Research On Black Spot Recognition And Early Warning Model Of Traffic Accidents Based On Adaptive Kernel Density Estimation

According to the report of the 16th International Road Safety Conference,the total length of accident blackspot accounts for 0.25% of the total length of the road network,while the total number of accidents that occurred at accident blackspot accounts for 25% of the total number of accidents on the road network.Therefore,identifying blackspot in road traffic accidents and taking targeted measures against them are the key to reduce traffic accidents and improve road traffic efficiency.This paper studies the traffic safety of arterial highways,and proposes methods to identify blackspot and early warning of blackspot.The specific contents are as follows:Firstly,based on the basic theory of nuclear density estimation,the paper analyzes the characteristics,searchs methods of nuclear density estimation,and determines its application process in blackspot recognition.At the same time,the influencing factors of kernel density estimation,the relationship between kernel function,bandwidth and recognition results are analyzed.Based on Epanechnikov kernel function,the formula of optimal bandwidth is deduced by using insertion method and minimized mean square error.In addition,it analyzes the insufficiency of nuclear density estimation to identify traffic accident blackspot,proposes an adaptive kernel density estimation method,introduces the accident risk index,and builds a traffic accident blackspot recognition model.Secondly,based on the theory of accident blackspot,the paper clarifies the definition of traffic accident blackspot on arterial highways,introduces traditional traffic accident blackspot recognition methods,analyzes their advantages and disadvantages and scope of application.At the same time,Arc GIS software is used to construct the road network model,and the traffic accident data information is used to determine the spatial distribution of traffic accidents.Based on the traffic accident blackspot recognition model,the paper identifies the traffic accident blackspot on the main highway,selects different accident blackspot recognition methods,and introduces Crash Predictive Accuracy Index(CPAI)and Road Repair and Maintenance Cost Limit Indicators(RRMCLI)to test and evaluate the effectiveness of the traffic accident blackspot recognition model.The results show that the blackspot recognition model constructed in this paper can identify about 69% of traffic accidents,which is 1.13 times and 1.27 times of nuclear density estimation and accident frequency method,respectively,and has the best effectiveness and accuracy.Finally,based on the theory of automotive safety technology,the advantages and disadvantages of automotive safety technology and the safety distance warning model are analyzed.On this basis,combined with driving characteristics and using the minimum safe distance,a traffic accident blackspot warning model is constructed;and the relationship between the pavement adhesion coefficient,slip rate and the ABS system is analyzed,and the traffic accident blackspot warning model is revised.In addition,Car Sim and Simulink software are used for joint simulation and compared with Honda,PATH,and NHTSA safety distance early warning models to verify the effectiveness of the model.