Study On Decision-Making Behavior Of Vehicles’ Discretionary Lane Change On Urban Expressway

Discretionary lane change is a frequent and ubiquitous phenomenon in urban expressway.Discretionary lane change is more random and sudden than mandatory lane change,so it is is usually difficult to predict by the surrounding vehicles.Moreover,the duration of discretionary lane change is too short for the following vehicle in the destination lane to take safe and efficient countermeasures in time.This often brings serious interference to the traffic flow,may cause traffic jams even traffic accidents.Therefore,in order to alleviate the interference of discretionary lane changing on traffic flow and improve driving safety,this paper conducts study on the decision-making behavior of discretionary lane change from the aspects of driver cognitive psychology and micro behavior according to different driving groups.Firstly,the scenario of discretionary lane change is described,the motivation and conditions of discretionary lane change behavior are analyzed,and the influence of traffic conditions and driver characteristics on the decision-making behavior of discretionary lane change is discussed.On this basis,the vehicle motion state and driving style are taken as the decision variables of discretionary lane change modeling.Secondly,according to the research object of this paper,the data acquisition scheme is designed,and the vehicle motion data is collected by video detection method.Through Simi Motion platform,the motion states data of 940 groups of vehicles are obtained,and the data are cleaned by using multi-step filtering method and improved two-step reconstruction method based on symmetric exponential moving average method.On this basis,the statistical analysis of discretionary lane change duration is carried out,and the distribution of discretionary lane change duration is deduced.Thirdly,the full velocity difference model is applied to the discretionary lane change scenario,and the calculation method of the acceleration of merging vehicle and the following vehicle in the destination lane is proposed.On this basis,the relationship between the minimum safety spacing of lane change and the involved vehicles’ motion states as well as the lane change duration is analyzed by means of numerical simulation,and the relative velocity,spacing and lane change duration required for safe lane change under different conditions are obtained.In addition,the model of minimum safety spacing for lane keeping is established.Then,three characteristic variables that can represent driving style are determined.Based on spectral clustering algorithm,driving styles are classified into conservative,moderate and radical types.Furthermore,the differences of lane change durations among different driving styles are analyzed.The results show that the average lane change duration of conservative,moderate and radical driving styles decreases successively,but the difference between them is very small,which proves that the driving style has no significant influence on the lane change duration.Finally,based on cumulative prospect theory,decision-making models of discretionary lane change under different driving styles are established,which take the minimum safety spacings for lane change and for lane keeping as reference points and the acceleration spacing as utility.The parameter calibration and performance evaluation of the model are carried out,and it is found that the prediction accuracy of the model is 86.30%,which is higher than that of the traditional stochastic utility model.Furthermore,the influences of vehicles’ motion states on the decision-making behavior of discretionary lane change aer discussed.At the end,according to the research results of this paper,the conception of discretionary lane change warning system developing is proposed.