Abstract: Aiming at the disadvantage of linear congruential method widely used in traffic flow simulation, MT random number generator is adopted after performance comparison. The traffic conditions is divided into the idle state and the busy one, and different stochastic distributions are applied in stochastic vehicle generation models based on random numbers generated by MT random number generator. The traffic flow procedures are developed in VC environment to achieve simulation experiment, then the results are compared with the field data, which proves the advantage of stochastic vehicle generation models based on MT random number generator.

Key words: traffic flow simulation, random number generator, stochastic vehicle generation model, stochastic distribution

摘要： 针对交通流仿真技术中广泛使用的线性同余发生器在周期长度和随机性方面的不足，引入MT随机数发生器，通过性能评价验证其优越性，然后在空闲和拥挤两种交通状态下，基于MT发生器产生的随机数，应用不同的随机分布构建相应的随机发车模型。在VC环境下开发车辆生成程序进行仿真实验，将仿真数据与实测数据进行对比分析，结果表明，基于MT发生器的随机发车模型更好地逼近了实际交通状态。

关键词: 交通流仿真, 随机数发生器, 车辆随机生成模型, 随机分布