A highway crash risk assessment method based on traffic safety state division.

Dongye Sun,Yunhua Sun,Yunfei Ai,Liping Zhao,Feng Chen

doi:10.1371/journal.pone.0227609

Abstract

In order to quantitatively analyze the influence of different traffic conditions on highway crash risk, a method of crash risk assessment based on traffic safety state division is proposed in this paper. Firstly, the highway crash data and corresponding traffic data of upstream and downstream are extracted and processed by using the matched case-control method to exclude the influence of other factors on the model. Secondly, considering the weight of traffic volume, speed and occupancy, a multi-parameter fusion cluster method is applied to divide traffic safety state. In addition, the quantitative relationship between different traffic states and highway crash risk is analyzed by using Bayesian conditional logistic regression model. Finally, the results of case study show that different traffic safety conditions are in different crash risk levels. The highway traffic management department can improve the safety risk management level by focusing on the prevention and control of high-risk traffic safety conditions.

Highlights

Highway Traffic Safety Performance Assessment is one of the important means to guarantee its safe and efficient operation
A highway crash risk assessment method based on traffic safety state division roadway segments [7].the Bayesian multivariate random-parameter Tobit model and spatio-temporal correlation model are incorporated to analyze the relationship between the crash rates with the injury severity [8,9]
The input of Fuzzy c-mean (FCM) model is a data set with four characteristics, which is composed of the traffic state comprehensive evaluation index on the four detectors (i.e. U2, U1, D1, and D2) located upstream and downstream of the crash site

Summary

Introduction

Highway Traffic Safety Performance Assessment is one of the important means to guarantee its safe and efficient operation. The crash and non-crash traffic flow data samples on the upstream and downstream of the crash location are collected and matched in case-control sample structure for cluster analysis. The Bayesian conditional logistic regression model is proposed to evaluate the highway crash risk under different traffic safety conditions.

Results

Conclusion