Traffic Oscilations Evolution Mechanisim And Driver’s Carfollowing Behavior Characteristics Analysis Based On Trajectory Data

Traffic oscillations(also known as ‘stop-and-go’ traffic phenomenon),are the group performance of individual driving behavior,which refer to the conditions that traffic waves propagate to the upstream in the vehicles’ deceleration or acceleration process,often followed by traffic congestions in urban expressways.Deep research on the mechanism of the traffic oscillations is conducive to formulating traffic congestion control methods and improving driving safety.Due to the lack of micro data support,the interrelationship between the traffic oscillations evolution mechanism and micro car-following behavior has not been clarified.Therefore,based on the vehicle trajectory data extracted from the real road traffic environment by video image recognition technology,this study made a systematic and quantitative research on the traffic oscillations evolution mechanism and drivers’ car-following behavior characteristics.The main contents of this thesis are as follows:Firstly,the vehicle trajectory data cleaning and denoising algorithm was explored.This study selected US-101 data set and Nanjing Yingtian data set as databases.The causes of abnormal values and measurement errors in the data and the proportion of erroneous data were analyzed.A two-step data correction method was proposed: 1)the sudden change of vehicle velocity was reduced by first-order difference method and 2)the data noise was filtered by Kalman denoising method.The algorithm for data construction was proved the effectiveness by acceleration statistical distribution and jerk analysis.The reconstructed data was verified the smoothness and maintaining the data characteristics as well as structure of the original trajectory features.Secondly,the traffic oscillations evolution mechanism was analyzed based on highprecision trajectory data.The Mexican hat wavelet analysis technology was developed to identify formation and propagation characteristics(such as amplitude,duration and intensity)of traffic oscillations.Based on the comparative analysis of the two sets of trajectory data,the propagation characteristics of traffic oscillation waves between China and the United States were analyzed.The asymmetric driving behavior theory and the vehicles’ running phase transition law were employed for analyzing the evolution characteristics of generation,growth and dissipation stages of traffic oscillations.Thirdly,the drivers’ car-following series were extracted and car-following operating characteristics during traffic oscillations were analyzed.According to the determinant conditions of car-following behavior,the car-following series were extracted.Mathematical statistics analysis methods were employed to conduct systematic and quantitative analysis on the variation law of American and Chinese drivers’ car-following operating characteristics.The main characteristic parameters included velocity,acceleration,space headway,following distance,headway,time gap,acceleration/deceleration response time,etc.Finally,the drivers’ sense and response capability in the process of traffic oscillations was analyzed.The drivers’ sense and response parameters extraction method were improved,and the driving styles were classified into aggressive,timid and common types by K-means clustering algorithm.The change of drivers’ sense and response characteristics with different driving styles in the process of traffic oscillations were analyzed to explain the asymmetric driving behavior.The sense and response parameters were introduced into the car-following model to study how driver’s sense and response capability work on the stability of traffic flow.