台结构，试验过程，数据采集，分析计算和数 据可靠性分析等几个部分。试验中由15个驾驶员在同一场景中进行模拟驾驶，同时采集各自车辆的运动参数，包 括车辆的行驶轨迹、各种速度加速度、刹车使用情况以及方向盘使用情况等。运用模糊神经网络及风险概率矩阵 分析上述数据，计算出车群发生风险的情况概率，提出了一种研究车群行为安全的方法。

关键词: 虚拟驾驶, 车群风险, 模糊神经网络, 风险概率矩阵