Home About ATSIP Program Awards Venue Sponsors/Exhibitors

Modeling Freeway Crashes Using Lane-Specific Artificial Traffic Data

Abstract: At the event level, the prevailing traffic conditions under which a crash takes place are one of the major contributors. Using real-time traffic data just prior to a crash helps to better understand crash causes associated with traffic speed, volume and density; and supports informed decision-making on effective traffic operational strategies for improving safety. Travel conditions can shift rapidly, and the traffic that a vehicle experienced immediately prior to or at the time of a crash is more relevant than earlier or later traffic conditions. Inductive loop detectors data have been a popular source for real-time traffic data in previous real time crash prediction studies. However, many studies did not consider the traffic conditions occurring right before a crash (e.g. 0-5 minutes), citing that preventative actions may take extra time to identify, notify and activate the prevention system. Instead, traffic data from earlier time periods (e.g. 5-10 minutes) were used in these studies. Moreover, the loop detector spacing can vary substantially from site to site, resulting in a lack of spatial consistency. The loop detector location may affect the estimation of traffic flow at crash locations by producing inconsistently biased traffic data. The discrepancies in the spatial-tempo domain mean that crash prediction models developed with traffic data collected directly from loop detector stations may be inadequate. Such data issues may severely undermine the prediction power of crash models. The objective of this study is to develop a real-time crash prediction models (RTCPM) using artificial traffic data generated from macroscopic traffic simulation. The approach would identify crash-prone conditions with higher accuracy by accounting for the tempo-spatial issue of loop detector data. First, a macroscopic traffic simulation model, a lane-specific cell transmission model (LSCTM), is proposed to instrument a corridor of highway with virtual detector stations on each lane and measure traffic data where physical stations were not available. The lane-specific model would improve the simulation accuracy by accounting for different traffic characteristics across lanes and lane-change activities between lanes compared with traditional CTMs which do not differentiate lanes. Then a RTCPM will be developed using lane-specific simulated traffic data. The lane-specific RTCPM model will be compared with models developed from field loop detector data to evaluate its performance.

Zhi Chen - is currently a research assistant in the Department of Civil & Environmental Engineering of the University of Wisconsin-Milwaukee. Mr. Chen has solid knowledge of statistics, data mining, machine learning, transportation engineering and transportation planning. Mr. Chen has authored and co-authored 5 journal papers and 4 conference papers. The focuses include the calibration of the Highway Safety Manual predictive methods, the model development for predicting freeway crashes using loop detector data, and the development of novel crash prediction model. Mr. Chen has also worked as a key researcher in several projects including evaluating local and tribal rural road design, calibrating the Highway Safety Manual predictive methods for rural roadway segments and intersections, and developing faulty loop detection tools.



Quick Links


Chris Osbourn


fb_logo   linkedin