基于改进FCM聚类的交通控制时段划分  

Division of Traffic Control Periods Based on Improved FCM Clustering

作　　者：于德新[1,2] 田秀娟[1] 杨兆升[1,2] YU De-xin TIAN Xiu-juan YANG Zhao-sheng(School of Transportation, Jilin University, Changchun 130022, Jilin, China Jilin Province Key Laboratory of Road Traffic, Changchun 130022, Jilin, China)

机构地区：[1]吉林大学交通学院,吉林长春130022 [2]吉林省道路交通重点实验室,吉林长春130022

出　　处：《华南理工大学学报（自然科学版）》2016年第12期53-60,共8页Journal of South China University of Technology(Natural Science Edition)

基　　金：国家科技支撑计划项目(2014BAG03B03)~~

摘　　要：对传统的模糊c-均值聚类算法进行改进,提出一种基于改进FCM聚类的交通信号控制时段划分方法.首先,引入模糊聚类隶属度基数,对聚类数目自动选取;然后,运用模拟退火遗传混合算法对初始聚类中心进行优化.最后,根据交叉口实际流量数据,进行时段划分,利用仿真软件进行方案效果评价.结果表明,与传统FCM算法相比,文中方法能有效实现控制时段划分,更加符合实际交通特性,且能得到全局最优解.与原有控制方案相比,FCM方案和文中方案都能有效降低车辆平均延误,文中方案效果更明显.In this paper, the traditional fuzzy c-means clustering （FCM） algorithm is improved, and a method to divide the traffic signal control periods is proposed based on the improved FCM algorithm. In the method, first, a cardinal number of fuzzy clustering membership degree is introduced to automatically select the cluster number. Then, the hybrid simulated annealing genetic algorithm is employed to optimize the initial clustering center. Final- ly, the traffic control periods are divided according to the actual traffic flow data, and the performance of the schemes is evaluated by using the simulation software. The results show that, as compared with the traditional FCM algorithm, the proposed method can divide the traffic control periods more effectively and reflect the actual traffic characteristics more accurately, and it achieves a global optimal solution. In addition, in comparison with the origi- nal signal control scheme, although both the scheme based on the FCM algorithm and the proposed scheme can re- duce the average vehicle delay, the proposed scheme has a more obvious effect.

关 键 词：交通控制 TOD控制 时段划分 FCM聚类 模拟退火遗传算法 

分 类 号：U491[交通运输工程—交通运输规划与管理]