Research On Road Traffic Safety Risk Identification Method Driven By Driving Behavior Data

The urban road traffic environment is complex and changeable,and the problem of vehicle conflicts is becoming increasingly prominent.Abnormal driving behavior will lead to further aggravation of vehicle conflict,causing the chaos of traffic system operation,thus increasing the risk of road traffic safety.The existing traffic safety risk identification methods are mostly mathematical statistical analysis after the accident,which is difficult to accurately describe the traffic running state before the accident.With the rapid development of vehicle network technology,a large number of abnormal driving behavior data have been accumulated,but these data have not been fully mined,which has virtually caused a waste of data resources.Based on the OBD abnormal driving behavior data of the Internet of vehicles,this paper analyzes in detail the influence of single road conditions and combined road conditions on the abnormal driving behavior rate of vehicles,and establishes the spatial analysis model of abnormal driving behavior.Finally,combined with the road safety risk assessment method,the road traffic safety risk status is identified.The specific research contents are as follows:(1)A systematic review of the research on road traffic safety risk identification methods and driving behavior data.The research data types,sources and mathematical models of current traffic safety risk identification methods are analyzed in detail,and the influencing factors of driving behavior and the research on driving behavior and safety are expounded.This paper analyzes the limitations and shortcomings of current research,and analyzes the opportunities and challenges of current risk identification methods and driving behavior research.(2)According to the location of abnormal driving behavior,road slope,curve,bus station and opening were selected as influencing factors.Taking 6 main roads in the main urban area of Chongqing as the research object,the above roads were divided into 130 road sections according to the "homogeneity method".The influence of different road conditions on the rate of abnormal driving behavior was qualitatively analyzed,and the correlation degree between different road conditions and abnormal driving behavior was quantitatively analyzed by referring to the grey relational model.(3)Two-factor analysis of abnormal driving behavior.The frequency of abnormal driving behavior was analyzed under the conditions of road slope,bend,bus station and opening combination.(4)Spatial analysis model of abnormal driving behavior.Based on the analysis of the applicable characteristics of the Possion regression model and the zero-inflation series regression model,on the basis of this,a model of the spatial distribution of rapid acceleration,rapid deceleration,sharp turns,and overspeed behavior was established.The results show that the rapid acceleration rate and the rapid deceleration rate are suitable for the Possion regression model,and the zero-inflation negative binomial distribution(ZINB)regression model for the sharp turn rate is more effective.The zero-inflation Possion(ZIP)regression model for the over speed is the best.(5)Based on the calculation model of abnormal driving behavior rate and combined with the road safety risk assessment method,an integrated road traffic safety risk identification method driven by driving behavior data is established.(6)Case study.Select 3 roads in Chongqing and use the calculation model of abnormal driving behavior rate in this paper to get the theoretical values of rapid acceleration rate,rapid deceleration rate,sharp turn rate and over speed in each section,The error between the theoretical value and the actual value of each road section was analyzed to verify the feasibility of the model in this paper.At the same time,based on the integrated road traffic safety risk identification method driven by driving behavior data,road traffic safety risks are classified into high and low risks according to the road traffic safety entropy value.The classification threshold is 0.051,and the accuracy rate is83.78%.