ATIS条件下驾驶员出行途中路径选择行为研究

1)深圳市综合交通运行指挥中心,广东深圳 518041; 2)东南大学交通学院,江苏南京 210096; 3)深圳市城市交通规划研究中心,广东深圳 518021; 4)宁波大学海运学院,浙江宁波 315211

城市交通; 多源信息参考; BP神经网络; 自变量筛选; 路径选择; 多元logit模型

Drivers' route choice behavior analysis under ATIS
Li Chunyan1, Chen Jun2, Sun Zheng'an3, and Ye Xiaofei4

1)Shenzhen Transportation Operation Command Center, Shenzhen 518041, Guangdong Province, P.R.China2)School of Transportation, Southeast University, Nanjing 210096, Jiangsu Province, P.R.China3)Shenzhen Urban Transport Planning Center, Shenzhen 518021, Gua

urban traffic; multi-source traffic information reference; BP neural theory; influence factor selection; route choice; binary logit model

DOI: 10.3724/SP.J.1249.2016.02164

备注

在假设驾驶人对不同来源交通信息的参考程度不同的情况下,研究驾驶人在先进的出行者交通信息服务系统(advanced traveler information systems, ATIS)提供的多源实时信息条件下的出行途中路径选择行为,针对传统logit模型不能模拟影响变量对目标变量的非线性影响的缺点,利用BP(back propagation)神经网络模型进行信息参考行为的影响因素筛选,通过二元logit模型分析交通信息的参考概率,且作为自变量建立驾驶人出行途中路径选择模型.以南京驾驶人在广播和可变信息板(variable message signs, VMS)两种信息发布源条件下的出行路径选择行为进行模型验证,发现年龄和学历是影响广播信息参考的主要因素,年收入和对本地道路的熟悉程度是影响VMS信息参考的主要因素,驾驶人在接收到外界信息时改变出行路径的概率较大,在未接收到外界信息时选择原路径概率较大.

To analyze drivers' en-route route choice behavior under multi-source traffic information provided by advanced traveler information systems(ATIS), this paper assumes that believed drivers' reference behavior of different sources for information is different. To overcome the shortcoming of the logit model which cannot simulate the nonlinear effects of the independent variables on the target variables, influence factors were first selected using BP neural theory. Then the traffic information reference probability was analyzed using binary logit model, and the en-route route choice model was founded. To verify the effectiveness of the model, drivers' en-route route choice behavior under broadcast and variable message signs(VMS)traffic information condition in Nanjing was taken for instance. We find that age and education are the main factors for broadcast information while income and familiarity to Nanjing road network are the main factors for VMS information. The probability of changing route is much bigger when drivers accept travel information.

·