Off-campus UNL users: To download campus access dissertations, please use the following link to log into our proxy server with your NU ID and password. When you are done browsing please remember to return to this page and log out.
Non-UNL users: Please talk to your librarian about requesting this dissertation through interlibrary loan.
Safety Modeling of Highway-Rail Grade Crossings Using Intelligent Transportation System Data
The primary goal of this research is to investigate driver behaviors and the dynamic feature of the driver’s decision-making process as they approach a highway-rail grade crossing (HRGC) where the warning lights and automatic gates have been activated. The ultimate goal is to find out which factors contributing to drivers’ risky behaviors can be improved to increase safety at HRGCs. Through data collection and sample observation, this research starts with an investigation of driver behaviors when confronting an on-coming train at HRGCs in a large-range area. Unlike the traditional decision-making model, which assumes drivers make a simple one-time decision, this study divided the decision-making process into 3A’s zones: the awareness zone, the assessment zone, and the action zone. Each zone is detailed in this dissertation. The probability of drivers falling into a dynamic dilemma zone and the corresponding decision-making were modeled. The relationship between the violation occurrence and the dilemma zone entrapment was studied. Together with the dilemma zone entrapment probability, traffic violation occurrence was predicted. The higher the dynamic dilemma zone entrapment probability, the more likely the occurrence of a violation. The results indicated that the dynamic dilemma zone model had a significant upgrade on the violation estimation compared to the traditional static dilemma zone model. No study in literature has done the HRGC violation prediction using the vehicle profile data. In this dissertation, predictive models of drivers committing violations was developed based on how the drivers reacted toward approaching trains at different situations using two tree-based ensemble methods, i.e., bootstrap forest and boosted tree. Using the speed, time, and acceleration profile data, the model can have a prediction accuracy of over 80 percent. While the study is location specific, this study provides a method that can be easily expanded to a wide range of traffic locations and situations to determine the coefficients between the number of vehicles entrapped in the dilemma zone and the number of violations that will actually occur. This is important in order to do the violation-need study such as driver behavior based HRGC safety estimation and the optimization of HRGC warning and control system.
Zhao, Li, "Safety Modeling of Highway-Rail Grade Crossings Using Intelligent Transportation System Data" (2017). ETD collection for University of Nebraska - Lincoln. AAI10683674.