Traffic Risk Prediction Of Expressway Based On Big Data

Nowadays,traditional methods of analyzing expressway safety risks are subject to subjective limitations of analysts,and the designed methods are still limited by experience or knowledge,making it impossible to accurately predict traffic safety.The large scale and wide variety of data generated by the operation of the expressway make the traditional attribution theory model unable to process and analyze multiple heterogeneous data at the same time.Based on the big data-driven data warehouse and data mining technology,it can analyze the operating data of different ranges and regions in a unified manner,mine the spatial and temporal distribution characteristics,improve the scientific utilization efficiency of traffic data resources,and improve the information service level for traffic safety and early warning.Therefore,this paper focuses on the characteristics of China’s road traffic accident information collection data and the key issues in the application of data analysis.It conducts comprehensive and multi-angle analysis and research,and is committed to reducing the occurrence of road traffic accidents.This paper mainly discusses in detail from the aspects of accident cause analysis,classification of road accident risk levels and prediction of road accident risk levels,and uses related theories and methods of big data analysis such as classification,regression,association rule mining,and visual analysis.The above three aspects have been studied in detail:(1)Based on the expressway accident data of Hunan Province,basic geographic information data,and road geometric data of Hunan Province,the paper puts forward the emphasis on the identification and repair of fault data,the transformation of traffic data and the complementation of traffic data matching.Traffic data preprocessing method.Aiming at the problems of large dimension and large scale of big data of expressway operation characteristics,an accident feature extraction method is proposed to filter redundant attributes in big data of characteristics and obtain accident-related characteristic parameters.(2)Carry out statistical analysis on the overall accident situation of expressway,and especially on the typical sections of expressway such as bridges,tunnels,entrances and exits,service areas,harbors,etc.,to study the spatial and temporal distribution characteristics of accidents on the upstream and downstream adjacent sections before and after the accident It is used to define the accident characteristic parameters of dangerous traffic states and determine different safety states.(3)Build a expressway big data analysis and visualization platform,visually display the accident situation in Hunan Province through the visualization platform,and build an analysis of the risk factors of expressway accidents based on different traffic safety conditions for expressway use Decision Trees.Combining the data imbalance technology SMOTE+ENN and machine learning algorithms,including Decision Trees,Random Forests and Generalized Regression Neural Network algorithms,the accident risk diagnosis models of SMOTE+ENN-RF and SMOTE+ENN-DT-GRNN and the quantitative assessment were built The influence of road factors,weather conditions,time,and abnormal driving behavior on the risk of accidents,and the prediction accuracy of the model is tested.