Measurement And Risk Analysis Of Pedestrian-traffic Conflicts At Signalized Intersections

In recent years,with China’s economic development and technological progress,the traditional rural society dominated by agriculture has gradually transformed into a modern urban society dominated by non-agriculture.With the continuous development of urbanization in China,the scale of urban population and industry continues to expand,and the pressure of urban transportation also continues to expand,especially the location of intersections.In particular,as an important part of urban roads,signalized intersection plays an important role in connecting roads,but it is also the traffic bottleneck of the whole urban road network.Intersections often gather traffic from two or more directions.Most of them are mixed traffic flow of motor vehicles,non-motor vehicles and pedestrians.As a vulnerable group,pedestrians are often more vulnerable.Through the analysis of traffic safety technology,it can be seen that due to the scarcity of traffic accident samples and the limitations of accident occurrence,traffic conflict technology can be selected to analyze the data of human vehicle traffic conflict at the signal intersection,evaluate the traffic safety of the intersection,put forward the good faith view of the intersection,and improve the safety of pedestrian crossing.In this paper,based on the basic principle of traffic conflict technology,video recording method is used to obtain on-site video data,and physical tracking software tracker is used to calibrate the benchmarking pole and set the coordinate axis,select the mass points,track manually or automatically the running track of the mass points in each frame,obtain the basic data of coordinate changes of motor vehicles,and then realize the extraction of micro data of traffic conflict.In this study,a total of 400 four-dimensional samples are obtained,each of which includes conflict time,conflict speed,maximum acceleration,vehicle steering and other parameters.All samples are deeply studied by MATLAB software,and cluster analysis is carried out.Through classification and dimensionality reduction,these samples can be divided into four categories,representing the four levels of traffic conflict severity.Based on the results of deep learning,the safety evaluation of signalized intersection based on the severity of human vehicle traffic conflict is quantified.In addition,this study establishes a risk analysis and evaluation system based on 17indicators of four dimensions of people,vehicles,roads and management,and analyzes the weight of each indicator using AHP technology.Combined with the risk evaluation index system,based on the traffic conflict data and the coupling degree theory,a coupling model of urban intersection risk factors is constructed to analyze the coupling effect among the risk components of urban intersection,and a conflict severity model is proposed to evaluate the coupling effect of multiple factors of urban intersection risk.The output of the coupling model is beneficial to the improvement of intersections.In this study,the micro data of traffic conflict is extracted based on trajectory tracking technology,and in-depth learning is carried out to identify the types of traffic conflict severity,initially quantify the safety level of the intersection,and assess the risk level of the intersection.In addition,according to the sample data of traffic conflict,this study establishes the risk analysis and evaluation system of signal intersection based on AHP and coupling degree model,and analyzes the coupling effect of various risk factors.This research can provide a new try angle for traditional safety evaluation methods and a new direction for intersection improvement methods.