Research On Accident Risk Measurement Of Bus Drivers Under The Joint Framework Of Accident Liability And Frequency

The scale of domestic bus operation is gradually increasing,and bus accidents are also increasing year by year.The professional bus drivers urgently need to be scientifically managed.In the previous literature,when studying the relationship between bus drivers and accidents,it was not possible to deeply explore the differences in violation types,driver groups,and accident risk measurement.Therefore,by excavating the correlation between the violations of different groups of bus drivers and their accident consequences and accident risks,reasonably measuring the accident risk of drivers,and identifying potential high-risk bus drivers,it can effectively improve the safety management level of bus operators to drivers,so as to ensure road traffic safety.Based on the above background,this study uses the data of 3697 violations and accidents of the operation safety management system of a municipal bus company in China within two years,and takes several violations and demographic factors of drivers as the research variables.First,the driver is analyzed for latent class analysis,dividing the driver into different categories.Secondly,taking the driver’s accident liability type and accident frequency as the research dependent variables,the modeling and analysis of accident liability based on the standard Probit model and the Probit model with random parameters were carried out,and the accident frequency was modeled and solved based on the zero-inflated Poisson model and the zeroinflated model of random parameters.Third,to ascertain the disparity in the effect of accidents between different types of drivers,independent accident liability and accident frequency models of each type of driver are established.Finally,The TOPSIS-gray correlation analysis method is employed to measure the driver’s accident risk degree,which is then divided by the K-means cluster analysis method,after constructing an evaluation framework for bus driver accident risk based on the driver’s responsible accident probability and accident frequency as indicators.Results demonstrate that(1)the most advantageous outcome is when the driver should be classified into three groups.The results of the full data set and the independent data modeling results based on potential categories were significantly different,and behaviors such as not wearing seat belts and using cell phones had different effects on different types of drivers.(2)There are obvious differences between different types of violations and the impact of driver accident liability and frequency.When the driver has several specific violations at the same time,the probability and frequency of accidents are significantly higher than those of other drivers;(3)The modeling of accident liability showed that the impact of pressing against the lane’s edge while driving,not yielding to the pedestrians,illegal parking,etc.have random characteristics,and the average value of the illegal parking behavior parameters was affected by behaviors such as wearing uniform untidily,changing lanes suddenly,driving outside the bus lane,and not stopping at the bus station;(4)In the accident frequency modeling,it was found that the parameters of variables such as not wearing seat belts,not yielding to the pedestrians and not updated license on time have random characteristics,and unobserved heterogeneity in the data was detected in the random parameter of not yielding to the pedestrians,which was significantly correlated with the number of pressing against the lane’s edge while driving and driving outside the bus lane,which would further increase the frequency of accidents.(5)Finally,the results of the driver’s potential accident risk assessment show that drivers in class 3 have the highest latent risky accident rate;Different risk levels also have significant characteristics,high-risk drivers have the characteristics of young age and short driving experience,and low-risk drivers have the characteristics of older age and long driving experience.The research results provide a theoretical basis for bus operators to find potential high-risk drivers and formulate reasonable and effective targeted management measures for different risk levels of risk drivers.