Abstract: With the defects of the Biham-Middleton-Levine (BML) model when analyzing the effect from traffic accident in the road network, an improved BML model is proposed. In which, the car driving behavior of changing direction under traffic accident is considered. With this, the car driving rule is more analogous to the actual car action. On the basis of the improved model, the effect of accident amount to average speed is analyzed when the density of traffic flow is given. Then, the effect of traffic flow density to average speed is analyzed when the amount of accident is determined. Furthermore, the effect of the turn probability of car before accident to average speed is estimated. Finally, the relationship between the turn probability distribution function and the average speed of traffic flow is investigated, and some conclusions are drawn with simulation.

Key words: triffic engineering, cellular automation (CA), traffic accident, turning probability, Biham-Middleton-Levine (BML) model

摘要： 针对BML模型在分析交通事故对路网交通运行状态影响方面的不足，考虑事故发生时车辆为回避阻塞点而改变行驶路径等行为，对BML模型中事故点前车辆的行驶规则进行改进，使之与实际车辆的驾驶行为更加接近. 在改进模型的基础上，分别就路网交通流密度一定时，事故点数量变化对交通流平均速度的影响；事故点数量一定时，路网交通流密度变化对交通流平均速度的影响；事故点前各格点车辆转移概率对交通流平均速度的影响；车辆在事故点前各格点的转移概率分布函数不同对系统运行状态的影响等四个问题进行模拟研究，通过对模拟结果进行分析，得出了一系列有意义的结论.

关键词: 交通工程, 元胞自动机, 交通事故, 转移概率, BML模型