A Data-driven Approach To Safety Analysis Of Highway Tunnel Pavement Traffic Operation

Up to now,China has the world's fastest growing number of expressway tunnels,but also the world's largest number of expressway tunnel projcets,the most complex conditons of the country.The number of traffic accidents occurred in tunnels has also increased with the increase in the number of tunnels.At present,the research on the characteristics of tunnel traffic accidents and the prediction of accident rate of different pavement materials in China is slightly inadequate,the conmon characteristics of tunnel traffic accidents are limited to a single dimension,and the applicable conditions of tunnel accident rate prediction model are often conmplex,which can not provide a strong theoretical support for the safety operation managers of expressway tunnels.Therefore,this study relies on the collected expressway traffic accidents data,some tunnel linear data and traffic volume data of Shanxi Expressway for 2012-2018 years,and obtains the characteristics of tunnel accidents and the prediction model of tunnel accident rate under different conditions,which provides the working basis for the traffic safety operation of the expressway later.First,extensively collect traffic accidents data of expressway tunnels in Shanxi Province,linear data of accidents tunnels,traffic volume data and climate zoning data and so on.Based on different definitions of tunnel segments and traffic volume levels,the collected data are divided to different categories,doing the statistics and analysis of characteristics of 1)accidents type distribution,2)time distribution,3)spatial distribution,4)vehicles involving accidents under the conditions of 1)The whole Provice,2)different length types,3)different climate zones,4)different traffic volume levels.Then,Comparing and analyzing the characteristics of tunnel traffic accidents on asphalt pavement tunnels and cement pavement tunnels,in the mean time,Comparing and analyzing the safety of two kinds pavement material tunnel from 4 angles: 1)number of traffic accidents,2)rate of traffic accidents,3)severity of traffic accidents(number of deaths,number of injuries,number of vehicles involved,number of severe accidents),4)the degree of accident impact(duration of accidents),based on which,the influence factors of exprssway tunnel traffic accidents are analyzed from four aspects of man-car-tunnel pavement-tunnel environment.Third,SPSS software is used to analyze and get the regression of tunnel accidents rate under different pavement materials and different traffic volume levels and each individual index(the longest straight line segment in the tunnel,the longest flat curve in the tunnel,the shortest easing curve in the tunnel,the smallest radius of circle curve in the tunnel,the maximum longitudinal slope in the tunnel,the length of the largest longtitudinal slope in the tunnel,and the proportion of trucks in the tunnle).The regression model between tunnel accident rate and individual indexes can be obtained,and a multivariate regression model and multiple indexes and tunnel accident rate can be established.Through the analysis above,it can be obtained: 1)the type of accident,time distribution,spatial distribution and characteristics of vehicle types and their differences under the conditions of the whole province,different length types,different climate zones,different traffic levels and different pavement materials;2)The safety difference of asphalt pavement and cement pavement in accident number,accident rate,accident severity and degree of impact;3)single factor regression model of tunnel accident rate and different linear index of tunnel;4)tunnel accident rate and different linear index of tunnel Multi-factor linear regression model.