中文摘要：

利用递阶结构和模糊神经网络来进行交通系统的实时协调控制。其基本思想是把交通干线作为一个大系统。子系统为干线上的各交叉口，用模糊神经网络综合调控绿信比，相位差以及周期时长等3方面因素，在模糊神经网络控制器的设计中使用4层网络，将3种数据模糊化后输入，进而得到相应的输出结果，目的在于更好的协调交叉口间的信号，使得主干路的排队长度最小，从而减少车辆延误。最后将设计的控制器作用于北京市海淀区学院路与成府路和清华东路所构成的系统进行仿真研究，结果显示，该方法在减少主干道车辆延误的同时，综合考虑了次干道路段的需求，使得在减少车辆延误方面有较为显著的效果。

英文摘要：

By using hierarchical structure and fuzzy neural network, real-time coordination control of traffic system was performed. Its main idea is that regarding the main roads as the large-scale multi-objective programming system. Its sub-problems are the intersections of the arterial roads. And the fuzzy neural network synthetically takes green signal ratio, phase difference and＇ cycle length into account, while using four-tier networks in the design of its controller, then inputting three kinds of fuzzy data to obtain the corresponding result. The purpose is to coordinate control signals of the intersections of urban arterial roads, and to make the queue length of the arterial road shorter, so as to reduce vehicle delay. Finally, the simulation study for the system composed of Xueyuan Road, Chengfu Road, and Qinghua East Road, Haidian District in Beijing was performed. Its result shows that the purposed method is more effective in reducing vehicle delay due to consideration of both reducing the vehicle delay on the main roads and the demands of the sub-streets.