Data-driven Based Traffic Demand Pattern Exploration Of Urban Expressway

Expressway is an important part of the city’s multi-level comprehensive road traffic road network.With the intensification of urban congestion,the amount of traffic undertaken by expressways has increased significantly.In this context,mastering the spatial and temporal distribution of expressway traffic demand and serving real-time active management and control of expressway demand are important ways to improve the operational efficiency of urban road networks.In order to deeply analyze the spatial and temporal distribution of expressway traffic demand,this paper studies the expressway traffic demand mode based on data driving.The traffic demand pattern is divided into the commuting demand pattern and the noncommuting demand pattern.The key to distinguish the two patterns is to distinguish the commuting demand.Compared with non-commuting trips,commuting trips have relatively stable characteristics at a certain period of time,such as departure time,travel path,and the starting and ending points of travel.Existing commuting travel research is mostly oriented to traffic planning and traffic policy formulation.In order to guide the real-time active control of urban expressways,this paper proposes a data-driven based commuting travel identification framework for analyzing expressway traffic demand patterns.On this basis,considering that the existing traffic demand estimation is limited by the difficulty of obtaining individual vehicle travel feature,and the spatial and temporal distribution difference between commuting demand and non-commuting demand is not considered.With the emerging of traffic detection equipment,the spatial and temporal correlation characteristics of the traffic demand between the on-ramps and off-ramps under different demand patterns was analyzed,and based on this,the traffic flow distribution on the off-ramps is estimated using the information of the on-ramps traffic flow,First,based on the data of high-definition bayonet vehicle detectors and the characteristics of expressway,this paper proposes a commuting travel identification framework,which is consisted by three parts: commuting trip spatial and temporal feature extraction,commuting vehicle identification using Technique for Order Preference by Similarity to Ideal Solution(TOPSIS)and threshold method and traffic demand spatial and temporal pattern analysis.On this basis,a traffic demand estimation allocation matrix is constructed to estimate traffic demand different patterns.Finally,the Zhonghuan Expressway in Kunshan City,Jiangsu Province was used as the test site for case study and performance evaluation.The case study results show that in the time dimension,the peak hours of commuting trips captured by the proposed method are consistent with the actual peak time of the road network;in the spatial dimension,there is a clear correspondence between the commonly used on ramps and the off ramp,that is,the identified commuters often use the on-ramps that departs in the morning is close to the commonly used off-ramps that returns in the afternoon.The study verified that the proposed commuter vehicle identification framework can reasonably identify commuter vehicles.In addition,through the analysis of the results of the demand estimation,it is found that the error of the traffic demand estimation can be decreased to a certain extent considering different traffic patterns.