Research On Traffic Safety Evaluation Indexes Of Mountain Urban Road Based On Vehicle OBD Data

Urban road traffic safety has always been an important part of the development of urban transport systems.In recent years,although the number of casualties in traffic accidents in China has decreased,the number of casualties is still huge.Therefore,the research on urban road traffic safety still has a long way to go,and the road traffic safety evaluation after the traffic accident is difficult to describe the road traffic safety risk before the accident.The development of driving behavior information data collection technology provides a favorable opportunity for the development of urban transportation.Driving behavior is the response of the driver under the influence of road conditions,weather conditions,traffic conditions,etc.Driven by driving behavior data,the road traffic safety risk status evaluation system is constructed,and the influencing factors of abnormal driving behavior are further studied,which has a positive effect on identifying road traffic risk points and preventing traffic accidents.Based on the vehicle OBD data,this paper analyzes the relationship between the high-risk position of abnormal driving behavior and the location of road traffic accidents,and proposes a road traffic safety risk status evaluation indexes with road safety entropy as the first-level indicator and abnormal driving behavior rate as the second-level indicator.Determine the state of road traffic safety risk before the accident occurs.At the same time,based on the rough set theory of artificial intelligence field,the factors affecting the abnormal driving behavior of road traffic safety risk are studied.The specific research contents are as follows:(1)The system reviews the research history of road traffic safety evaluation,and analyzes the research object,evaluation index selection and index weight calculation method of current road traffic safety evaluation.By analyzing the research results of the past 20 years,the development trends of research objects,research data and research methods involved in road traffic safety evaluation are expounded.Combined with the limitations of existing evaluation methods and data in road traffic safety assessment,the research hotspots and challenges of road traffic safety assessment are analyzed.(2)Analyze the characteristics and advantages of vehicle OBD data,and give the general processing method of vehicle OBD data.The characteristics of traditional road traffic safety evaluation data and vehicle OBD data are compared,and the advantages and limitations of the two are summarized.The advantages of vehicle OBD data for road traffic safety risk assessment are clarified,and the process of data classification,association and feature data extraction based on feature fields is given for vehicle OBD data types.(3)Construction of road traffic safety risk assessment indexes.The theory of information entropy is introduced,and the road safety entropy is proposed as the firstlevel indicator of road traffic safety risk assessment.The rapid acceleration rate,rapid deceleration rate,sharp turn rate,over-speed and high-speed neutral taxi rate are the second-level indicators.Based on the original entropy weight method,the base selection of log logarithm,the second-level index zero-value processing and the weight calculation method are optimized.Based on the improved entropy weight method,the weight of each abnormal driving behavior rate is objectively calculated,and then the road safety entropy value is calculated.(4)Propose a method for dividing road traffic safety risk levels.Combined with road traffic accident data and considering the randomness of traffic accidents,a road traffic safety risk classification method based on density clustering and k-means two-step clustering is proposed.Firstly,based on density clustering,the data points with relatively distributed distribution are removed,and then clustering is performed based on k-means.At the same time,the contour coefficients under different cluster categories are calculated,and the number of categories with the largest contour coefficient is used as the road safety entropy level.At the same time,the two adjacent cluster centers are the upper and lower limits,and 0.01 is the growth unit.The classification accuracy under different thresholds is calculated,and the highest accuracy value is used as the road safety risk classification threshold.(5)Research on the influencing factors of abnormal driving behavior.Abnormal driving behavior is an important indicator of road traffic safety risk.In the research of road traffic safety risk assessment system,it is necessary to explore its influencing factors.Focusing on the influence of road gradient,turning radius,traffic state and weather conditions on abnormal driving behavior,based on rough set theory of artificial intelligence field,objectively calculate different influencing factors to abnormal driving behavior by calculating the information quantity and importance of different influencing factors.The significance of the impact.(6)Empirical research.Four sections of Chongqing(with a total length of about 38km)were selected as the empirical research objects.Considering the slope of the road,the turning radius and the road opening situation,the empirical road sections are further divided into 8 types.Based on the optimized entropy weight method,the abnormal driving behavior rate weights of different types of road sections are objectively calculated,and the safety entropy value of the road sections is calculated.At the same time,comparing the road safety entropy and traffic accident data,the results show that the road safety entropy is consistent with the trend of road traffic safety status characterized by traffic accidents.Based on the density clustering and k-means two-step clustering classification method,the road traffic safety risk is divided into two levels: high and low risk.The threshold is 0.042 and the accuracy is 87.88%.Based on the rough set theory,the influencing factors of abnormal driving behavior are calculated.The results show that the curve is from large to small,the road turning radius,traffic state,road gradient and weather conditions.