Principal Investigator/Researcher

Dr. Joseph W. Palese, Tyler Bernstein (Student), University of Delaware

Project Description

Utilizing the national grade crossing inventory database and other readily available demographic data, a Bayesian Network will be developed to predict optimal crossing protection and accident/collision risk. Data will be assembled and distilled and an exploratory data analysis performed to identify critical variables in grade crossing accident risk. A literature search will be performed and an exposure metric employed based on train and highway traffic density. This metric along with other variables will be used to develop a Bayesian Network that defines the protection level required for an individual crossing based on historic performance of similar crossings, and predicts the probability of train/road vehicle collision.

Implementation of Research Outcomes

The result of the research will be a self-learning model that can be used to determine high risk grade crossings and the optimal level of grade crossing protection.

Impacts/Benefits of Implementation

The benefit of this research will be the ability to use a data driven model to analyze grad crossing accident rick and prioritize remedial actions for high risk grade crossings.