基于路口相关性的交通流量修复研究

1)北京工业大学北京市交通工程重点实验室,北京 100124; 2)中国公路工程咨询集团有限公司,北京 100124

交通运输工程; 绿波协调; 路口相关性; 主成分分析; 径向基函数神经网络; 交通流量修复

Restoration of traffic flow data based on intersection correlation
YU Quan1, LIU Yang1, and GUO Xiaowei2

1)Beijing Key Laboratory of Traffic Engineering, Beijing University of Technology, Beijing 100124, P.R.China2)China Highway Engineering Consulting Corporation, Beijing 100124, P.R.China

transport engineering; green wave coordinate; intersection correlation; principal component analysis(PCA); radical basis function neural network; traffic flow restoration

DOI: 10.3724/SP.J.1249.2019.03304

备注

为实现对交通流异常数据的有效修复,根据流量-占有率函数模型,利用主成分分析法对绿波协调控制交叉口群中各路口的占有率参数进行相关性分析,间接得到各路口流量相关性的大小,构建相关路口集; 根据相关路口集的历史数据,分别利用流量-占有率模型、径向基函数神经网络模型以及基于两种方法的组合模型,对目标路口的缺失交通流数据进行修复; 通过实例分析验证模型性能.结果表明,与单独流量-占有率模型或径向基函数神经网络模型相比,组合模型可更精确地修复交通数据,在实际数据验证中表现出更好的适应性.

In order to restore the fault data of traffic flow effectively, according to the flow-occupancy function model, we use the principal component analysis(PCA)method to analyze the correlation coefficient of occupancy of each intersection in signalized intersections under green wave coordinated control, and obtain the correlation of traffic flow data at each intersection indirectly. Finally, we construct the relevant intersection set. According to the historical data of the relevant intersection set, we restore the missing traffic flow data at the target intersection by using the traffic-occupancy model, the radical basis function(RBF)neural network model, and the combination model based on the two models, respectively. The performances of the three models are verified on the basis of case analysis. The results show that the combined model can be used to recover traffic data more accurately than the other two models and shows better adaptability in actual data validation.

·