城市区域交通智能分散控制研究  

Study on intelligent decentralized control for urban region traffic

作　　者：沈国江[1] 

机构地区：[1]浙江大学工业控制技术国家重点实验室,浙江杭州310027

出　　处：《浙江大学学报（工学版）》2006年第4期585-589,共5页Journal of Zhejiang University：Engineering Science

基　　金：国家自然科学基金资助项目(NSFC-60304018);杭州市科技创新计划资助项目(20051331B02)

摘　　要：针对城市区域交通非线性、不确定性和模糊性特点,提出了一种新颖的实时智能分散控制策略.把整个城市区域交通作为一个大系统,区域中的各交叉口作为子系统,在每个交叉口设置一个独立的控制器,该控制器根据自己和相邻交叉口的交通流信息对交叉口的相序、相位切换、信号周期和绿信比进行动态优化.每个控制器有3个模块组成:相序优化模块、绿灯判断模块和相位切换模块.对每个控制模块设计了相应的模糊优化控制算法,并用改进的BP神经网络实现算法的模糊关系.控制目标是保持区域内各交叉口前的交通畅通和车辆延误最小.仿真研究表明,在交通流量较大和流量时变的环境下,智能分散控制方法比普通单交叉口车辆感应控制方法的控制效果更好,实用性更强.A novel real-time intelligent decentralized control strategy applicable to the nonlinear, uncertain, fuzzy system of urban region traffic was presented. The whole urban region traffic was regarded as a large scale system and the subsystem was every single intersection in the region. Each intersection had its own traffic controller which managed the phase sequence, phase switch, cycle time and splits dynamically according to its own and its neighbor＇s traffic situations. The controller consisted of three modules that were the phase sequence optimizing module, the green-phase judging module and the phase switching module. Fuzzy optimal control arithmetic was designed for each module, and an improved BP neural network was introduced to implement the fuzzy relation. The method aimed at making the intersections unblocked and making the average vehicle delay time shortest. The results of simulation show that this method has better performances in the cases of time varying traffic patterns and heavy traffic conditions than the vehicle actuated method. This intelligent decentralized control strategy can be used for urban traffic control.

关 键 词：控制理论与工程 城市区域 模糊控制 神经网络 分散控制 

分 类 号：TP393.04[自动化与计算机技术—计算机应用技术] TN915.04[自动化与计算机技术—计算机科学与技术]