Based On AD-DBSCAN Algorithm And Grey Theory Traffic Accident Correlation Study

With the rapid growth of road mileage,motor vehicle ownership and the number of drivers,the total number of traffic accidents and the number of casualties remain high.In road traffic safety research,the study of the geographical distribution of traffic accident characteristics is an important step in the research.By studying the characteristic distribution of regional traffic accidents and the living areas of residents,accidents with certain district or village specific characteristics are determined.The study of the geographical distribution of traffic accident characteristics has important theoretical significance and practical value,and can provide reference for traffic management departments to develop targeted accident prevention countermeasures.In addition,another important study is the prediction study,which evaluates the relevance of each influencing factor on road safety by studying the impact of various accident factors on road traffic,so as to make targeted rule-making on road safety and ensure a safe environment for public transportation.This paper firstly studied and researched the geographical distribution of traffic accident characteristics and the spatial clustering theory,combined the spatial characteristics of road traffic and the DBSCAN clustering algorithm to identify the spatial characteristics of accident-prone point segments,divided the traffic accident areas on the urban roads in Dallas,USA,carried out the validity verification of the identification method,and came up with the accident-prone areas and road sections.Second,the number of road traffic accidents in the next few years was predicted and analyzed by using the gray GM(1,1)prediction model.The system of road traffic is composed of various road environmental factors,and each influencing factor of road traffic accidents was analyzed to find out the main environmental factors affecting road traffic accidents by using the gray correlation analysis.The specific research work is as follows:(1)For the shortcomings of the traditional density-based spatial clustering algorithm-(Density-Based Spatial Clustering of Applications with Noise,DBSCAN),the clustering effect is not significant and the selection of parameter combinations.Therefore,an AD-DBSCAN algorithm with adaptive parameters is proposed,which makes the algorithm more difficult in parameter selection.By establishing the DBSCAN algorithm model,which is adapted to find the optimal distance threshold and the minimum number of adjacent points,the clustering is more accurate,and the noise points identified in the data are more accurate.By observing the clustering algorithm evaluation index of the Calinski-Harabasz index calculation model,selecting the optimal distance threshold and the minimum number of neighbor points,the accuracy of noise point recognition in the clustering algorithm is increased by 5 times,and the Calinski-Harabasz index is increased by about 39.84 %,which verifies the applicability of the algorithm in the clustering of urban road traffic accident locations.(2)To improve the accuracy of traffic accident data prediction by designing an improved gray prediction model(Grey Model(1,1))for the road traffic safety problems brought by the increase of motor vehicle use in society,the GM(1,1)prediction model was used to predict the number of road traffic accidents in New York City,USA,from 2021 to 2023,and the results of one fitting were adjusted according to the accuracy of the prediction results The residual repair was performed to derive the prediction results with scientific accuracy.In addition,the gray correlation between road temperature,wind chill value,humidity,pressure value,visibility,wind speed and other factors and each cause of accidents and the importance of each cause of vehicle traffic accidents were analyzed by gray correlation degree analysis,and the laws of vehicle traffic accidents were explored by analyzing the optimal governing factors affecting vehicle traffic accidents,and the corresponding measures of road traffic accident prevention were proposed for these key factors,for traffic management departments to effectively control the occurrence of traffic accidents to provide reference.