Study On Discretionary Lane-Changing Behavior On Urban Streets

In recent years,with the increase of automobile inventory,traffic congestion has become increasingly prominent.As one of the most common driving behaviors,lane-changing behavior not only has a negative impact on the operational efficiency of traffic flow,but also poses a potential threat to the traffic safety,especially in peak periods or bottlenecks.The research on the driver's lane changing plays an important role in the research field of traffic flow and road traffic safety.At present,most lane-changing models are established based on road traffic conditions in foreign countries,and mainly for highways.Taking into account the large number of vehicles and the complex traffic conditions on the road in our country,these models may not apply.Therefore,it is very necessary to study lane-changing models suitable for China's traffic situation,especially for urban roads.This study takes the video of vehicles on a section of the Southwest Road of Dalian as the raw data.A large number of vehicle trajectory data was extracted from the video.Based on the data,the driving speed,space headway and duration time of lane-changing were analyzed.Then,based on the measured data,the traditional binary logit lane-changing model was established.The results show that the relative speed of the vehicle and its leading vehicle on the current/target lane,the relative speed of the vehicle and its following vehicle on the target lane,the space headway between the vehicle and its leading vehicle on the current lane,the space headway between the leading and following vehicles on the target lane,and the existence of a bus in front of the vehicle have significant impact on the decision of changing lane.Taking into account the different sensitivity of the drivers to factors affecting lane-changing caused by the individual differences of drivers,the mixed logit lane-changing model considering drivers' heterogeneity was established on the basis of the binary logit lane-changing model.The differences of drivers' decision on lane-changing were also explored.The results show that compared with the traditional logit lane-change model,the mixed logit lane-changing model has better fitting effect.The drivers appear significant heterogeneity on the relative speed of the vehicle and its leading vehicle on the current/target lane,the space headway between the leading and following vehicles on the target lane during the lane-changing.In addition,the good applicability of the mixed logit lane-changing model to the new data was validated using the measured data.The questionnaire was used to investigate the driver's willingness of changing lane.Based on the survey data,the K-means clustering method was applied to classify the drivers into three types: conservative,neutral and aggressive.Also,the distributions of drivers' driving style with different attributes such as gender,age,monthly income,numbers and mileage of weekly driving were analyzed.Finally,the specific driver attributes were incorporated into the binary logit lane-changing model.The discretionary lane-changing model considering driver factors was established.The result shows that the driver's monthly income,the numbers and mileage of weekly driving have significant impact on the decision of changing lane.