Evaluation And Modeling Of Operational Safety Of Signalized Intersection Under E-bike Violation Behaviors

With the continuous advancement of urbanization,the problem of traffic congestion in many cities in China is becoming more and more serious.Facing the problem of traffic congestion,electric bicycles are favored by urban residents because of their flexibility,economy and convenience.While electric bicycle brings convenience to urban residents,it also brings new potential safety hazards.The characteristics of fast acceleration,flexibility and compactness make electric bicycles a more unstable safety factor on urban roads,and e-bike violation behaviors have a great negative impact on signal crossings.Therefore,in order to grasp the relationship between electric bicycle violations and traffic operation environment,improve the safety of signalized intersections,,and help traffic management departments improve the road traffic environment,it is necessary to conduct in-depth research on the safety of signalized intersections under electric e-bike violations.Firstly,this paper selects typical signalized intersections in Beijing for video filming survey,analyzes the difference of the occurrence rate of e-bike violations,and models e-bike violations in multiple categories based on a multivariate logistic regression model.The results show that e-bike violations at signalized intersections are mainly running red lights,occupying motor vehicle lanes,reverse riding and waiting across lanes.The violation rates of electric bicycles of different non motor vehicle riders and different age groups are significantly different.14 factors such as peak hours,isolation mode between motor vehicles and non motor vehicles and the width of non motor lanes have a significant impact on the multi category violations of electric bicycle riders.Secondly,the violation rate function of electric bicycles is constructed according to the survival analysis method,and the violation rate product limit estimation method and the violation rate model based on Cox proportional risk regression of electric bicycles are proposed to estimate the violation rate of electric bicycle under the influence of single factor and multi factor respectively.The estimation results show that there is no significant difference in the violation rate of electric bicycles riders of different gender and age,and the violation rate of e-bike is significantly higher than that of conventional bicycle.It is estimated that the violation rate of electric bicycle riders under the joint influence of six significant factors,such as red light duration,non motorized lane width and motor vehicle traffic volume.Then,the traffic conflicts of illegal riding electric e-bikes are defined and identified,and the severity of traffic conflict is classified by k-means clustering based on post-intrusion time PET,traffic conflict time TTC and safety deceleration speed DST indicators.The traffic conflict prediction model based on Generalized Linear Model(GLM)and Generalized Linear Mixed Model(GLMM)for illegal riding e-bike is further constructed,and the traffic conflicts of different severity are predicted.The results show that the proportion of general conflict is the highest,the division of traffic conflict severity is in line with the actual situation,the prediction effect of GLMM is better than GLM,and the prediction result of general conflict frequency is the best.Finally,the evaluation index system is established considering the signal intersection characteristics,violation behaviors,traffic conflicts and other factors,and the safety evaluation model of signalized intersections under electric bicycle violations based on entropy weight TOPSIS method is constructed,and the empirical study is conducted from the aspects of signal intersection and the inlet lane.Based on the evaluation results,aiming at the safety of signalized intersection,some improvement measures are proposed,such as the installation of additional wardens,special left-turn phases,physical machine-office separation and widening of non-motorized lanes.The research results can provide support for urban management to develop signal intersection management measures and optimization strategies.There are 20 figures,30 tables,58 references in this paper.