Prediction Of The Severity Of Multiple Conflicts And Risk Assessment At Non-signalized Controlled Crosswalk

In recent years,the number of motor vehicles and electric bicycles in China has been continuously increasing,and the environment for pedestrians and electric bicycles to cross the street has become increasingly severe.In order to improve the safety of crossing the street,China advocates for motor vehicles to give way,but it also increases the complexity of the crossing environment while protecting it.There is a common situation where motor vehicles conflict with another motor vehicle when pedestrians or electric bicycles cross the street.This article is based on the theory of traffic conflict technology and combined with survey data to study the conflict of crossing a pedestrian crossing without signal control,and evaluate and improve its safety.Reduced the number of traffic conflicts,making motor vehicles more inclined to give way to pedestrians,and improving the safety of pedestrians and cyclists crossing the street.Firstly,taking pedestrians and electric bicycles crossing the street as research objects,the characteristics of pedestrians crossing the street,electric bicycles,and the yielding characteristics of motor vehicles are studied.Based on the analysis,this article proposes multiple threat conflicts and clarifies their definition and discrimination methods,distinguishing them from single conflicts.To compare the two types of conflicts,the severity of the conflict is quantified through conflict indicators.However,considering the one-sidedness and applicability of individual conflict indicators,this article proposes a motor vehicle conflict indicator(TTZ_veh)that considers the impact of driver visual impairments,and combines the post intrusion time（PET）and safety deceleration（DST）conflict indicators to quantify the severity of traffic conflicts.Two unmanned pedestrian crossings in Zhenjiang City were recorded using drones for a total of 5 hours of video.Adobe software was used to extract the data and calculate the values of TTZ_veh,PET and DST conflict indicators.Secondly,based on the actual data of pedestrian crossings without signal control,455 sets of conflict numbers were extracted.Based on the values of three conflict indicators,a severity level classification model for two types of traffic conflicts was established using the fuzzy C-means clustering algorithm.The conflicts were divided into severe and non severe,and the partition values were solved to identify the severe and non severe conflicts of the two types of conflicts,Analyzed the severity of multiple threat conflicts and single conflicts.The results show that for the severity of conflicts,multiple threat conflicts are higher than single conflicts,and the severity of conflicts of the same type is higher for electric bicycles than for pedestrians.Then,by analyzing the behavior characteristics of road users,nine explanatory variables related to conflict were extracted,and the correlation and collinearity among explanatory variables were tested.The severity of the conflict is used as the dependent variable,with severity being 1 and non severity being 0.Establish prediction models for the severity of multiple threat conflicts and single conflict,and construct two Markov chains based on Win Bugs software to estimate model parameters.Determine significant influencing factors through 90%and 95%confidence intervals.The results indicate that the legal parking behavior of vehicles,crossing speed,motor vehicle speed,waiting time for crossing,crossing direction,and crossing flow at the time of conflict have a significant impact on the severity of the conflict.Finally,a risk assessment model for pedestrian crossings without signal control was established based on the probability and potential severity of multiple threat conflicts.Based on the environment where multiple threat conflicts occur,a conflict probability model was established and a potential collision severity index was extracted to calculate the conflict risk values of different crossing objects during the morning peak,flat peak,and evening peak periods.The fuzzy C-means clustering algorithm was used to determine the level of cross street safety risk,and the evaluation results were analyzed based on the current situation,verifying the effectiveness of the model.Based on the research results of this article,it can reduce the incidence of pedestrian crossing conflicts without signal control,improve the safety and comfort of pedestrians and electric bicycles crossing the street,and provide theoretical reference for improving the yield rate of motor vehicles.It helps road participants to comply with laws and regulations,and facilitates traffic managers to manage and implement related traffic issues.