Font Size: a A A

Traffic Accident Prediction Model Based On IHSDM Framework

Posted on:2015-11-10Degree:MasterType:Thesis
Country:ChinaCandidate:Q Z HouFull Text:PDF
GTID:2272330422492298Subject:Traffic and Transportation Engineering
Abstract/Summary:PDF Full Text Request
The poor alignment condition is an important cause of accident. Analyzing the relation between accident and alignment and building accident prediction model is very important for optimizing alignment design and decreasing accident.The IHSDM crash prediction model with reliable modeling method developed by FHWA is widely respected and studied. In this paper, we use this modeling method for reference to build crash prediction models for freeways in mountainous, rolling and plain areas in China based on freeway alignment and accident data.Firstly, the freeways are divided into a series of subsections based on alignment conditions by―same attributes method‖. Then the accident data collected in advance and AADT data of subsections collected by video investigation method are distributed to subsections. Then accident analysis database are created containing geometric alignment indexes, accidents, AADT.Then the relations between accidents and alignment indexes on single alignment such as radius of horizontal curve, longitudinal gradient and combination indexes subsections such as horizontal curve versus vertical curve are analyzed. And the alignment indexes which has significant impact on accident are determined. After that, the alignment indexes ranges are determined when the corresponding accident rate is minimal which is the base alignment conditions in this paper.The accident prediction model with base conditions are proposed after analyze the relation between accident rate and AADT on subsections with base conditions. After that, the modification factors of radius of horizontal curve, longitudinal gradient, external angle of deflection, radius of vertical curve, length of tangent for mountainous and rolling area freeways and the modification factors of length of tangent for plain area freeways are determined based on the relations between actual accident and prediction accident by accident prediction model with base conditions. Then the accident prediction models are established for freeways in different terrain conditions after combine the accident prediction model with base conditions and alignment modification factors.The prediction models for mountainous freeway, rolling freeway, plain area freeway validation data comes from Shendan freeway,Shenda freeway and Tiefu freeway in Liaoning province. The validation result shows that the prediction models of this paper have certain accuracy. At last, the prediction models are used to predict accident for Chaohui freeway in Guangdong province.
Keywords/Search Tags:expressway, accident prediction model with base conditions, alignmentmodification factors, IHSDM, regression analyze
PDF Full Text Request
Related items