Study Of Free Lane Change Behavior In Merging Zone Of Expressway Based On Traffic Flow Characteristics Analysis

Urban expressway is an important component of the urban transportation system,which plays a vital role in the fast and convenient passage of medium and long distances between inner city areas.The merge zone is a notable section of the expressway,which has a high risk of traffic accidents due to the intertwined traffic flow and frequent vehicle lane changes.Most of the existing expressway merge zone studies use foreign data or data from domestic megacities and megalopoleis.However,megacities and megalopoleis only account for 20% of domestic cities with demand for expressway construction,and the remaining 80% are Type I and Type II large cities.There are certain differences in urban traffic flow data of different scales.Therefore,this paper studies the lane change behavior and its risk in the expressway merging area of Type I and II large cities represented by Nanchang City,which can be used to guide the safe lane change of vehicles in this type of merging area and improve regional traffic safety.The main content of the paper is divided into the following aspects: Firstly,the data investigation area was selected,and the merging area of Beijing West Road on Hongdu Expressway in Nanchang City was photographed by an unmanned aerial vehicle at a high altitude to obtain the traffic flow video data.The vehicle trajectory was extracted by Tracker video analysis software.Then the exponential moving average method was applied to denoising and smoothing.The purpose of this step was to achieve the high-resolution extraction and processing of the vehicle trajectory data and analyze the merging area traffic flow characteristics.Secondly,the lane change types in the merging zone of the expressway were distinguished according to the purpose of lane change,and the object of study was determined to be the free lane change behavior with general regularity.The free lane change phase was divided by identifying the feature points of the lane change phase.Lane-change samples were constructed for extracting the spatial and temporal feature information of the lane-change process.The vehicle lane-change pattern was analyzed from lane change duration and lane change spatial characteristics.Then,the data preprocessing of the characteristic variables related to the free lane-changing behavior of vehicles was performed by principal component analysis,and the combined parameters were optimized by using single-parameter optimization and grid search algorithms.The XGboost multi-classification model was built for free lane-changing phase classification,which provided data support for the subsequent lane-changing risk research.Finally,the risk identification indexes of lane change behavior in both space and time were selected as the evaluation indexes of traffic conflicts based on the data which used the XGboost model to identify the free lane change phases of vehicles.The risk identification method of lane change behavior was constructed to distinguish the degree of vehicle lane change risk which was based on the CRITIC empowerment method and matter-element extension model.Early warning measures such as roadside information were proposed after identifying the road sections with a high risk of free lane change behavior which was based on the risk identification results of the merging zone.And this research can provide a reference for the design of a vehicle-road cooperation system assisted by vehicle lane change and the formulation of safety control measures in the merging zone.The results of the study found that: First,the type I and II large cities represented by Nanchang City have smaller traffic flow rates,a higher proportion of large vehicles,faster lane speed,and larger headway than the expressway merge area of megacities and megalopoleis represented by Nanjing City;Secondly,the free lane change behavior of vehicles in the expressway merge area of type I and II large cities is easily influenced by the traffic flow condition,and its trend is increasing first and then decreasing with the increase of traffic volume or traffic density.The average duration in the probability distribution is in line with the Weibull distribution,and the longitudinal spatial distribution is concentrated in the merging section and its downstream;Thirdly,based on the example of the risk identification method in this paper,after studying the spatial distribution of the free lane change behavior with higher risk degree,it is found that the highest frequency distribution is from 60 m before the downstream section to 10 m after the downstream section on the outermost lane of the merging area.