Abstract: In the light of the urban traffic flow characteristics，the traffic status of an isolated intersection is divided into four levels，and a real-time traffic signal control method is proposed on the basis of this traffic status identification.Based on the traffic flow characteristics and control targets of the isolated intersection，this method can choose appropriate evaluation parameters and establish signal timing models for different traffic levels.Simultaneously，an improved genetic algorithm is presented to solve the timing models.This algorithm uses the penalty mechanism of sorting and classifying to handle constraints，and introduces simulated annealing algorithm to enhance the local optimal capability of genetic algorithm.Lastly，the traffic signal timing optimization case of an actual isolated intersection is extensively computed by three kinds of algorithm，and simulated via Paramics.The results indicate that the improved genetic algorithm is effective and the signal strategy derived from the proposed method has better performance.

Key words: traffic signal control, traffic status, optimization model, genetic algorithm, Paramics simulation

摘要： 针对城市道路单点交叉口交通流的到达特性，将路口到达的交通流划分为4种状态，提出了“基于状态划分的多相位交通信号实时控制方法”，该方法根据路口各状态下交通流的到达特征和控制目标，为不同交通状态选择合适的性能指标，并建立各状态下交通信号的动态配时模型。同时，设计了一种改进的自适应实数编码遗传算法对交通信号配时模型进行求解，该算法采用基于分类的排序惩罚机制对约束进行处理，并引入模拟退火算子增强遗传算法的局部搜索。最后，采用3种算法对实例进行大量的数值计算和Paramics仿真，计算和仿真结果均表明所设计的算法求解精度高且模型具有良好的控制效果。

关键词: 交通信号控制, 交通状态, 优化模型, 遗传算法, Paramics仿真